둘러보기

Unity Shader 노트

• github: nf.example.shader

  • collections
  • SRP

• ShaderArt 관련

• ComputeShader 관련

Pipeline

텍스쳐

메쉬 VertexBuffer pos/uv/normal/color IndexBuffer

Gamma / Linear Pipeline (HDR)

Rendering Forward Forward Plus Deferred

CPU =DrawCall=> GPU

Culling

Input Assembly Shader:Vertex object/world/camera/clip Shader:Hull Shader:Tesselator Shader:Domain Shader:Geometry

Rasterizer
    정점간 보간
Early Depth Testing

Shader:Pixel

Depth Testing

Render Target Output

포스트프로세스 (HDR)Luminance (HDR)Bloom (HDR)Eye Adaptation (HDR)Tone Mapping

SSR SSAO SSGI Motion Blur

Depth of Field - Bokeh(geometry / gather) Color Filtering Gamma Control

AA

LOD HLOD Lumen

Vector

Swizzling

벡터의 요소들을 이용 임의의 순서로 구성가능

float4 A = float4(1, 2, 3, 4);

A.x     == 1
A.xy    == float2(1, 2)
A.wwxy  == float4(4, 4, 1, 2)
A.rgba  == float4(1, 2, 3, 4)

내적 외적

• 내적과 외적 공식.
• 내적과 외적을 시각적으로 생각할 수 있어야 함.
• 이거 이름 햇갈리기 쉬움.

| 내적 | Dot Product | Inner Product |

• 닷은 점이니까 모이는건 내적
• 점이니까 두개 모아서 하나가 됨.
• 하나로 모이니 두 벡터 사이의 각도를 구할 수 있음.
• 각도니까 cos연산 들어감.
• https://rfriend.tistory.com/145
• 교환법칙이 성립

| 각도 | 값  |
| ---- | --- |
|    0 |  1  |
|   90 |  0  |
|  180 | -1  |
| -270 |  0  |

        1
        |
        |
0-------+------ 0
        |
        |
       -1

| 외적 | Cross Product | Outer Product |

• 크로스는 삐죽하니까 외적으로 외울껏.
• X 니까 삐저나옴.
• X가 직각이니 수직 구할때 씀.
• https://rfriend.tistory.com/146
• 교환법칙 성립안함

Matrix

If w == 1, then the vector (x,y,z,1) is a position in space. If w == 0, then the vector (x,y,z,0) is a direction

// 순서주의
TransformedVector = TranslationMatrix * RotationMatrix * ScaleMatrix * OriginalVector;

// ref: https://www.3dgep.com/3d-math-primer-for-game-programmers-matrices/#Rotation_about_an_arbitrary_axis

이동행렬
|  1  0  0  x  |
|  0  1  0  y  |
|  0  0  1  z  |
|  0  0  0  1  |

스케일
|  x  0  0  0  |
|  0  y  0  0  |
|  0  0  z  0  |
|  0  0  0  1  |


X축 회전
|    1    0    0    0 |
|    0  cos -sin    0 |
|    0  sin  cos    0 |
|    0    0    0    1 |

Y축 회전
|  cos    0  sin    0 |
|    0    1    0    0 |
| -sin    0  cos    0 |
|    0    0    0    1 |

Z축 회전
|  cos -sin    0    0 |
|  sin  cos    0    0 |
|    0    0    1    0 |
|    0    0    0    1 |


임의의 N축 회전
s : sin
c : cos
ic: 1 - cos

| ic * NxNx + c      | ic * NxNy - s * Nz | ic * NzNx + s * Ny | 0 |
| ic * NxNy + s * Nz | ic * NyNy + c      | ic * NyNz - s * Nx | 0 |
| ic * NzNx - s * Ny | ic * NyNz + s * Nx | ic * NzNz + c      | 0 |
|                  0 |                  0 |                  0 | 1 |

Mesh

• https://docs.unity3d.com/ScriptReference/Mesh.html

// (0,1) +----+ (1,1)
//       |    |
// (0,0) +----+ (1,0)
//
//     2 +----+ 3
//       |    |
//     0 +----+ 1


Mesh mesh = new Mesh();
Vector3[] vertices = new Vector3[4] {
    new Vector3(0, 0, 0),
    new Vector3(1, 0, 0),
    new Vector3(0, 1, 0),
    new Vector3(1, 1, 0)
};

int[] tris = new int[6] {
    // lower left triangle
    0, 2, 1,

    // upper right triangle
    2, 3, 1
};

Vector2[] uv = new Vector2[4] {
    new Vector2(0, 0),
    new Vector2(1, 0),
    new Vector2(0, 1),
    new Vector2(1, 1)
};

Vector3[] normals = new Vector3[4] {
    -Vector3.forward,
    -Vector3.forward,
    -Vector3.forward,
    -Vector3.forward
};

mesh.vertices = vertices;
mesh.triangles = tris;
mesh.uv = uv;
mesh.normals = normals;

Topology

• https://docs.unity3d.com/ScriptReference/MeshTopology.html

MehsFilter mf = GetComponent<MeshFilter>();
mf.mesh.SetIndice(mf.mesh.GetIndices(0), MeshTopology.Points, 0);

public void SetIndices(int[] indices, MeshTopology topology, int submesh, bool calculateBounds = true, int baseVertex = 0);

메쉬토폴로지를 변경시켜 좀 더 그럴듯한 효과를 얻을 수 있다.

Ref

• https://docs.unity3d.com/ScriptReference/Mesh.SetIndices.html

Coordinate

• 좌표공간

유니티 정의 positions

• 동차좌표 homogeneous coordinate
  • 동차좌표는 n차원의 직교좌표를 n+1차원으로 확장한 것이다
  • float3 => float4로 되면서 Homogeneous를 붙임
• Perspective-Correct Interpolation을 위해 Normalized Device Coordinate (Homogeneous)를 사용함.

    uint2  positionSS;  // Screen space pixel coordinates                       : [0, NumPixels)
    uint2  tileCoord;   // Screen tile coordinates                              : [0, NumTiles)
    float  deviceDepth; // Depth from the depth buffer                          : [0, 1] (typically reversed)
    float  linearDepth; // View space Z coordinate                              : [Near, Far]

OS  ----------------------- Object Space
 | UNITY_MATRIX_M * OS
WS  ----------------------- World Space
 | UNITY_MATRIX_V * WS
VS  ----------------------- View Space
 | UNITY_MATRIX_P * VS
CS  ----------------------- Clip Space (Homogeneous)
 | NDC    = CS * 0.5
 | NDC.x  =  NDC.x + NDC.w
 | NDC.y  =  NDC.y + NDC.w // DirectX
 | NDC.y  = -NDC.y + NDC.w // OpenGL
 | NDC.zw = CS.zw
NDC ---------------------- Normalized Device Coordinate (Homogeneous) [0..w]
 | pd = (NDC.xyz / NDC.w); // [0, 1] : perspective divide
 | ndc = pd * 2.0 - 1.0;   // [-1, 1]
 |
 |
ndc ---------------------- Normalized Device Coordinate (Nonhomogeneous) [-1..1]
 | float2 uv_Screen = positionNDC.xy / positionNDC.w;
 | float2 uv_Screen = GetNormalizedScreenSpaceUV(positionCS);
uv_Screen ---------------- [0, 1]
 | float2 positionScreen = uv_Screen * _ScreenParams.xy;
positionScreen ----------- [0, screenWidth] / [0, screenHeight]

Transform

• com.unity.render-pipelines.core/ShaderLibrary/SpaceTransforms.hlsl

---
title: Positions
---

flowchart LR
    ObjectPos --TransformObjectToWorld--> WorldPos
    ObjectPos --TransformObjectToHClip--> HClipPos

    WorldPos --TransformWorldToObject--> ObjectPos
    WorldPos --TransformWorldToView--> ViewPos
    WorldPos --TransformWorldToHClip--> HClipPos

    ViewPos --TransformViewToWorld--> WorldPos
    ViewPos --TransformWViewToHClip--> HClipPos

---
title: Directions
---

flowchart LR
    ObjectDir --TransformObjectToTangent--> TangentDir
    ObjectDir --TransformObjectToWorldDir--> WorldDir

    TangentDir --TransformTangentToObject--> ObjectDir
    TangentDir --TransformTangentToWorldDir --> WorldDir

    WorldDir --TransformWorldToTangentDir--> TangentDir
    WorldDir --TransformWorldToObjectDir--> ObjectDir
    WorldDir --TransformWorldToViewDir--> ViewDir
    WorldDir --TransformWorldToHClipDir--> HClipDir


    ViewDir --TransformViewToWorldDir--> WorldDir

---
title: Surface Normals
---

flowchart LR
    ObjectNormal --TransformObjectToWorldNormal--> WorldNormal
    
    WorldNormal --TransformWorldToViewNormal--> ViewNormal
    WorldNormal --TransformWorldToTangent--> TangentNormal
    WorldNormal --TransformWorldToObjectNormal--> ObjectNormal
    
    ViewNormal --TransformViewToWorldNormal--> WorldNormal
    
    TangentNormal --TransformTangentToWorld --> WorldNormal

UNITY_MATRIX

• localToWorldMatrix
  • 유니티 4까지는 GPU에 넘겨주기전에 스케일을 가공하여
  • renderer.localToWorldMatrix, transform.localToWorldMatrix가 달랐으나 지금은 같음.

Pserspective Camera

• https://chengkehan.github.io/ReconstructPositionFromDepth.html

// Find our current location in the camera's projection space.
Vector3 pt = Camera.main.projectionMatrix.MultiplyPoint(transform.position);

// Matrix4x4.MultiplyPoint
public Vector3 MultiplyPoint(Matrix4x4 mat, Vector3 v)
{
    Vector3 result;
    result.x = mat.m00 * v.x + mat.m01 * v.y + mat.m02 * v.z + mat.m03;
    result.y = mat.m10 * v.x + mat.m11 * v.y + mat.m12 * v.z + mat.m13;
    result.z = mat.m20 * v.x + mat.m21 * v.y + mat.m22 * v.z + mat.m23;
    float num = mat.m30 * v.x + mat.m31 * v.y + mat.m32 * v.z + mat.m33;
    num = 1 / num;
    result.x *= num;
    result.y *= num;
    result.z *= num;
    return result;
}

// z값 구하지 않으면
public Vector3 MultiplyPoint(Matrix4x4 mat, Vector3 v)
{
    Vector3 result;
    result.x = mat.m00 * v.x + mat.m01 * v.y + mat.m02 * v.z + mat.m03;
    result.y = mat.m10 * v.x + mat.m11 * v.y + mat.m12 * v.z + mat.m13;
    float num = mat.m30 * v.x + mat.m31 * v.y + mat.m32 * v.z + mat.m33;
    num = 1 / num;
    result.x *= num;
    result.y *= num;
    return result;
}

// 값을 대입하면
public Vector3 MultiplyPoint(Matrix4x4 mat, Vector3 v)
{
    Vector3 result;
    result.x = mat.m00 * v.x + 0 * v.y + 0 * v.z + 0;
    result.y = 0 * v.x + mat.m11 * v.y + 0 * v.z + 0;
    float num = 0 * v.x + 0 * v.y + -1 * v.z + 0;
    num = 1 / num;
    result.x *= num;
    result.y *= num;
    return result;
}

// 최종적으로
public Vector3 MultiplyPoint(Matrix4x4 mat, Vector3 v)
{
    Vector3 result;
    result.x = mat.m00 * v.x;
    result.y = mat.m11 * v.y;
    float num = -1 * v.z;
    num = 1 / num;
    result.x *= num;
    result.y *= num;
    return result;
}

(X, Y, linearEyeDepth)
positionNDC // [-1, 1]
X = positionNDC.x * linearEyeDepth / mat.m00
Y = positionNDC.x * linearEyeDepth / mat.m11


The zero-based row-column position:
|  _m00, _m01, _m02, _m03 |
|  _m10, _m11, _m12, _m13 |
|  _m20, _m21, _m22, _m23 |
|  _m30, _m31, _m32, _m33 |

The one-based row-column position:
|   _11,  _12,  _13,  _14 |
|   _21,  _22,  _23,  _24 |
|   _31,  _32,  _33,  _34 |
|   _41,  _42,  _43,  _44 |

UV

• texel(TExture + piXEL) coordinate

Direct X
(0,0)            (1,0)
  +-------+-------+
  |       |       |
  |       |       |
  +-------+-------+
  |       |       |
  |       |       |
  +-------+-------+
(0,1)            (1,1)


OpenGL / UnityEngine
(0,1)            (1,1)
  +-------+-------+
  |       |       |
  |       |       |
  +-------+-------+
  |       |       |
  |       |       |
  +-------+-------+
(0,0)            (1,0)

• 수학적으로 바라보면 모든 2D좌표계를 OpenGL방식으로하면 좌표계를 헷갈릴 걱정이 없다. 하지만, 프로그래밍 하는 입장에서는 DirectX방식이 좀 더 와닿을 것이다.

Ref

• Computergrafik - Vorlesung 6 - Coordinate Systems
• Unity - shader의 World matrix(unity_ObjectToWorld)를 수작업으로 구성
• Unity - shader의 Camera matrix(UNITY_MATRIX_V)를 수작업으로 구성
• Unity - unity_CameraWorldClipPlanes 내장 변수 의미
• Unity - shader의 원근 투영(Perspective projection) 행렬(UNITY_MATRIX_P)을 수작업으로 구성
• 렌더링 파이프라인의 좌표 공간들
• Look At Transformation Matrix in Vertex Shader
• transform.localToWorldMatrix
• Renderer.localToWorldMatrix
• Camera.worldToCameraMatrix
• Camera.projectionMatrix
• GL.GetGPUProjectionMatrix
• http://blog.hvidtfeldts.net/index.php/2014/01/combining-ray-tracing-and-polygons/

Alpha

• 알파쓰면 Testing(discard)이나 Blend가 할 것 없이 성능 잡아먹는다.
  • 구형 모바일 디바이스에선 Blend쪽이 성능이 잘 나오는 경향이 있었다.

SubShader
{
    Tags // SubShader의 Tags는 Pass의 Tags와 다름.
    {
        "RenderPipeline" = "UniversalRenderPipeline"

        // "IgnoreProjector" <<< 요놈은 URP에서 안씀

        // for cutout
        "Queue" = "AlphaTest"              // 렌더순서
        "RenderType" = "TransparentCutout" // 그룹핑(전체 노말맵같이 한꺼번에 바꾸어 그릴때 이용)

        // for blend
        "Queue" = "Transparent"
        "RenderType" = "Transparent"
    }

    Pass
    {
        Tags
        {
            "LightMode" = "UniversalForward"
        }

        // https://docs.unity3d.com/Manual/SL-Blend.html
        Blend A B

        // http://docs.unity3d.com/Manual/SL-CullAndDepth.html
        ZWrite <On | Off> // default: On 
        ZTest <(Less | Greater | LEqual | GEqual | Equal | NotEqual | Always)> // default: LEqual 
    }
}

• Blend
  • 색 혼합 방법
• ZWrite
  • Z 값을 기록할지 안할지 결정.
• ZTest
  • Z 값이 씌여져 있는 상태를 읽어서(ZRead), 그려져도 되는지를 결정.

Alpha Cutout / Alpha Testing

• clip(texkill)을 이용
• 간편. sorting걱정 안해도 됨.
• 구형 모바일에서는 AlphaBlend 보다 성능이 안나오는 경향이 있음.
  • 요즘은 AlphaTesting이 더 낳을지도
  • 모바일(A11(ios), PowerVR 등)은 메모리와 대역폭을 줄이기위해 타일별 렌더링을 하는 TBDR(tile-based deferred rendering)을 이용함.
  • 알파테스팅을 이용할시, 실제 보여지는지 여부를 알파테스팅이 끝날때까지 알 수 없으므로 Deffered 최적화를 방해함.
• 풀, 나무, 머리카락, 털 등...
• clip하여 너무 각지는게 보기 싫어질 정도면 blend를 잘 쓰자
• // if ZWrite is Off, clip() is fast enough on mobile, because it won't write the DepthBuffer, so no GPU pipeline stall(confirmed by ARM staff).

SubShader
{
    Tags // SubShader의 Tags는 Pass의 Tags와 다름.
    {
        "RenderPipeline" = "UniversalRenderPipeline"
        "Queue" = "AlphaTest"
        "RenderType" = "TransparentCutout"
    }

    Pass
    {
        Tags
        {
            "LightMode" = "UniversalForward"
        }

        HLSLPROGRAM
        ...
        half4 frag(VStoFS IN) : SV_Target
        {
            half alpha = ...;
            clip(alpha - _Cutoff);

            return half4(1, 0, 0, 1);
        }
        half4 
        ENDHLSL
    }
}

// URP에선 `_ALPHATEST_ON` 여부로 할지 말지 결정하는 함수가 있다.
// https://github.com/Unity-Technologies/Graphics/blob/master/com.unity.render-pipelines.universal/ShaderLibrary/ShaderVariablesFunctions.hlsl
void AlphaDiscard(real alpha, real cutoff, real offset = real(0.0))
{
    #ifdef _ALPHATEST_ON
        clip(alpha - cutoff + offset);
    #endif
}

Alpha Blend

• 이펙트에서 주로 쓰임
• Alpha Testing보다 디테일 살릴때...
• 불투명 유리

SubShader
{
    Tags // SubShader의 Tags는 Pass의 Tags와 다름.
    {
        "RenderPipeline" = "UniversalRenderPipeline"
        "Queue" = "Transparent"
        "RenderType" = "Transparent"
    }

    Pass
    {
        ZWrite Off // 픽셀 중복으로 출력됨.
        Blend SrcAlpha OneMinusSrcAlpha

        Tags
        {
            "LightMode" = "UniversalForward"
        }
    }
}

• ZWrite Off - 뒷면까지 렌더링하는게 문제가됨
  • 2Pass로 보이는면 랜더링

SubShader
{
    Tags // SubShader의 Tags는 Pass의 Tags와 다름.
    {
        "RenderPipeline" = "UniversalRenderPipeline"
        "Queue" = "Transparent"
        "RenderType" = "Transparent"
    }

    Pass
    {
        Tags
        {
            "LightMode" = "SRPDefaultUnlit"
        }
        ZWrite On
        ColorMask 0 // 색 렌더링 안함
        Cull Front

        HLSLPROGRAM
        ...
        ENDHLSL
    }

    Pass
    {
        Tags
        {
            "LightMode" = "UniversalForward"
        }

        ZWrite Off
        Cull Back
        Blend SrcAlpha OneMinusSrcAlpha

        HLSLPROGRAM
        ...
        ENDHLSL
    }
}

Ref

• Jihoo Oh - 이펙트 쉐이더 2강 - 알파 / 블랜딩
• https://www.gamedev.net/forums/topic/656826-why-the-alphablend-is-a-better-choice-than-alphatest-to-implement-transparent-on-mobile-device/5154785/
• 모바일 기기의 Tile Based Rendering(타일 기반 렌더링)과 유니티에서의 주의 사항 #2 : TBR 대응 리소스 제작시 주의점
• https://blog.naver.com/dasoong15/221356325748
• PowerVR Performance Recommendations The Golden Rules
• Z 버퍼의 Read / Write 개념 by 김윤정
  • 1부. Z 값과 Z 버퍼
  • 2부. 개악마 알파 블렌딩1편
  • 3부. 개악마 알파 블렌딩2편
  • 4부. 개악마 알파 블렌딩3편
  • 5부, 알파 테스팅
  • 6부, Z Read/Write
  • 7부, Z Read On / Write Off

NormalMap

inline void ExtractTBN(in half3 normalOS, in float4 tangent, inout half3 T, inout half3  B, inout half3 N)
{
    N = TransformObjectToWorldNormal(normalOS);
    T = TransformObjectToWorldDir(tangent.xyz);
    B = cross(N, T) * tangent.w * unity_WorldTransformParams.w;
}

inline half3 CombineTBN(in half3 tangentNormal, in half3 T, in half3  B, in half3 N)
{
    return mul(tangentNormal, float3x3(normalize(T), normalize(B), normalize(N)));
}

Varyings vert(Attributes IN)
{
    ExtractTBN(IN.normalOS, IN.tangent, OUT.T, OUT.B, OUT.N);
}

half4 frag(Varyings IN) : SV_Target
{
    half3 normalTex = UnpackNormal(SAMPLE_TEXTURE2D(_NormalTex, sampler_NormalTex, IN.uv));
    half3 N = CombineTBN(normalTex, IN.T, IN.B, IN.N);
}

• NormalMap(법선맵)을 쓰는 이유?
• TBN이란?
• world-Normal 구하는 법?
• 노말맵 혹은 법선맵(tangent space)에서 g채널을 뒤집는 이유?

법선맵을 쓰는 이유

• 정점(vertex)을 많이 밖아서 디테일을 표시하면, 실시간으로 정점을 처리하는데 부하가 걸린다(주로 CPU).
• 셰이더 계산시 법선맵에서 가상의 정점을 생성해 빛을 계산하면 디테일을 살릴 수 있다.

Object Space vs Tangent Space

• 리깅을 사용하는 모델의 경우 정점이 몰핑되면서 노말 벡터의 방향이 바뀌게 되는데 이때는 고정된 오브젝트 의 공간좌표계는 의미가 없어짐.

TBN

N = mul(mat_I_M, normalOS);
T = mul(tangentOS, mat_M);
B = mul(binormalOS, mat_M);
// unity같이 binormalOS를 못어올 경우 N, T를 이용하여 B를 만들 수 있다.
// B = cross(N, T) * tangentOS.w

======== 월드공간 T / B / N 을 구하고 TBN매트릭스(tangent -> world)를 만든다
float3x3 TBN_Tangent2World = float3x3(normalize(Input.T), normalize(Input.B), normalize(Input.N));
| Tx Ty Tz |
| Bx By Bn |
| Nx Ny Nz |

mul(tangentNormal, TBN_Tangent2World);   // 왠지 이케 해버리면 앞서 말한 NormalScaleProblem에 걸릴것 같음


======== TBN은 직교행렬, 직교행렬의 역행렬은 전치행렬.
TBN_World2Tangent = transpose(TBN_Tangent2World);
| Tx Bx Nx |
| Ty By Ny |
| Yz Bz Nz |

mul(TBN_World2Tangent, tangentNormal);   // 이케하면 되겠지?

======== 뇌피셜
// 위에꺼도 맞긴 맞는데...
// TBN은 직교행렬, 직교행렬의 역행렬은 전치행렬.
// traspose(inverse(M)) == M
mul(tangentNormal, TBN_Tangent2World);  // 따라서 이케해도 문제될꺼 없음? 확인해봐야함

normal flatten

//                               T, B, N
const float3 vec_TBN_UP = float3(0, 0, 1);

normalTS = lerp(normalTS, vec_TBN_UP, _Flatteness);

Block Compression

• https://docs.microsoft.com/en-us/windows/win32/direct3d10/d3d10-graphics-programming-guide-resources-block-compression#bc3

UNITY_NO_DXT5nm

DXT5nm이 아닌 경우(UNITY_NO_DXT5nm) 는 다음과 같은 공식을 썼으나,

real3 UnpackNormalRGBNoScale(real4 packedNormal)
{
    return packedNormal.rgb * 2.0 - 1.0;
}

아닌경우 UnpackNormalmapRGorAG을 사용 DXT5, DXT5nm을 처리할 수 있게한다.

real3 UnpackNormal(real4 packedNormal)
{
#if defined(UNITY_ASTC_NORMALMAP_ENCODING)
    return UnpackNormalAG(packedNormal, 1.0);
#elif defined(UNITY_NO_DXT5nm)
    return UnpackNormalRGBNoScale(packedNormal);
#else
    // Compiler will optimize the scale away
    return UnpackNormalmapRGorAG(packedNormal, 1.0);
#endif
}

// Unpack normal as DXT5nm (1, y, 0, x) or BC5 (x, y, 0, 1)
real3 UnpackNormalmapRGorAG(real4 packedNormal, real scale = 1.0)
{
    // Convert to (?, y, 0, x)
    packedNormal.a *= packedNormal.r;
    return UnpackNormalAG(packedNormal, scale);
}

real3 UnpackNormalAG(real4 packedNormal, real scale = 1.0)
{
    real3 normal;
    normal.xy = packedNormal.ag * 2.0 - 1.0;
    normal.z = max(1.0e-16, sqrt(1.0 - saturate(dot(normal.xy, normal.xy))));

    // must scale after reconstruction of normal.z which also
    // mirrors UnpackNormalRGB(). This does imply normal is not returned
    // as a unit length vector but doesn't need it since it will get normalized after TBN transformation.
    // If we ever need to blend contributions with built-in shaders for URP
    // then we should consider using UnpackDerivativeNormalAG() instead like
    // HDRP does since derivatives do not use renormalization and unlike tangent space
    // normals allow you to blend, accumulate and scale contributions correctly.
    normal.xy *= scale;
    return normal;
}

• xyzw, wy => _g_r => rg => xyn // r이 뒤로 있으므로, 한바퀴 돌려줘야함.
• normal.xy = packednormal.wy * 2 - 1; (0 ~ 1 => -1 ~ 1)
• Z는 쉐이더에서 계산. 단위 벡터의 크기는 1인것을 이용.(sqrt(x^2 + y^2 + z^2) = 1) sqrt(1 - saturate(dot(normal.xy, normal.xy)))

Normal Scale Problem

• Normal Transformation

오브젝트를 스케일시킬때 Normal의 변화의 문제

A라는 도형을 x에 대해서 2만큼 스케일 업하고 싶다고 가정하면,

C처럼 x의 스케일 2배 됐다고, 노멀의 x값에 곱하기 2를 해서는 안된다. 역인 나누기 2 를 해야한다.

위치(position)에 대해서는 world-Position = mul(obj-Position, M )이 정상적으로 성립되었다.

하지만, world-Normal = mul( obj-Normal, M ) 처럼 적용했을시 앞+선 B와 같은 문제가 발생한다.

월드행렬(M)식으로 나타내면

우리가 구하고 싶은 행렬을 M-want라 했을시 world-Normal = mul(M-want, obj-Normal)

즉 M-want = traspose(inverse(M)).

DirectX기준 row-major에서의 메트릭스와 벡터의 인자 순서: mul(벡터, 메트릭스) = mul( transpose(메트릭스), 벡터 )

아레 예는 row-major 기준으로 작성.

M          = ObjectToWorld
inverse(M) = WorldToObject
M-want     = traspose(inverse(M))

world-Normal
= mul(obj-Normal   , M-want              )
= mul(obj-Normal   , traspose(inverse(M)))
= mul(inverse(M)   , obj-Normal          )
= mul(WorldToObject, obj-Normal          )

노말맵 혹은 법선맵(tangent space)에서 g채널을 뒤집는 이유

• pope - 노말맵은 왜 파란가?

• https://victorkarp.wordpress.com/2020/06/26/inverting-a-normal-map-in-blender/

• 단위벡터의 크기가 1이지만, (-1, 0, 0)과 같은게 있으므로, 정규화된 법선벡터의 범위는 -1...1이다.

• 바이너리로 저장하기위해 범위를 0...1로 줄이려면 0.5를 곱하고 다시 0.5를 더해주면 된다.

• 셰이더에서 저장된 범위 0...1을 -1...1로 확장시키려면 2를 곱하고 1을 빼주면 -1...1의 범위로 확장된다.

노말맵에서 z값이 강한 경우가 있는데 그럼 이미지가 퍼렇게 보이면서 돌출이 아닌 움푹 패인듯한 느낌이 든다.

• 표면 안쪽으로 향하는(z의 값이 음수인) 경우가 없다.
  • 범위는 0 ~ 1
  • 바이너리 저장시 범위가 0.5 ~ 1로 변경되면서 0.5부터 값이 시작된다.
• 따라서 맵이 퍼렇게 보이면서, 돌출되는 부분이 이미지상 움푹들어간 모습처럼 보인다.

그러므로, g채널을 뒤집어주면 돌출된 부분을 아티스트가 쉽게 인지할 수 있다.

OpenGLY+ / DirectX Y-

• https://blender.stackexchange.com/questions/100017/directx-vs-opengl-normal-maps

• 유니티는 OpenGL(Y+) 를 써서 보기 비교적 편하다.
• DirectX와 같은 엔진에서는 작업자를 위해 Y+텍스쳐 제작 쉐이더에서 y에 -1을 곱해 뒤집어 주는 코드를 넣어주면 작업자들이 편해진다.

MikkTSpace

• Mikkelsen tangent space normal
  • https://github.com/mmikk/MikkTSpace

• 노말맵 베이크
  • 방식1. 하이트맵이나 일반 이미지를 이용해서 노멀맵 변환
  • 방식2. 로우폴리곤과 하이폴리곤을 가지고 베이크
    • 3D프로그램 별 에버레지 버텍스 노말에 동일한 연산을 하지 않음.
      • 동일한 연산을 하도록 MikkTSpace로 통일.

Ref

• youtube: 안콜3D - (전체공개) 노말맵의 모든것
  • PART 01, 02
• xNormal 프로그램
  • 사용법
    • youtube: 엑스 노말(x-Normal) 사용법
      • 1, 2
• https://bgolus.medium.com/generating-perfect-normal-maps-for-unity-f929e673fc57

Cubemap

• 텍스쳐를 받을 수 있는 Cubemap생성하기 : Create > Legacy > Cubemap

TEXTURECUBE(_CubeMap);  SAMPLER(sampler_CubeMap);

half3 reflectVN = reflect(-V, N);
half4 cubeReflect = SAMPLE_TEXTURECUBE_LOD(_CubeMap, sampler_CubeMap, reflectVN, 0);

half3 refractVN = refract(-V, N, 1 / _RefractiveIndex);
half4 cubeRefract = SAMPLE_TEXTURECUBE_LOD(_CubeMap, sampler_CubeMap, refractVN, 0);

CubemapGen

• https://gpuopen.com/archived/cubemapgen/
• https://seblagarde.wordpress.com/2012/06/10/amd-cubemapgen-for-physically-based-rendering/
• https://github.com/gscept/CubeMapGen

// | AMD CubeMapGen | Unity |
// | -------------- | ----- |
// | X+             | -X    |
// | X-             | +X    |
// | Y+             | +Y    |
// | Y-             | -Y    |
// | Z+             | +Z    |
// | Z-             | -Z    |

using System.IO;
using UnityEditor;
using UnityEngine;

public class BakeStaticCubemap : ScriptableWizard
{
    static string imageDirectory = "Assets/CubemapImages";
    static string[] cubemapImage = new string[6] {
        "top+Y", "bottom-Y",
        "left-X", "right+X",
        "front+Z","back-Z",
    };
    static Vector3[] eulerAngles = new Vector3[6] {
        new Vector3(-90.0f, 0.0f, 0.0f), new Vector3(90.0f, 0.0f, 0.0f),
        new Vector3(0.0f, 90.0f, 0.0f), new Vector3(0.0f, -90.0f, 0.0f), 
        new Vector3(0.0f, 0.0f, 0.0f), new Vector3(0.0f, 180.0f, 0.0f),
    };


    public Transform renderPosition;
    public Cubemap cubemap;
    // Camera settings.
    public int cameraDepth = 24;
    public LayerMask cameraLayerMask = -1;
    public Color cameraBackgroundColor;
    public float cameraNearPlane = 0.1f;
    public float cameraFarPlane = 2500.0f;
    public bool cameraUseOcclusion = true;
    // Cubemap settings.
    public FilterMode cubemapFilterMode = FilterMode.Trilinear;
    // Quality settings.
    public int antiAliasing = 4;

    public bool IsCreateIndividualImages = false;

    [MenuItem("GameObject/Bake Cubemap")]
    static void RenderCubemap()
    {
        DisplayWizard("Bake CubeMap", typeof(BakeStaticCubemap), "Bake!");
    }

    void OnWizardUpdate()
    {
        helpString = "Set the position to render from and the cubemap to bake.";
        if (renderPosition != null && cubemap != null)
        {
            isValid = true;
        }
        else
        {
            isValid = false;
        }
    }

    void OnWizardCreate()
    {
        QualitySettings.antiAliasing = antiAliasing;
        cubemap.filterMode = cubemapFilterMode;

        GameObject go = new GameObject("CubemapCam", typeof(Camera));
        go.transform.position = renderPosition.position;
        go.transform.rotation = Quaternion.identity;

        Camera camera = go.GetComponent<Camera>();
        camera.depth = cameraDepth;
        camera.backgroundColor = cameraBackgroundColor;
        camera.cullingMask = cameraLayerMask;
        camera.nearClipPlane = cameraNearPlane;
        camera.farClipPlane = cameraFarPlane;
        camera.useOcclusionCulling = cameraUseOcclusion;

        camera.RenderToCubemap(cubemap);
        if (IsCreateIndividualImages)
        {
            if (!Directory.Exists(imageDirectory))
            {
                Directory.CreateDirectory(imageDirectory);
            }
            RenderIndividualCubemapImages(camera);
        }
        DestroyImmediate(go);
    }

    void RenderIndividualCubemapImages(Camera camera)
    {
        camera.backgroundColor = Color.black;
        camera.clearFlags = CameraClearFlags.Skybox;
        camera.fieldOfView = 90;
        camera.aspect = 1.0f;
        camera.transform.rotation = Quaternion.identity;

        for (int camOrientation = 0; camOrientation < eulerAngles.Length; camOrientation++)
        {
            string imageName = Path.Combine(imageDirectory, cubemap.name + "_" + cubemapImage[camOrientation] + ".png");
            camera.transform.eulerAngles = eulerAngles[camOrientation];
            RenderTexture renderTex = new RenderTexture(cubemap.height, cubemap.height, cameraDepth);
            camera.targetTexture = renderTex;
            camera.Render();
            RenderTexture.active = renderTex;
            Texture2D img = new Texture2D(cubemap.height, cubemap.height, TextureFormat.RGB24, false);
            img.ReadPixels(new Rect(0, 0, cubemap.height, cubemap.height), 0, 0);
            RenderTexture.active = null;
            DestroyImmediate(renderTex);
            byte[] imgBytes = img.EncodeToPNG();
            File.WriteAllBytes(imageName, imgBytes);
            AssetDatabase.ImportAsset(imageName, ImportAssetOptions.ForceUpdate);
        }
        AssetDatabase.Refresh();
    }
}

Etc

• 좋은 큐브맵
  • https://youtu.be/mnuKwAV-MBA?si=g-NaYv2cyln2jQes&t=239
  • 밝고 어둠/ 중간명도 /반사광이 충분히 포함
  • GI가 표현될 수 있는 밝음
  • 태양 반대편에 빛을 받는 물체가 있어야 함
• 라이팅셋팅
  • https://youtu.be/mnuKwAV-MBA?si=HZsut7AbGQ0C-OPE&t=576
  • albedo 명도 벨런스 필수
    • 팔레트준비
  • 색온도 조절이 편함
    • 하지만, 노을질때든지 라이트가 너무 진하면
      • directional light색온도 대신 postprocess 색온도 활용
• 대기
  • https://youtu.be/mnuKwAV-MBA?si=JGrQpz3kCFf7M3i1&t=747
  • atmospheric fog - 하늘,대기
  • exponential

Ref

• https://developer.arm.com/documentation/102179/0100/Implement-reflections-with-a-local-cubemap
• NDC2011 - PRT(Precomputed Radiance Transfer) 및 SH(Spherical Harmonics) 개괄

Stencil

• vert > Depth Test > Stencil Test > Render
• frag > AlphaTest > Blending

템플릿

// 기본값
Pass
{
    Stencil
    { 
        Ref         0      // [0 ... 255]
        ReadMask    255    // [0 ... 255]
        WriteMask   255    // [0 ... 255]
        Comp        Always
        Pass        Keep
        Fail        Keep
        ZFail       Keep
    }
    ZWrite On              // On | Off
    ZTest LEqual           // Less | Greater | LEqual | GEqual | Equal | NotEqual | Always
}

Properties
{
    [IntRange]
    _StencilRef("Stencil ID [0-255]",      Range(0, 255)) = 0
    
    [IntRange]
    _StencilReadMask("ReadMask [0-255]",   Range(0, 255)) = 255
    
    [IntRange]
    _StencilWriteMask("WriteMask [0-255]", Range(0, 255)) = 255

    [Enum(UnityEngine.Rendering.CompareFunction)]
    _StencilComp("Stencil Comparison",     Float) = 8 // Always

    [Enum(UnityEngine.Rendering.StencilOp)]
    _StencilPass("Stencil Pass",           Float) = 0 // Keep

    [Enum(UnityEngine.Rendering.StencilOp)]
    _StencilFail("Stencil Fail",           Float) = 0 // Keep

    [Enum(UnityEngine.Rendering.StencilOp)]
    _StencilZFail("Stencil ZFail",         Float) = 0 // Keep
}

Pass
{
    Stencil
    { 
        Ref         [_StencilRef]
        ReadMask    [_StencilReadMask]
        WriteMask   [_StencilWriteMask]
        Comp        [_StencilComp]
        Pass        [_StencilPass]
        Fail        [_StencilFail]
        ZFail       [_StencilZFail]
    }
}

table

Ex

마스킹

• ZTest 실패: 깊이 X / 색상 X

// 마스크.
// 일부러 비교(Comp)하지 않아서 실패상태로 만들고(Fail) Ref값을 덮어씌운다(Replace).
// 마스킹 작업이 오브젝트 보다 먼저 렌더링 되어야 함으로, 렌더큐 확인.
Stencil
{
    Ref     1
    Comp    Never
    Fail    Replace
}

// 오브젝트.
// 앞서 마스크가 1로 덮어씌운 부분과 같은지 비교(Equal).
// 마스킹 작업이 오브젝트 보다 먼저 렌더링 되어야 함으로, 렌더큐 확인.
Stencil
{ 
    Ref     1
    Comp    Equal
}

실루엣

• 가려져 있는 물체 그리기
  • 1. 일반
    • 스텐실 버퍼 Write
  • 2. 가려지면
    • 가려져있는가 : ZTest Greater
    • 스텐실 버퍼 비교

Pass
{
    Tags
    {
        "LightMode" = "SRPDefaultUnlit"
    }
    
    ZTest Greater
    ZWrite Off

    Stencil
    {
        Ref 2
        Comp NotEqual
    }
}

Pass
{
    Tags
    {
        "LightMode" = "UniversalForward"
    }

    Stencil
    {
        Ref 2
        Pass Replace
    }
}

Ref

• https://docs.unity3d.com/Manual/SL-CullAndDepth.html
• https://docs.unity3d.com/Manual/SL-Stencil.html
• https://www.ronja-tutorials.com/post/022-stencil-buffers/
• https://rito15.github.io/posts/unity-transparent-stencil/#스텐실
• 유니티 URP 멀티 렌더 오브젝트 기법으로 겹쳐진면 투명화하기

Depth

• LinearEyeDepth : distance from the eye in world units
• Linear01Depth : distance from the eye in [0;1]

#include "Packages/com.unity.render-pipelines.universal/ShaderLibrary/DeclareDepthTexture.hlsl"
half3 pd            = IN.positionNDC.xyz / IN.positionNDC.w; // perspectiveDivide
half2 uv_Screen     = pd.xy;

half  sceneRawDepth = SampleSceneDepth(uv_Screen);
half  sceneEyeDepth = LinearEyeDepth(sceneRawDepth, _ZBufferParams);
half  scene01Depth  = Linear01Depth (sceneRawDepth, _ZBufferParams);

• SampleSceneDepth
• Linear01Depth / LinearEyeDepth

// mirror: com.unity.render-pipelines.universal/ShaderLibrary/DeclareDepthTexture.hlsl
float SampleSceneDepth(float2 uv)
{
    return SAMPLE_TEXTURE2D_X(_CameraDepthTexture, sampler_CameraDepthTexture, UnityStereoTransformScreenSpaceTex(uv)).r;
}

// mirror: com.unity.render-pipelines.core/ShaderLibrary/Common.hlsl
// Z buffer to linear 0..1 depth (0 at camera position, 1 at far plane).
// Does NOT work with orthographic projections.
// Does NOT correctly handle oblique view frustums.
// zBufferParam = { (f-n)/n, 1, (f-n)/n*f, 1/f }
float Linear01Depth(float depth, float4 zBufferParam)
{
    return 1.0 / (zBufferParam.x * depth + zBufferParam.y);
}

// Z buffer to linear depth.
// Does NOT correctly handle oblique view frustums.
// Does NOT work with orthographic projection.
// zBufferParam = { (f-n)/n, 1, (f-n)/n*f, 1/f }
float LinearEyeDepth(float depth, float4 zBufferParam)
{
    return 1.0 / (zBufferParam.z * depth + zBufferParam.w);
}

depth buffer value non-linear (in view space)

Sample

// vert
float currEyeDepth = -positionVS.z;
float curr01Depth = -positionVS.z * _ProjectionParams.w;
float4 positionNDC = GetVertexPositionInputs(positionOS).positionNDC;

// frag
half2 uv_Screen = IN.positionNDC.xy / IN.positionNDC.w;
half sceneRawDepth = SampleSceneDepth(uv_Screen);

// --------------------------------------------
half scene01Depth = Linear01Depth(sceneRawDepth, _ZBufferParams);   //  [near/far, 1]

// -----------------------------------------------
// scene01Depth을 _ProjectionParams.z(far plane)으로 늘리면 sceneEyeDepth
half sceneEyeDepth = scene01Depth * _ProjectionParams.z;            //  [near, far]
half sceneEyeDepth = LinearEyeDepth(sceneRawDepth, _ZBufferParams); //  [near, far]

// -----------------------------------------------
// 물체와의 거리를 빼면, 얼마나 앞에 나와있는지 알 수 있다.
half diffEyeDepth = sceneEyeDepth - IN.currEyeDepth;
half intersectGradient = 1 - min(diffEyeDepth, 1.0f);

Reversed-z

TODO

• https://developer.nvidia.com/content/depth-precision-visualized

ReconstructPositionWS

TODO 둘 중 하나 문제있음

// https://www.cyanilux.com/tutorials/depth

OUT.toViewVectorWS = _WorldSpaceCameraPos - vertexInputs.positionWS;

float2 screenUV = (IN.positionNDC.xy / IN.positionNDC.w);

float sceneRawDepth = SampleSceneDepth(screenUV);
float sceneEyeDepth = LinearEyeDepth(sceneRawDepth, _ZBufferParams);

float fragmentEyeDepth = -IN.positionVS.z;
float3 scenePositionWS = _WorldSpaceCameraPos + (-IN.toViewVectorWS / fragmentEyeDepth) * sceneEyeDepth;

// https://docs.unity3d.com/Packages/com.unity.render-pipelines.universal@12.0/manual/writing-shaders-urp-reconstruct-world-position.html
float2 screenUV = IN.positionCS.xy / _ScaledScreenParams.xy;

// Sample the depth from the Camera depth texture.
#if UNITY_REVERSED_Z
    real sceneRawDepth = SampleSceneDepth(screenUV);
#else
    // Adjust Z to match NDC for OpenGL ([-1, 1])
    real sceneRawDepth = lerp(UNITY_NEAR_CLIP_VALUE, 1, SampleSceneDepth(screenUV));
#endif

// Reconstruct the world space positions.
float3 scenePositionWS = ComputeWorldSpacePosition(screenUV, sceneRawDepth, UNITY_MATRIX_I_VP);


// https://github.com/Unity-Technologies/Graphics/blob/master/Packages/com.unity.render-pipelines.core/ShaderLibrary/Common.hlsl
float3 ComputeWorldSpacePosition(float2 positionNDC, float deviceDepth, float4x4 invViewProjMatrix)
{
    float4 positionCS  = ComputeClipSpacePosition(positionNDC, deviceDepth);
    float4 hpositionWS = mul(invViewProjMatrix, positionCS);
    return hpositionWS.xyz / hpositionWS.w;
}

float4 ComputeClipSpacePosition(float2 positionNDC, float deviceDepth)
{
    float4 positionCS = float4(positionNDC * 2.0 - 1.0, deviceDepth, 1.0);

#if UNITY_UV_STARTS_AT_TOP
    // Our world space, view space, screen space and NDC space are Y-up.
    // Our clip space is flipped upside-down due to poor legacy Unity design.
    // The flip is baked into the projection matrix, so we only have to flip
    // manually when going from CS to NDC and back.
    positionCS.y = -positionCS.y;
#endif

    return positionCS;
}

ReconstructNormalVS

• https://wickedengine.net/2019/09/22/improved-normal-reconstruction-from-depth/
• https://gist.github.com/bgolus/a07ed65602c009d5e2f753826e8078a0

// 3 tap
const float2 offset_u = float2(0, _CameraDepthTexture_TexelSize.y); // up
const float2 offset_r = float2(_CameraDepthTexture_TexelSize.x, 0); // right

float depth_c = LinearEyeDepth(SampleSceneDepth(IN.uv           ), _ZBufferParams);  // center
float depth_u = LinearEyeDepth(SampleSceneDepth(IN.uv + offset_u), _ZBufferParams);  // up
float depth_r = LinearEyeDepth(SampleSceneDepth(IN.uv + offset_r), _ZBufferParams);  // right

float3 diff_h = float3(offset_u, depth_u - depth_c);  // horizontal
float3 diff_v = float3(offset_r, depth_r - depth_c);  // vertical

float3 normalVS = normalize(cross(diff_h, diff_v));

Ref

• https://www.cyanilux.com/tutorials/depth/
• https://beta.unity3d.com/talks/Siggraph2011_SpecialEffectsWithDepth_WithNotes.pdf
• https://www.reedbeta.com/blog/depth-precision-visualized/

Lighitng Model

• https://www.jordanstevenstechart.com/lighting-models
• 년도를 보면서 발전상황과 왜 쓰는지 왜 안쓰는지 확인필요.

Lighitng Model - NPR

비 물리기반

Lambert - 람버트

• Johann Heinrich Lambert
• 1760 - Photometria

half NdotL = max(0.0, dot(N, L));
half diffuse = NdotL;

Minnaert - 미네르트

• 1954 - Marcel Minnaert

• 달표면 반사를 표현하기 위해 고안됨. moon shader라 불리기도 함

• https://blog.naver.com/canny708/221551395976

half NdotL = max(0.0, dot(N, L));
half NdotV = max(0.0, dot(N, V));
half diffuse = NdotL * pow(NdotL * NdotV, _MinnaertDarkness);

Phong - 퐁

• 1973 - Bui Tuong Phong

half3 R = reflect(-L, N);
half RdotV = max(0.0f, dot(R, V));
half specular = pow(RdotV, _SpecularPower) * _SpecularNormFactor;

Blinn Phong - 블린 퐁

• 1977 - Jim Blinn

half3 H = normalize(V + L); 
half NdotH = max(0.0, dot(N, H));

half specular = pow(NdotH ,_SpecularPower) * _SpecularNormFactor;

Strauss - 스트라우스

• 1990 - Paul Strauss
• https://blog.naver.com/sorkelf/401550597481

Gooch - 구치

• 1998 - Gooch
• SIGGRAPH1998 - A Non-Photorealistic Lighting Model For Automatic Technical Illustration
• GDC2008 - Stylization with a Purpose: The Illustrative World of TEAM FORTRESS 2
• 따뜻함과 차가움 영역을 나누어 표시

Half Lambert & Wrapped Lambert - 하프 람버트 & 와프드 람버트

• 2004 Half-Life2 - Valve
• SIGGRAPH2006 - Shading In Valve's Source Engine

// half lambert
half NdotL = max(0.0, dot(N, L));
half diffuse = pow(NdotL * 0.5 + 0.5, 2);

// wrapped lambert
half diffuse = pow(NdotL * wrapValue + (1.0 - wrapValue), 2);
half diffuse = max(0.0, (NdotL + _wrapped) / (1.0 - _wrapped));

// ref: https://blog.naver.com/eryners/220144182154
// Harf Lambert사용시 명암 차이가 너무 없어져서 무게감이 없어보인다.
half diffuse = ​pow((dot(N, L) * 0.5) + 0.5, 4)  // Half Lambert + Pow

half diffuse = max(0, ((dot(L, N) + warp) / (1 + wrap + wrap^2)) ^ (1 + wrap));

LUT

• Look Up Texture : 룩업텍스쳐
• Ramp Texture라고도 함
  • Ramp : 증감. 경사(gradient)

Lake

• 2000 - Lake
• Stylized Rendering Techniques For Scalable Real-Time 3D Animation
• 룩업텍스쳐 사용
  • NdotL - LUT Texture(1D)

BARLA

• 2006 - BARLA
• X-Toon: An extended toon shader - Pascal Barla, Joëlle Thollot, Lee Markosian
• 룩업텍스쳐 사용
  • NdotL, Detail - LUT Texture(2D)

Lighitng Model - PBR

물리기반

Cook Torrance - 쿡토렌스

• 1982 - Robert L.Cook & Kenneth E. Torrance - A Reflectance Model For Computer Graphics
• 미세면이론
• 거친표면 specular 초점

Ward - 알드

• 1992 - Gregory J. Ward - Measuring and modeling anisotropic reflection
• 경험적 데이터 기반, 거의 사용되지 않음.

Oren-Nayar - 오렌네이어

• 1994 - Michael Oren & Shree K. Nayar - Generalization of Lambert’s Reflectance Model
• 거친포면 diffuse 초점

half NdotL = max(0.0, dot(N, L));
half NdotV = max(0.0, dot(N, V));
half VdotL = max(0.0, dot(V, L));

half s = VdotL - NdotL * NdotV;
half t = lerp(1.0, max(NdotL, NdotV), step(0.0, s));

half3 A = 1.0 + _OrenNayarAlbedo * (_OrenNayarAlbedo / (_OrenNayarSigma + 0.13) + 0.5 / (_OrenNayarSigma + 0.33));
half3 B = 0.45 * _OrenNayarSigma / (_OrenNayarSigma + 0.09);

half3 diffuse = _OrenNayarAlbedo * max(0.0, NdotL) * (A + B * s / t) / 3.14159265;

Modified Phong - 모디파이드 퐁

• Lafortune and Willems (1994)

half norm = (shininess + 2.0) / (2.0 * PI);

half3 R = reflect(-L, N);
half3 VdotR = max(0.0, dot(V, R));

half3 specular = norm * pow(VdotR, shininess);

Ashikhmin Shirley - 어크먼 셜리

• 2000 - Michael Ashikhmin & Peter Shirley - An Anisotropic Phong BRDF Model
• 퐁 스펙큘러

Fakey Oren-Nayar - 최적화 오렌네이어

• 2011 - pope - Rendering Tech of Space Marine
  • https://kblog.popekim.com/2011/11/blog-post_16.html

half OrenNayar_Fakey(half3 N, half3 L, half3 V, half roughness)
{
    half LdotN = dot(L, N);
    half VdotN = dot(V, N);

    half result = saturate(LdotN);
    half soft_rim = saturate(1 - VdotN / 2);

    const half FAKEY_MAGIC = 0.62;
    half fakey = pow(1 - result * soft_rim, 2);
    fakey = FAKEY_MAGIC - fakey * FAKEY_MAGIC;
    return lerp(result, fakey, roughness);
}

Disney - 디즈니

• SIGGRAPH 2012 - Brent Burley - Physically Based Shading at Disney
• 여러 파라미터

Ref

• https://www.cnblogs.com/timlly/p/10631718.html
• https://www.jordanstevenstechart.com/physically-based-rendering

HemisphereLight

if (degree <= 90)
    x = 1 - (0.5 * sin(degree));
else
    x = 0.5 * sin(degree);

RealColor = x * TopColor + (1 - a) * BottomColor;
=================================================

// 비교적 편차가 적은 간소화 버전으로 변경가능.
x = 0.5 + (0.5 * cos(degree));
// - 분기가 없어졌다.
// - cos(degree)는 dot 연산으로 대처가능
// x = 0.5 + (0.5 * dot(N, L));

FakeColor = x * TopColor + (1 - a) * BottomColor;

half hemiWeight = 0.5 + 0.5 * dot(N, L);
half3 diffuse = lerp(_GroundColor, _SkyColor, hemiWeight);

half3 camPositionWS = GetCurrentViewPosition();
half3 L_VS = GetWorldSpaceViewDir(L);
half skyWeight = 0.5f + 0.5 * max(0, dot(N, normalize(camPositionWS + L_VS)));
half groundWeight = 0.5f + 0.5 * max(0, dot(N, normalize(camPositionWS - L_VS)));
half3 specular = (max(0, pow(skyWeight, _SpecularPower)) + max(0, pow(skyWeight, _SpecularPower)))
    *_SpecularNormFactor
    * hemiWeight
    * diffuse;
half3 result = diffuse + specular;

Ref

• https://github.com/hughsk/glsl-hemisphere-light
• https://emunix.emich.edu/~mevett/GraphicsCourse/Labs/MattsLab5Lighting/index.html#Hemisphere
• https://en.wikibooks.org/wiki/GLSL_Programming/Unity/Diffuse_Reflection_of_Skylight
• Hemisphere Lights - Interactive 3D Graphics

Hair Anisotropic

비등방성(非等方性)(anisotropy)은 방향에 따라 물체의 물리적 성질이 다른 것을 말한다. 
예를 들어, 솔질이 된 알루미늄, 섬유, 옷감, 근육 등의 표면은
들어오는 빛의 방향에 따라 반사율이 다른 광학적 비등방성을 띈다. 
- https://ko.wikipedia.org/wiki/비등방성

• Kajya-Kay 모델 - SIGGRAPH 1989
  • 짧은머리는 괜춘. 빛의 산란효과는 별로
• Steve Marschner 모델 - SIGGRAPH 2003
  • 빛의 산란효과 개선(반사/내부산란/투과)
• Scheuermann - Hair Rendering and Shading - GDC 2004

• 【Unite Tokyo 2018】『崩壊3rd』開発者が語るアニメ風レンダリングの極意
• https://graphics.pixar.com/library/DataDrivenHairScattering/

Kajiya-Kay

• SIGGRAPH 1989

• blog - Hair Rendering Lighting Model - (Kajiya-Kay)

// Sphere
// T | r | 오른쪽
// B | g | 위쪽
// N | b | 직각

// 논문에서 T. 방향은 머리를향한 위쪽 방향.
// half3 T = normalize(IN.T);

// Sphere에서는 B가 위쪽이므로 B로해야 원하는 방향이 나온다.
half3 T = normalize(IN.B);

half sinTH    = sqrt(1 - dot(T, H) * dot(T, H));
half specular = pow(sinTH, specularPower);

Marschner

• SIGGRAPH 2003
• blog - Hair Rendering Lighting Model - (Marschner)
• pdf - Light Scattering from Human Hair Fibers - Stephen R. Marschner, Henrik Wann Jensen, Mike Cammarano
• 2020 - Dev Weeks: A3 Still Alive - Technical Art Review

• 2개의 반사를 이용.
  • Tangent를 이동 (+ TangentShiftTex)
  • 1번째 반사(RR)
  • 두번째반사(TRT) (+ SpecularMaskTex)

Scheuermann

• GDC 2004 Hair Rendering and Shading
• Kajiya-Kay랑 Marschner를 믹스함

에니메이션

• 버텍스 셰이더로 하는 머리카락 애니메이션

Ref

• Hair in Tomb Raider
• ShaderX3 Advanced Rendering with DirectX and OpenGL
  • 2.14 Hair Rendering and Shading

BRDF

• Bidirectional reflectance distribution function

BRDF Texture

• BRDF Fake라고도 함.

half u = dot(L, N) * 0.5 + 0.5;
half v = dot(V, N);

half3 brdfTex = SAMPLE_TEXTURE2D(_BrdfTex, sampler_BrdfTex, half2(u, v)).rgb;

TODO - Ambient BRDF

• Gotanda 2010

TODO - Environment IBL Map

• Schlick's approximation // fresnel
• Lazarov 2013

DFG LUT

예

Color c = diffuse * intensity + fresnelReflectionColor * fresnelTerm + translucentColor * t + Color(0, 0 ,0, specular);
c *= intensity;

half u = dot(L, N) * 0.5 + 0.5;
half v = dot(H, N);

half3 brdfTex = SAMPLE_TEXTURE2D(_BrdfTex, sampler_BrdfTex, half2(u, v)).rgb;
half3 color = albedo * (brdfTex.rgb + gloss * brdfTex.a) * 2;

//   +--- B ---+    A : 빛과 마주치는 면
//   |         |    B : 빛과 반대방향의 면
//   D         C    C : 카메라와 마주치는 면
//   |         |    D : 카메라와 90도 되는 면
//   +--- A ---+ 

OffsetU // [-1, 1]
OffsetV // [-1, 1]

half2 brdfUV = float2(saturate(NdotV + OffsetU), saturate((LdotN + 1) * 0.5) + OffsetV);
brdfUV.y = 1 - brdfUV.y;
half3 brdfTex = tex2D(BRDFSampler, brdfUV).rgb;

half3 color = ambient + brdfTex;

Ref

• SIGGRAH 2013 Real Shading in Unreal Engine 4
  • https://lifeisforu.tistory.com/348
• https://learnopengl.com/PBR/IBL/Specular-IBL
• SIGGRAH2019 - A Journey Through Implementing Multiscattering BRDFs and Area Lights
• https://teodutra.com/unity/shaders/cook-torrance/lookup-texture/2019/03/28/Lookup-The-Cook-Torrance/
• SIGGRAH2017 - Physically-Based Materials: Where Are We?
• Multi-Textured BRDF-based Lighting - Chris Wynn
• http://www.mentalwarp.com/~brice/brdf.php
• http://wiki.polycount.com/wiki/BDRF_map
• Stoyan Dimitrov: How Space Ape uses Uber Shaders in Unity for Mobile Games
  • https://stoyan3d.wordpress.com/2018/11/21/uber-shaders-for-mobile-games/

PBR

• PBR(Physical based rendering) / PBS(Physical based shader)

Energy = diffuse + specular + transmission

• https://renderwonk.com/publications/

[Ndc13]Ndc 2013 김동석:UDK로 물리기반 셰이더 만들기

siggraph 2010 tri-Ace Practical Implementation of Physically-Based Shading Models at tri-Ace (Yoshiharu Gotanda) slide cource notes

siggraph 2011 Lazarov Physically Based Lighting in Call of Duty: Black Ops Dimitar Lazarov, Lead Graphics Engineer, Treyarch

Sébastien Lagarde Moving Frostbite to PBR (Sébastien Lagarde & Charles de Rousiers) https://blog.selfshadow.com/publications/s2014-shading-course/frostbite/s2014_pbs_frostbite_slides.pdf

• Etc
  • BSDF(Bidirectional Scattering Distribution Function)
  • BTDF(Bidirectional Transmission Distribution Function)
  • BSSRDF(Bidirectional Scattering Surface Reflectance Distribution Function)
  • SPDF(Scattering Probability Density Function)

Custom PBR

TODO

• https://www.slideshare.net/dongminpark71/ndc19-pbr-143928930
• 언차티드4 테크아트 파트4 Special Case Materials - Moss & Wetness & Glass
• 언차티드4 테크아트 파트3 MicroShadowBRDF Term

Ref

• https://www.slideshare.net/MRESC/pbr-vol2-131205432
• Adobe The PBR Guide
  • https://substance3d.adobe.com/tutorials/courses/the-pbr-guide-part-1
  • https://substance3d.adobe.com/tutorials/courses/the-pbr-guide-part-2

https://leegoonz.blog/2020/01/05/energy-conserved-specular-blinn-phong/ https://www.rorydriscoll.com/2009/01/25/energy-conservation-in-games/

재질

• https://dev.epicgames.com/documentation/en-us/unreal-engine/physically-based-materials-in-unreal-engine
• https://creativecloud.adobe.com/learn/substance-3d-designer/web/the-pbr-guide-part-1
• https://creativecloud.adobe.com/learn/substance-3d-designer/web/the-pbr-guide-part-2

custom

난반사(Diffuse Reflection) 정반사(Specluar Reflection)

• 방식
  • ref
    • 물리기반 머터리얼 상식
      • 직접광보다 간접광(Environment map)이 중요.
        • https://marmoset.co/shop/
          • Perpetual License $31900one-time fee (USD)
    • 디퓨즈와 스페큘러
      • How To Split Specular And Diffuse In Real Images
      • https://lifeisforu.tistory.com/382
      • https://www.virial.com/reflection-models.html
  • metal - roughness
    • base color : brdf color
    • metallic : reflectance ( specular level, Index of Refelection - 별명이 artistic metallic - 1.6)
    • roughness: glossiness를 선형화 시켜서 뒤짚은 값.
      • 사람은 대략 0.5
  • specular / glossiness
    • diffuse(albedo (알비도)) / specular / glossiness

base color/metalic / roughness 방식

• 단점
  • 텍셀 밀도가 낮을시 metalic edge 현상 발생
• base color
  • PBR Safe Color
    • https://helpx.adobe.com/substance-3d-designer/substance-compositing-graphs/nodes-reference-for-substance-compositing-graphs/node -library/material-filters/pbr-utilities/pbr-albedo-safe-color.html
    • https://helpx.adobe.com/substance-3d-designer/substance-compositing-graphs/nodes-reference-for-substance-compositing-graphs/node-library/material-filters/pbr-utilities/pbr-basecolor-metallic-validate.html
• 메탈릭
  • albedo : 밝아야함(생각보다 어둡게 나오는 경우가 많음)
  • 금속/비금속을 나누는 기준임으로 어중간한 값들의 사용은 자제해야한다.
• 거칠기 : 높을수록 정반사 비율이 낮아짐.
  • 기울여 봐야 Fresnel의 차이를 확인 할 수 있음.

FlatShader

• 노말의 앵글을 없에므로 메쉬간 평평한(Flat)효과를 얻을 수 있다.
• 속성을 변경하거나, 런타임에 변경할 수도 있다.

속성을 변경

• https://gamedevelopment.tutsplus.com/articles/go-beyond-retro-pixel-art-with-flat-shaded-3d-in-unity--gamedev-12259

fbx> Normals & Tangents > Normals> Calculate
fbx> Normals & Tangents > Smoothing Angle> 0

런타임

void FlatShading ()
{
    MeshFilter mf = GetComponent<MeshFilter>();
    Mesh mesh = Instantiate (mf.sharedMesh) as Mesh;
    mf.sharedMesh = mesh;

    Vector3[] oldVerts = mesh.vertices;
    int[] triangles = mesh.triangles;
    Vector3[] vertices = new Vector3[triangles.Length];

    for (int i = 0; i < triangles.Length; i++) 
    {
        vertices[i] = oldVerts[triangles[i]];
        triangles[i] = i;
    }

    mesh.vertices = vertices;
    mesh.triangles = triangles;
    mesh.RecalculateNormals();
}

Shader

• 그게 아니면 shader를 이용해도...
• Unity로 실습하는 Shader (5) - Flat Shading
• https://catlikecoding.com/unity/tutorials/advanced-rendering/flat-and-wireframe-shading/

half3 x = ddx(IN.positionWS);
half3 y = ddy(IN.positionWS);

half3 N = normalize(-cross(x, y));

LOD

• LOD : Level Of Detail
• tex2Dlod와 tex2Dbias의 비교연구

// ref: https://www.unity3dtips.com/unity-fix-blurry-textures-on-mipmap/

UnityEditor.EditorPrefs.SetBool("DeveloperMode", true);

// 인스펙터에 Debug-Internal로 들어가서
// Texture Settings > Mip Bias 부분 설정 가능

밉맵 날카롭게 만들기

• 밉맵 디테일 높이기
• DDS로 밉맵을 따로 제작해서 만들거나
• AssetPostprocessor를 이용해서 처리

1. 밉맵 0으로부터 밉맵 1 생성 (bilinear filter)
2. 밉맵 1에 sharpening filter 적용
3. 2번 결과물로부터 밉맵 2 생성(bilinear filter)
4. 밉맵 2에 sharpening filter 적용
5. 밉맵 끝까지 만들때까지 반복...

AssetPostprocessor

public class MipmapsSharperImporter : AssetPostprocessor
{
    void OnPostprocessTexture(Texture2D texture)
    {
        if (!Path.GetFileNameWithoutExtension(assetPath).EndsWith("_sharppen"))
        {
            return;
        }

        if (texture.mipmapCount == 0)
        {
            return;
        }

        for (int mipmapLevel = 1; mipmapLevel < texture.mipmapCount; ++mipmapLevel)
        {
            ApplyBilinearFilter(texture, mipmapLevel);
            ApplySharpeningFilter(texture, mipmapLevel);
        }
        texture.Apply(updateMipmaps: false, makeNoLongerReadable: true);
    }

    void ApplyBilinearFilter(Texture2D texture, int currMipmapLevel)
    {
        int currMipmapWidth = texture.width / (1 << currMipmapLevel);
        int currMipmapHeight = texture.height / (1 << currMipmapLevel);
        Color[] currPixels = new Color[currMipmapWidth * currMipmapHeight];

        int prevMipmapLevel = currMipmapLevel - 1;
        int prevMipmapWidth = texture.width / (1 << prevMipmapLevel);
        Color[] prevPixels = texture.GetPixels(prevMipmapLevel);
        
        for (int y = 0; y < currMipmapHeight; ++y)
        {
            for (int x = 0; x < currMipmapWidth; ++x)
            {
                int px = 2 * x;
                int py = 2 * y;

                Color c00 = prevPixels[(py) * prevMipmapWidth + (px)];
                Color c10 = prevPixels[(py) * prevMipmapWidth + (px + 1)];
                Color c01 = prevPixels[(py + 1) * prevMipmapWidth + (px)];
                Color c11 = prevPixels[(py + 1) * prevMipmapWidth + (px + 1)];

                Color b0 = Color.Lerp(c00, c10, 0.5f);
                Color b1 = Color.Lerp(c01, c11, 0.5f);
                Color final = Color.Lerp(b0, b1, 0.5f);

                currPixels[y * currMipmapWidth + x] = final;
            }
        }
        texture.SetPixels(currPixels, currMipmapLevel);
    }

    private void ApplySharpeningFilter(Texture2D texture, int mipmapLevel)
    {
        float _Sharpness = 0.1f;
        Color[] pixels = texture.GetPixels(mipmapLevel);
        int mipmapWidth = texture.width / (1 << mipmapLevel);
        int mipmapHeight = texture.height / (1 << mipmapLevel);
        const int HALF_RANGE = 1;
        for (int y = 0; y < mipmapHeight; ++y)
        {
            for (int x = 0; x < mipmapWidth; ++x)
            {
                Color color = pixels[y * mipmapWidth + x];
                Color sum = Color.black;
                for (int i = -HALF_RANGE; i <= HALF_RANGE; i++)
                {
                    for (int j = -HALF_RANGE; j <= HALF_RANGE; j++)
                    {
                        sum += pixels[Mathf.Clamp(y + j, 0, mipmapHeight - 1) * mipmapWidth + Mathf.Clamp(x + i, 0, mipmapWidth - 1)];
                    }
                }
                Color sobel8 = color * Mathf.Pow(HALF_RANGE * 2 + 1, 2) - sum;
                Color addColor = sobel8 * _Sharpness;
                color += addColor;
                pixels[y * mipmapWidth + x] = color;
            }

        }
        texture.SetPixels(pixels, mipmapLevel);
    }
}

HLOD

• HLOD : Hierarchical Level Of Detail
• Unite2019 HLOD를 활용한 대규모 씬 제작 방법

Ref

• https://docs.unity3d.com/ScriptReference/AssetPostprocessor.OnPostprocessTexture.html
• https://zhuanlan.zhihu.com/p/413834301
• https://community.khronos.org/t/texture-lod-calculation-useful-for-atlasing/61475

Shadow

• ShadowCaster패스로 그림자를 그려주고
• 메인 패스에서
  • shadowCoord를 얻어와
    • OUT.shadowCoord = TransformWorldToShadowCoord(OUT.positionWS);로
  • 라이트를 얻고
    • Light mainLight = GetMainLight(inputData.shadowCoord);
  • 그림자를 적용시킨다
    • half shadow = mainLight.shadowAttenuation;
    • finalColor.rgb *= shadow;

// Light & Shadow
#pragma multi_compile _ _MAIN_LIGHT_SHADOWS
#pragma multi_compile _ _MAIN_LIGHT_SHADOWS_CASCADE
#pragma multi_compile _ _ADDITIONAL_LIGHTS
#pragma multi_compile _ _ADDITIONAL_LIGHTS_CASCADE
#pragma multi_compile _ _SHADOWS_SOFT

UnityEngine.Rendering.Universal.ShaderKeywordStrings

// com.unity.render-pipelines.universal/ShaderLibrary/Shadows.hlsl
float4 TransformWorldToShadowCoord(float3 positionWS)
{
#ifdef _MAIN_LIGHT_SHADOWS_CASCADE
    half cascadeIndex = ComputeCascadeIndex(positionWS);
#else
    half cascadeIndex = half(0.0);
#endif
    float4 shadowCoord = mul(_MainLightWorldToShadow[cascadeIndex], float4(positionWS, 1.0));
    return float4(shadowCoord.xyz, 0);
}

OUT.shadowCoord = TransformWorldToShadowCoord(positionWS);// float4

VertexPositionInputs vertexInput = GetVertexPositionInputs(v.vertex.xyz);
OUT.shadowCoord = GetShadowCoord(vertexInput);

Light mainLight = GetMainLight(inputData.shadowCoord);
half shadow = mainLight.shadowAttenuation;
finalColor.rgb *= shadow;

아 유니티 병신같은 문서어딧

// Toggle the alpha test
#define _ALPHATEST_ON

// Toggle fog on transparent
#define _ENABLE_FOG_ON_TRANSPARENT

UsePass "Universal Render Pipeline/Lit/ShadowCaster"

com.unity.render-pipelines.universal/Shaders/ShadowCasterPass.hlsl

// You can also optionally disable shadow receiving for transparent to improve performance. To do so, disable Transparent Receive Shadows in the Forward Renderer asset
_MAIN_LIGHT_SHADOWS _MAIN_LIGHT_SHADOWS_CASCADE _MAIN_LIGHT_SHADOWS_SCREEN ?? => MAIN_LIGHT_CALCULATE_SHADOWS
_MAIN_LIGHT_SHADOWS_CASCADE => REQUIRES_VERTEX_SHADOW_COORD_INTERPOLATOR
_ADDITIONAL_LIGHT_SHADOWS => ADDITIONAL_LIGHT_CALCULATE_SHADOWS

// cascade
// https://forum.unity.com/threads/what-does-shadows_screen-mean.568225/
// https://forum.unity.com/threads/water-shader-graph-transparency-and-shadows-universal-render-pipeline-order.748142/

PipelineAsset> Shadows > Cascades> No Cascades

// com.unity.render-pipelines.universal/ShaderLibrary/Shadows.hlsl
#if !defined(_RECEIVE_SHADOWS_OFF)
    #if defined(_MAIN_LIGHT_SHADOWS) || defined(_MAIN_LIGHT_SHADOWS_CASCADE) || defined(_MAIN_LIGHT_SHADOWS_SCREEN)
        #define MAIN_LIGHT_CALCULATE_SHADOWS

        #if !defined(_MAIN_LIGHT_SHADOWS_CASCADE)
            #define REQUIRES_VERTEX_SHADOW_COORD_INTERPOLATOR
        #endif
    #endif

    #if defined(_ADDITIONAL_LIGHT_SHADOWS)
        #define ADDITIONAL_LIGHT_CALCULATE_SHADOWS
    #endif
#endif

TEXTURE2D_SHADOW(_MainLightShadowmapTexture);
SAMPLER_CMP(sampler_MainLightShadowmapTexture);

half4       _MainLightShadowParams;   // (x: shadowStrength, y: 1.0 if soft shadows, 0.0 otherwise, z: main light fade scale, w: main light fade bias)
float4      _MainLightShadowmapSize;  // (xy: 1/width and 1/height, zw: width and height)

struct ShadowSamplingData
{
    half4 shadowOffset0;
    half4 shadowOffset1;
    half4 shadowOffset2;
    half4 shadowOffset3;
    float4 shadowmapSize;
};

// ShadowParams
// x: ShadowStrength
// y: 1.0 if shadow is soft, 0.0 otherwise
half4 GetMainLightShadowParams()
{
    return _MainLightShadowParams;
}

half MainLightRealtimeShadow(float4 shadowCoord)
    ShadowSamplingData shadowSamplingData = GetMainLightShadowSamplingData();
    half4 shadowParams = GetMainLightShadowParams();
    return SampleShadowmap(TEXTURE2D_ARGS(_MainLightShadowmapTexture, sampler_MainLightShadowmapTexture), shadowCoord, shadowSamplingData, shadowParams, false);

half AdditionalLightRealtimeShadow(int lightIndex, float3 positionWS, half3 lightDirection)
real SampleShadowmap(TEXTURE2D_SHADOW_PARAM(ShadowMap, sampler_ShadowMap), float4 shadowCoord, ShadowSamplingData samplingData, half4 shadowParams, bool isPerspectiveProjection = true) _SHADOWS_SOFT

PipelineAsset> Shadows > Cascades> Soft Shadows

_SHADOWS_SOFT : real SampleShadowmapFiltered(TEXTURE2D_SHADOW_PARAM(ShadowMap, sampler_ShadowMap), float4 shadowCoord, ShadowSamplingData samplingData)
float4 TransformWorldToShadowCoord(float3 positionWS)  : _MAIN_LIGHT_SHADOWS_CASCADE
_MAIN_LIGHT_SHADOWS_CASCADE : half ComputeCascadeIndex(float3 positionWS)


float3 ApplyShadowBias(float3 positionWS, float3 normalWS, float3 lightDirection)

LerpWhiteTo

#pragma multi_compile_fog
OUT.fogCoord = ComputeFogFactor(OUT.positonHCS.z);

half3 ambient = SampleSH(IN.N);
finalColor.rgb *= ambient;
finalColor.rgb = MixFog(finalColor.rgb, IN.fogCoord);

ShadowAttenuation

// URP
half4 shadowCoord = TransformWorldToShadowCoord(positionWS);
// or
// VertexPositionInputs vertexInput = GetVertexPositionInputs(IN.positionOS.xyz);
// half4 shadowCoord = GetShadowCoord(vertexInput);

half shadowAttenuation = MainLightRealtimeShadow(shadowCoord);
// or
// ShadowSamplingData shadowSamplingData = GetMainLightShadowSamplingData();
// half4 shadowParams = GetMainLightShadowParams();
// half shadowAttenuation = SampleShadowmap(TEXTURE2D_ARGS(_MainLightShadowmapTexture, sampler_MainLightShadowmapTexture), shadowCoord, shadowSamplingData, shadowParams, false);
// or
// Light mainLight = GetMainLight(i.shadowCoord);
// half shadowAttenuation = mainLight.shadowAttenuation;

ShadowCaster

// 그림자 그려주는놈
Pass
{
    Tags{"LightMode" = "ShadowCaster"}
}

vert()
{
    OUT.positionCS = TransformWorldToHClip(ApplyShadowBias(positionWS, normalWS, lightDirectionWS));
}

frag()
{
    1 : lit
    0 : shadow

    return 1 or 0;
}

Pass
{
    Name "ShadowCaster"
    Tags
    {
        "LightMode" = "ShadowCaster"
    }

    ZWrite On
    Cull Back

    HLSLPROGRAM
    #pragma target 3.5

    #pragma vertex shadowVert
    #pragma fragment shadowFrag

    #include "Packages/com.unity.render-pipelines.core/ShaderLibrary/Common.hlsl"  // real
    #include "Packages/com.unity.render-pipelines.core/ShaderLibrary/CommonMaterial.hlsl" // LerpWhiteTo
    #include "Packages/com.unity.render-pipelines.universal/ShaderLibrary/Core.hlsl"
    #include "Packages/com.unity.render-pipelines.universal/ShaderLibrary/Shadows.hlsl" // ApplyShadowBias

    struct Attributes
    {
        float4 positionOS   : POSITION;
        float4 normal       : NORMAL;
    };

    struct Varyings
    {
        float4 positionHCS  : SV_POSITION;
    };

    Varyings shadowVert(Attributes IN)
    {
        Varyings OUT = (Varyings)0;

        float3 positionWS = TransformObjectToWorld(IN.positionOS.xyz);
        float3 normalWS = TransformObjectToWorldNormal(IN.normal.xyz);
        OUT.positionHCS = TransformWorldToHClip(ApplyShadowBias(positionWS, normalWS, _MainLightPosition.xyz));

        return OUT;
    }

    half4 shadowFrag(Varyings IN) : SV_Target
    {
        return 0;
    }
    ENDHLSL
}

#include "Packages/com.unity.render-pipelines.core/ShaderLibrary/Common.hlsl"  // real
#if defined(SHADER_API_MOBILE) || defined(SHADER_API_SWITCH)
#define HAS_HALF 1
#else
#define HAS_HALF 0
#endif

#if REAL_IS_HALF
#define real half
#define real2 half2
#define real3 half3
#define real4 half4

#include "Packages/com.unity.render-pipelines.universal/ShaderLibrary/Shadows.hlsl" // ApplyShadowBias
float3 ApplyShadowBias(float3 positionWS, float3 normalWS, float3 lightDirection)


real SampleShadowmap(TEXTURE2D_SHADOW_PARAM(ShadowMap, sampler_ShadowMap), float4 shadowCoord, ShadowSamplingData samplingData, half4 shadowParams, bool isPerspectiveProjection = true)
// 안쓰는 놈인데.. LerpWhiteTo를 들고있다..

#include "Packages/com.unity.render-pipelines.core/ShaderLibrary/CommonMaterial.hlsl" // LerpWhiteTo
real LerpWhiteTo(real b, real t)

쉐도우맵

1. Z-depth구하기

   1. 씬 렌더링

   2. Z-depth를 깊이버퍼에 저장한다(depth map)

      world > View[Light] > Proj[Light]
            Light's View Matrix > Light's Projection Matrix
      > transform NDC
      > transform texture Space
      

2. 그림자그리기

   1. 씬 렌더링

   2. 깊이버퍼랑 Z-depth 테스트

      if (fragment Z-depth > sampled Z-depth)
      {
          shadow : 0
      }
      else
      {
          lit     : 1
      }
      
      
      

• SSSM(Screen Space Shadow Map)

• Unity Shader - Custom SSSM(Screen Space Shadow Map) 自定义屏幕空间阴影图

• OpenGL - 阴影映射 - Tutorial 16 : Shadow mapping

Shadow Acne

DepthOnly

Pass
{
    Tags
    {
        "LightMode" = "DepthOnly"
    }

    ZWrite On
    ColorMask 0

    HLSLPROGRAM
    
    ...

    half4 shadowFrag(Varyings IN) : SV_Target
    {
        return 0;
    }
    ENDHLSL
}

Meta

• 라이트맵 구울때 사용.
• 디버깅용 내부툴 만들때 유용.

Pass
{
    Tags
    {
        "LightMode" = "DepthOnly"
    }

    ...
}

Ref

• URP 셰이더 코딩 튜토리얼 : 제 1편 - Unlit Soft Shadow
• URP Default Unlit Based to Custom Lighting
• urp管线的自学hlsl之路 第十篇 主光源阴影投射和接收
• builtin - Rendering 7 Shadows
• [Unity] URP Custom Shadow Shader 도전하기 : Frame Debugger로 원인 찾기(1/3)
  • [Unity] URP Custom Shadow Shader 도전하기 : Frame Debugger로 원인 찾기(2/3)
  • [Unity] URP Custom Shadow Shader 도전하기 : Frame Debugger로 원인 찾기(3/3)
• Reading a depth value from Unity's shadow map?

Dithering

• Dithering : 이미지에 Noise를 입히는 행위
  • 이미지의 디테일을 향상 시킬 수 있음(ex 계단현상(Banding) 완화)
• jitter : 흐트러짐

Ref

• https://www.ronja-tutorials.com/post/042-dithering/
• GDC2016 - Low Complexity, High Fidelity: The Rendering of INSIDE

Gemoetry

• VS > HS > TS > GS > FS
• shader model 4.0
• https://roystan.net/articles/grass-shader.html
• https://halisavakis.com/my-take-on-shaders-geometry-shaders/

Type

• https://docs.microsoft.com/en-us/windows/win32/direct3dhlsl/dx-graphics-hlsl-geometry-shader

[maxvertexcount(NumVerts)]
void ShaderName ( PrimitiveType DataType Name [ NumElements ], inout StreamOutputObject )
{
}

• https://docs.microsoft.com/en-us/windows/win32/direct3dhlsl/dx-graphics-hlsl-so-type

Barebone

#pragma vertex vert
#pragma fragment frag
#pragma geometry geom

struct FromVS
{
    float4 positionOS : POSITION
}

struct VStoGS
{
    float4 positionOS : SV_POSITION
}

struct GStoFS
{
    float4 positionCS : SV_POSITION
}

VStoGS vert(FromVS IN)
{
}


[maxvertexcount(3)] // 최대 얼마나 많이 vertex를 추가할 것인가.
void geom(triangle float4 IN[3] : SV_POSITION, uint pid : SV_PrimitiveID, inout TriangleStream<GStoFS> STREAM)
void geom(triangle VStoGS IN[3], uint pid : SV_PrimitiveID, inout TriangleStream<GStoFS> STREAM)
{
    GStoFS OUT1;
    GStoFS OUT2;
    GStoFS OUT3;

    STREAM.Append(OUT1);
    STREAM.Append(OUT2);
    STREAM.Append(OUT3);

    // https://docs.microsoft.com/en-us/windows/win32/direct3dhlsl/dx-graphics-hlsl-so-restartstrip
    // Ends the current primitive strip and starts a new strip
    STREAM.RestartStrip();
}

half4 frag(GStoFS IN) : SV_Target
{
}

Etc

• https://medium.com/@andresgomezjr89/rain-snow-with-geometry-shaders-in-unity-83a757b767c1
  • https://github.com/tiredamage42/RainSnowGeometryShader
• https://jayjingyuliu.wordpress.com/2018/01/24/unity3d-wireframe-shader/

Lightmap

Ref

• KGC2008 -Build Lightmap system

LPV

• Light Propagation Volumes

Ref

• https://ericpolman.com/2016/06/28/light-propagation-volumes/
  • https://blog.naver.com/catllage/221830338176

Noise

Ref

• Unity Shader Graph - Electricity Shader Effect Tutorial

• sorkelf - 펄린 노이즈(Perln Noise)

• http://staffwww.itn.liu.se/~stegu/simplexnoise/simplexnoise.pdf

• https://blog.csdn.net/yolon3000/article/details/78106203

• https://blog.csdn.net/candycat1992/article/details/50346469

• https://blog.csdn.net/qq_38275140/article/details/84205711

• https://blog.csdn.net/qq_38275140/article/details/84349192

• https://blog.csdn.net/qq_38275140/article/details/96859384

Ray

Ray Marching

• http://jamie-wong.com/2016/07/15/ray-marching-signed-distance-functions/#signed-distance-functions
• https://iquilezles.org/www/index.htm
• http://blog.three-eyed-games.com/2018/05/03/gpu-ray-tracing-in-unity-part-1/
• http://blog.three-eyed-games.com/2018/05/12/gpu-path-tracing-in-unity-part-2/
  • https://blog.csdn.net/qq_38275140/article/details/90239218
  • https://blog.csdn.net/qq_38275140/article/details/90269865
  • https://blog.csdn.net/qq_38275140/article/details/90345582
• Coding Adventure: Ray Marching
• The Art of Code - Ray Marching playlist
• Peer Play - Raymarching Shader Tutorial playlist
• https://rito15.github.io/posts/ray-marching/

Ref

• SDF.md
• Disney's Practical Guide to Path Tracing

SDF

• SDf : Signed Distance Field

예

• 폰트 렌더링
• LOD를 위한 알파맵
  • https://blog.naver.com/kite3h/222500918662
• Ray Marching

Ref

• SIGGRAPH2015 - Dynamic Occlusion with Signed Distance Fields
• NVScene 2015 Session: How to Create Content with Signed Distance Functions (Johann Korndörfer)
• https://iquilezles.org/www/index.htm
  • https://iquilezles.org/www/articles/smin/smin.htm
  • https://iquilezles.org/www/articles/distfunctions/distfunctions.htm

SRP Overview

Sobel Filter

Bloom(with DualFilter)

Light Streak

Screen Space Ambient Occlusion

Screen Space Global Illumination

Light Shaft

FXAA

Linear / Gamma

Unity에는 Gamma와 Linear를 선택할 수 있는 Color Space항목이 있다.

Edit> Project Settings> Player> Other Settings> Rendering> Color Space

Gamma / Linear Color Space 결과물 차이

일단 차이부터 알아보자.

• unity: Linear or gamma workflow

• 감마 색 공간에서의 블렌딩은 결과적으로 채도와 밝기가 과도하게 높습니다

이러한 조명 강도, 블렌딩 차이는 왜 생기는 것일까?

Gamma와 Linear의 관계

같은 RGB값이라도 Linear와 Gamma상태에서 보여지는 색이 다르다.

• 감마 보정(wiki: Gamma correction)

Gamma Encode/Decode

Linear는 무엇이고 Gamma는 무엇인가?

• wiki: sRGB : standard RGB color space.

Gamma / Linear Color Space 작업 환경

Linear와 Gamma가 왜 작업 결과물에 영향을 주는가?

• Gamma Pipeline에서는 빛의 연산 결과가 Linear환경에서 연산되고 모니터에는 Gamam가 적용된 상태로 표시된다.
• 빛의 연산 결과도 Linear환경으로 표시하려면, 모니터에 Gamma가 적용되어 어두워지기전에, 미리 밝게해두면 Linear한 빛의 연산 결과를 모니터에서 확인할 수 있게 된다.

이미지 제작 환경(감마 보정 환경)

Rendering - Gamma Color Space

• 문제점
  • 광원 감쇠
    • 감마 파이프라인에서는 셰이더 연산이 어둡게 보임.(셰이딩 값 참조)
  • 광원 강도 반응
    • 광원의 강도에 따라 선형적이 아닌 비 선형적으로 밝아지거나 어두워진다.
  • 블렌딩
    • 채도와 밝기가 과도하게 높아질 수 있음.

Rendering - Linear Color Space

• Gamma Correction
  • Gamma를 1.0으로 하는게 Gamma Correction이라고 하는 인터넷 문서들이 있는데, 그렇게 이해하면 안됨.
  • Wiki에는 Gamma Correction 자체가 Gamma 연산을 하는 걸로 정의되어 있음.
  • 게임에서는 출력장치로 출력하기 좋게 Gamma를 보정하는 작업을 Gamma Correction이라 칭하는게 좀 더 게임개발에 알맞음.
    • 모니터로 출력시 어둡게 출력되는데, 출력 전에 밝게 후보정하는 작업.

sRGB 보정

• sRGB 체크시 RGB채널에 대한 Gamma Decode을 수행시(단, A채널은 그대로).
• alpha에 대해선 체크 여부에 상관없이 decode적용 안함.
  • 남는 alpha채널에 Mask맵 같은걸 찡겨 넣을 수 있음.
  • 다만, 게임에서의 리니어 알파가 포토샵같이 비선형의 알파가 다름으로써 UI 알파블렌딩에서 문제가 됨.
    • UI의 알파처리는 따로 처리해줘야 함.

sRGB 미보정

• 컬러 텍스쳐를 sRGB 체크를 하지 않으면, 색이 떠보이게됨.
• ORM
  • Normal 텍스쳐는 수치 그 자체이므로 sRGB옵션 자체가 없음.
  • Roughness/Occlusion는 sRGB 체크를 해지해야함.
• 기타 수치 텍스쳐
  • flowmap 등등...

종합

좌 감마 // 우 리니어

Linear Color Space에서 작업시 주의할 점

• 플렛폼 지원
• sRGB로 보정이 필요한 텍스쳐 구분
• UI 텍스쳐의 Alpha값

플렛폼 지원

• OpenGL ES 2.0 이하는 Gamma만 지원.

  • https://blogs.unity3d.com/kr/2016/12/07/linear-rendering-support-on-android-and-ios/
  • https://developer.android.com/about/dashboards/index.html#OpenGL

• Linear를 위한 모바일 최소 사양

sRGB로 보정이 필요한 텍스쳐 구분

1. 데이터를 그대로 다루는것은 Linear로
2. 나머지 Albedo / Emmission는 sRGB 체크로 Gamma Decode 하도록

UI 텍스쳐의 Alpha값

• Linear환경으로 보다 풍부한 표현력을 얻었지만, UI색상의 알파블랜딩이 제대로 되지 않는 현상이 있다.
  • Linear개념으로 보면 정확한 계산이지만, 포토샵 작업자 관점에서는 아니다.
• sRGB옵션은 RGB에만 영향을 줌으로, Alpha를 처리함에 있어 추가 작업을 해 주어야 한다.

몇가지 방법이 있다

• 포토샵 강제 설정하거나...
• UI카메라와 SRP의 활용하거나..

Photoshop 설정

• 처음부터 Linear로 저장시켜버리자
• 포토샵 Color Settings > Advanced > Blend RPG Colors Using Gamma: 1.00
• 작업비용
  • 디자이너들은 작업하기 불편...
  • 프로그래머의 추가 작업 불필요.

UI카메라 + SRP

• UI카메라를 따로 두어서 UI Alpha에 미리 감마를 적용시켜주자.
• 그리고 Game카메라와 잘 섞어주자.

1. UITexture sRPG해제
   • sRGB상태 데이터 그대로 쓰고 Alpha만 어떻게 잘 처리할 것이다.
2. Main Camera
   1. Camera> Rendering> Culling Mask> Uncheck UI
3. UI Camera
   1. Camera> Render Type> OverLay
   2. Camera> Rendering> Renderer> GameUIFix
   3. Camera> Rendering> Culling Mask> UI
4. UI Canvas
   1. Canvas> Render Camera> UI Camera
5. PipelineAsset 설정
   1. _CameraColorTexture를 활용: Quality> Anti Aliasing (MSAA)> 2x 이상
6. RenderFeature 작성
   1. Game 카메라(Linear공간)를 Gamma 공간으로 변환
   2. 변환된 Game카메라의 출력결과 + UI카메라 출력결과
   3. 합친 결과(Gamma Space)를 Linear Space로 변경시켜주기
7. 새로운 Renderer 추가와 작성한 Feature추가
   1. General> Renderer List> Add Last GammaUIFix

// _CameraColorTexture 활성화는 PipelineAsset> Quality> Anti Aliasing (MSAA)> 2x 이상으로 하면 됨.

// 1. DrawUIIntoRTPass
//    cmd.SetRenderTarget(UIRenderTargetID);
//    cmd.ClearRenderTarget(clearDepth: true, clearColor: true, Color.clear);
// 2. BlitPass
//    cmd.Blit(DrawUIIntoRTPass.UIRenderTargetID, _colorHandle, _material);

float4 uiColor = SAMPLE_TEXTURE2D(_MainTex, sampler_MainTex, i.uv);
uiColor.a = LinearToGamma22(uiColor.a);

float4 mainColor = SAMPLE_TEXTURE2D(_CameraColorTexture, sampler_CameraColorTexture, i.uv);
mainColor.rgb = LinearToGamma22(mainColor.rgb);

float4 finalColor;
finalColor.rgb = lerp(mainColor.rgb, uiColor.rgb, uiColor.a);
finalColor.rgb = Gamma22ToLinear(finalColor.rgb);
finalColor.a = 1;

Ref

• [GDC2010] GDCValue: Uncharted-2-HDR
  • slideshare: Lighting Shading by John Hable
• 정종필 linear/gamma 설명
  • 정종필님의 설명 볼때 주의점
    • 텍스쳐와 빛의 연산을 뚜렸히 분리하여 설명하지 않고, 리니어라는 것을 강조하기 위해 그래프를 옆에두고 설명함
    • 처음에는 이미지가 다른 걸 보고 아 그렇구나 해서 이해한것으로 착각하기 쉬운데, 텍스쳐와 빛의 연산을 분리하지 않고 설명해서 나중에 더 햇갈릴 수 있음.
    • Uncharted-2-HDR를 확인 할것.
  • 정종필 - Gamma Color space와 Linear Color space란?
    • 텍스쳐 저장 공간 설명 View/HDD/Display
  • 정종필 - 라이팅과 셰이더에서 연산을 위한 선형 파이프라인
  • 정종필 - 유니티 셰이더에서 sRGB/Linear 사용 및 응용
• GPU Gems 3 - Chapter 24. The Importance of Being Linear
• Article - Gamma and Linear Spaces
• [데브루키] Color space gamma correction
• 선형(Linear) 렌더링에서의 UI 작업할때 요령
• 201205 Unity Linear color space에서 UI의 alpha 값이 바뀌는 문제에 대하여..
• 3D scene need Linear but UI need Gamma
• https://nbertoa.wordpress.com/2016/06/20/gamma-correction/
• https://chrisbrejon.com/cg-cinematography/chapter-1-color-management/

셰이더 모델과 플렛폼 관계

• wiki: OpenGL ES
• wiki: Metal
• wiki: Vulkan

Shader Model

• https://docs.unity3d.com/Manual/SL-ShaderCompileTargets.html
• Metal
  • geometry 지원 여부(일단 5.0까지 지원 안함)
  • compute에서 GetDimensions 지원 안함
• es3.1
  • 4 compute buffer만 보장함.

Deferred support

• 최소 셰이더 모델 4.5이상
• OpenGL기반 API에서는 지원하지 않음.

안드로이드와 그래픽 라이브러리

Linear지원 사양

레퍼런스 디바이스

Android

• 모바일 디바이스는 PC와는 다르게 GPU전용 VRAM없이 그냥 shared memory임.

AI때문에 HBM(High Bandwidth Memory)이 모바일에 들어올 정도가 된다면....

ios

• https://iosref.com/ios
• https://en.wikipedia.org/wiki/IOS_version_history

Ref

• https://developer.android.com/guide/topics/graphics/opengl?hl=ko
• https://source.android.com/setup/start/build-numbers?hl=ko
• https://developer.android.com/ndk/guides/graphics/getting-started?hl=ko
• https://forum.unity.com/threads/severe-banding-with-webgl-on-chrome.310326/#post-2035482
• https://blog.mozilla.org/futurereleases/2015/03/03/an-early-look-at-webgl-2/
• SRP Batcher : https://blog.unity.com/kr/technology/srp-batcher-speed-up-your-rendering
• https://docs.unity3d.com/Packages/com.unity.render-pipelines.universal@12.0/manual/rendering/deferred-rendering-path.html

SRP (Scriptable Render Pipeline)

https://docs.unity3d.com/Manual/render-pipelines-feature-comparison.html

RenderPipelineAsset.asset

• 유니티에서 그래픽스 파이프 라인을 관리한다. 여러 Renderer를 가질 수 있다.

// 런타임 렌더파이프라인에셋 교체

// Edit> Project Settings> Scriptable Render Pipeline Settings

// 혹은, 이하 스크립트
public RenderPipelineAsset _renderPipelineAsset;

GraphicsSettings.renderPipelineAsset = _renderPipelineAsset;

// 클래스를 만들어서 사용자 렌더파이프라인에셋 만들기
[CreateAssetMenu(menuName = "Rendering/CustomRenderPipelineAsset")]
public class CustomRenderPipelineAsset : RenderPipelineAsset
{
    protected override RenderPipeline CreatePipeline()
    {
        return new CustomRenderPipeline();
    }
}

public class CustomRenderPipeline : RenderPipeline
{
    protected override void Render(ScriptableRenderContext context, Camera[] cameras);
}

Example

RenderPipeline

Pass
{
    Tags
    {
        // LightMode 태그는 라이팅 파이프 라인에서 패스의 역할을 정의.
        "LightMode" = "CustomLightMode"
    }
}

[CreateAssetMenu(menuName = "Rendering/CustomRenderPipelineAsset")]
public class CustomRenderPipelineAsset : RenderPipelineAsset
{
    protected override RenderPipeline CreatePipeline()
    {
        return new CustomRenderPipeline();
    }
}

// ==========================================================================
public class CustomRenderPipeline : RenderPipeline
{
    CustomRenderer _renderer = new CustomRenderer();

    protected override void Render(ScriptableRenderContext context, Camera[] cameras)
    {
        foreach (Camera cam in cameras)
        {
            _renderer.Render(ref context, cam);
        }
    }
}

// ==========================================================================
public class CustomRenderer
{
    readonly static ShaderTagId unlitShaderTagId = new ShaderTagId("CustomLightMode");

    public void Render(ref ScriptableRenderContext context, Camera cam)
    {
        // ...
        context.Submit();                 // 실행
    }
}

context.SetupCameraProperties(camera); // cmd전에 설정해주자(빠른 지우기)
var cmd = new CommandBuffer();
cmd.ClearRenderTarget
context.ExecuteCommandBuffer(cmd); // enqueue cmd
cmd.Release();
context.Submit();                 // 실행

var cmd = new CommandBuffer();
cmd.BeginSample(string sampleName); // profiler begin
cmd.EndSample(string sampleName);   // profiler end

// 컬링
if (!CulllResults.GetCullingParameters(camera, out ScriptableCullingParameters cullingParams))
{
    continue;
}
CullResults cullingResults = context.Cull(ref cullingParams);

SortingSettings sortingSettings = new SortingSettings(cam);
DrawingSettings drawingSettings = new DrawingSettings(unlitShaderTagId, sortingSettings);
FilteringSettings filteringSettings = new FilteringSettings(RenderQueueRange.opaque);

context.DrawRenderers(cullingResults, ref drawingSettings, ref filteringSettings);
context.DrawRenderers             // 렌더링
context.DrawSkybox(camera)        // Skybox

// cs
var cmd = new CommandBuffer();
cmd.SetGlobalVector("_LightDir", new Vector4(0, 1, 0, 0));
context.ExecuteCommandBuffer(cmd);
cmd.Release();

// shader
CBUFFER_START(_Light) // CommandBuffer에서 전송됨
float4 _LightDir;
CBUFFER_END

• CBUFFER_START(UnityPerMaterial) // 메터리얼별
• CBUFFER_START(UnityPerDraw) // draw별
• https://blogs.unity3d.com/kr/2019/02/28/srp-batcher-speed-up-your-rendering/

/// Render Texture 사용.

// RenderTarget Id가 필요
int _TmpShaderProperty = Shader.PropertyToID("_TmpShaderProperty");

{
    var cmd = new CommandBuffer();
    // https://docs.unity3d.com/ScriptReference/Rendering.CommandBuffer.GetTemporaryRT.html
    // GetTemporaryRT(int nameID, int width, int height, int depthBuffer, FilterMode filter, RenderTextureFormat format, RenderTextureReadWrite readWrite, int antiAliasing, bool enableRandomWrite);
    // GetTemporaryRT(int nameID, RenderTextureDescriptor desc, FilterMode filter);
    cmd.GetTemporaryRT(_TmpShaderProperty, )
    cmd.SetRenderTarget(RTID);
    cmd.ClearRenderTarget;
    context.ExecuteCommandBuffer(cmd);
    cmd.Release();
}

{
    var cmd = new CommandBuffer();
    cmd.Blit(RTID, BuiltinRenderTextureType.CameraTarget);
    cmd.ReleaseTemporaryRT(TemporaryRTID);
    context.ExecuteCommandBuffer(cmd);
    cmd.Release();
}

ScriptableRenderPass

...

CullResults cr = CullResults.Cull(ref cullingParams, context);
InitializeRenderingData(settings, ref cameraData, ref cullResults, out var renderingData);
renderer.Setup(context, ref renderingData); // RenderPass 쌓기.
renderer.Execute(context, ref renderingData);

public struct RenderingData
{
    public CullingResults cullResults;
    public CameraData cameraData;
    public LightData lightData;
    public ShadowData shadowData;
    public PostProcessingData postProcessingData;
    public bool supportsDynamicBatching;
    public PerObjectData perObjectData;
    public bool postProcessingEnabled;
}

- ScriptableRenderContext
- ScriptableRenderer (abstract class)
  - public abstract void Setup(ScriptableRenderContext context, ref RenderingData renderingData);

- ScriptableRendererFeature
RenderingData
RenderPassEvent
RenderTargetHandle

SubmitRenderRequest

• https://docs.unity3d.com/ScriptReference/Camera.SubmitRenderRequest.html
• https://docs.unity3d.com/ScriptReference/Rendering.RenderPipeline.SubmitRenderRequest.html
  • UniversalRenderPipeline은 다음을 지원합니다.
    • ScriptableRenderer.StandardRequest: 이 요청 유형은 전체 URP 카메라 스택을 렌더링하고 결과를 지정된 대상에 출력합니다. Base Camera에서만 호출할 수 있습니다.
    • UniversalRenderPipeline.SingleCameraRequest: 이 요청 유형은 단일 URP 카메라를 렌더링하고 그 결과를 지정된 대상에 출력합니다.
• https://docs.unity3d.com/ScriptReference/Rendering.RenderPipeline.ProcessRenderRequests.html

SubmitRenderRequest하면 파이프라인의 Rendering.RenderPipeline.ProcessRenderRequests 이 실행됨.

Ref

• https://github.com/cinight/CustomSRP/tree/master/Assets

• https://blogs.unity3d.com/2018/01/31/srp-overview/

• https://blogs.unity3d.com/kr/2019/02/28/srp-batcher-speed-up-your-rendering/

• https://docs.unity3d.com/Manual/ScriptableRenderPipeline.html

• https://docs.unity3d.com/Manual/srp-creating-render-pipeline-asset-and-render-pipeline-instance.html

• https://github.com/cinight/CustomSRP

• catlikecoding - Custom SRP

• 2019 - Unity SRP와 LWRP에 대한 모든 것!

• 2020 - Universal RenderPipeline의 Custom RenderPass를 활용하여 렌더링 기능을 구현해보자 Track1-2

• 2020 - Dev Weeks: URP 기본 구성과 흐름

  • Dev Weeks 2020.5. 세션자료

• 2020 - Dev Weeks: URP 셰이더 뜯어보기

  • Dev Weeks 2020.6. 세션자료

ShaderLab

// http://docs.unity3d.com/Manual/SL-Shader.html
Shader <shader-name>
{

    HLSLINCLUDE
    // ...
    ENDHLSL

    // http://docs.unity3d.com/Manual/SL-Properties.html
    Properties
    {
        _PropertyName ("displayed name", <property-type>) = <property-default-value>
    }

    // http://docs.unity3d.com/Manual/SL-SubShader.html
    SubShader
    {
        // http://docs.unity3d.com/Manual/SL-SubshaderTags.html
        Tags
        {
            // 주의. Pass의 Tag랑 다름
            <tag-name> = <tag-value>
        }
        
        // http://docs.unity3d.com/Manual/SL-ShaderLOD.html
        LOD <lod-number>
            
        // http://docs.unity3d.com/Manual/SL-UsePass.html
        UsePass "Shader/Name"
        
        
        // http://docs.unity3d.com/Manual/SL-Pass.html
        Pass
        {
            Name "PassName"
            
            // https://docs.unity3d.com/Manual/SL-PassTags.html
            Tags
            {
                // 주의. Subshader의 Tag랑 다름
                <tag-name> = <tag-value>
            }

            // https://docs.unity3d.com/Manual/SL-Stencil.html
            Stencil
            {
            }
            
            // http://docs.unity3d.com/Manual/SL-CullAndDepth.html
            Cull <Back | Front | Off> // default: Back 
            ZTest <(Less | Greater | LEqual | GEqual | Equal | NotEqual | Always)> // default: LEqual 
            ZWrite <On | Off> // default: On 
            Offset <OffsetFactor>, <OffsetUnits>

            // http://docs.unity3d.com/Manual/SL-Blend.html
            Blend <SourceBlendMode> <DestBlendMode>
            BlendOp <colorOp> // Instead of adding blended colors together, carry out a different operation on them
            BlendOp <colorOp, alphaOp> // Same as above, but use different blend operation for color (RGB) and alpha (A) channels.
            AlphaToMask <On | Off>

            ColorMask <RGB | A | 0 | any combination of R, G, B, A>
         
            HLSLPROGRAM
            ENDHLSL
        }
    }

    // http://docs.unity3d.com/Manual/SL-Fallback.html
    Fallback Off
    Fallback <other-shader-name>

    // http://docs.unity3d.com/Manual/SL-CustomEditor.html
    // http://docs.unity3d.com/Manual/SL-CustomMaterialEditors.html
    CustomEditor <custom-editor-class-name>
}

Properties

Float           | float  |
Range(min, max) | float  |

Vector          | float4 | (x, y, z, w)
Color           | float4 | (r, g, b, a)

2D              | float4 | "", "white", "black", "gray", "bump" // for     power of 2 size
Rect            | float4 | "", "white", "black", "gray", "bump" // for non-power of 2 size

Cube            | float4 | "", "white", "black", "gray", "bump"

// 주의해야할게 2D/Rect/Cube는 linear설정 관계없이 sRGB로 된다.
// ex) pow(gray, 2.2);

// color string
red
black
white
gray
grey
linearGray
linearGrey
grayscaleRamp
greyscaleRamp
bump
blackCube
lightmap
unity_Lightmap
unity_LightmapInd
unity_ShadowMask
unity_DynamicLightmap
unity_DynamicDirectionality
unity_DynamicNormal
unity_DitherMask
_DitherMaskLOD
_DitherMaskLOD2D
unity_RandomRotation16
unity_NHxRoughness
unity_SpecCube0
unity_SpecCube1

Properties attributes

[HideInInspector]
[NoScaleOffset]   - name##_ST 사용안할때
[Normal]          - 텍스쳐 설정 normal아니면 경고
[HDR]

[Gamma]           - indicates that a float/vector property is specified as sRGB value in the UI
(just like colors are), and possibly needs conversion according to color space used. See Properties in Shader Programs.
[PerRendererData]  - indicates that a texture property will be coming from per-renderer data in the form of a MaterialPropertyBlock. Material inspector changes the texture slot UI for these properties.

[MainTexture]
[MainColor]

SubShader's Tags

SubShader
{
    // http://docs.unity3d.com/Manual/SL-SubshaderTags.html
    Tags
    {
        // 주의. Pass의 Tag랑 다름
        "RenderPipeline" = "UniversalRenderPipeline"
        "RenderType" = "Opaque"
        "Queue" = "Geometry"
    }
}

// ex) cutout() 셰이더
Tags
{
    "RenderPipeline" = "UniversalRenderPipeline"
    "Queue" = "AlphaTest"
    "RenderType" = "TransparentCutout"
    "IgnoreProjector" = "True"
}

RenderPipeline

• https://docs.unity3d.com/2021.1/Documentation/ScriptReference/Shader-globalRenderPipeline.html

Queue

• 렌더링 순서 지정. Geometry+1, Geometry-1 과 같이 가중치 적용가능

RenderType

• 그룹을 짓는것. 해당 그룹의 셰이더를 바꿔 랜더링 할 수 있음.
  • 예를들어 Opaque의 노말버퍼를 만들고 싶을때 RenderWithShader(Shader shader, "Opaque") 이런 식으로..
• https://docs.unity3d.com/2021.1/Documentation/Manual/SL-ShaderReplacement.html
• https://github.com/Unity-Technologies/Graphics/blob/master/com.unity.shadergraph/Editor/Generation/Enumerations/RenderType.cs

IgnoreProjector

• https://docs.unity3d.com/Manual/class-Projector.html
  • URP is not compatible with the Projector component. URP does not currently include an alternative solution.
• https://github.com/Anatta336/driven-decals
• https://github.com/nyahoon-games/ProjectorForLWRP

Pass's Tags

Pass
{
    // http://docs.unity3d.com/Manual/SL-SubshaderTags.html
    Tags
    {
        // 주의. SubShader의 Tag랑 다름
        "LightMode" = "UniversalForward"
    }
}

LihgtMode

• https://docs.unity3d.com/Packages/com.unity.render-pipelines.universal@10.3/manual/urp-shaders/urp-shaderlab-pass-tags.html

Blend

• 대표적인 Blend 옵션 조합

Offset

Offset Factor, Units

Factor 및 units 파라미터 2개를 사용하여 뎁스 오프셋을 지정. Factor 는 폴리곤의 X 또는 Y를 기준으로 최대 Z 기울기를 스케일하고 units 는 최소 분석 가능 뎁스 버퍼 값을 스케일 하게된다.이를 통해 두개의 오브젝트가 겹칠경우 특정 오브젝트를 앞에 그리게 조절할 수 있다.

AlphaToMask

AlphaToMask On

포워드 렌더링을 사용하는 멀티샘플 안티앨리어싱(MSAA, QualitySettings 참조)을 사용하는 경우 알파 투 커버리지 기능을 사용해 알파채널 텍스쳐의 AA를 적용할 수 있다. MSAA 가 메시 외곽에만 AA를 적용하고 알파처럼 텍스쳐 안의 이미지에 대한 AA를 적용할수 없기 때문에 이와 같은 방식으로 AA를 적용한다

HLSLPROGRAM

HLSLPROGRAM
// https://docs.unity3d.com/Manual/SL-ShaderPrograms.html
#pragma target 3.5

#pragma vertex   <func>
#pragma fragment <func>
#pragma geometry <func> // target 4.0
#pragma hull     <func> // target 5.0
#pragma domain   <func> // target 5.0


#pragma only_renderers      <renderers>
#pragma exclude_renderers   <renderers>
// renderers
// d3d11    |Direct3D 11/12
// glcore   |OpenGL 3.x/4.x
// gles     |OpenGL ES 2.0
// gles3    |OpenGL ES 3.x
// metal    |iOS
// /Mac     |Metal
// vulkan   |Vulkan
// d3d11_9x |Direct3D 11 9.x , as commonly used on WSA platforms
// xboxone  |Xbox One
// ps4      |PlayStation 4
// n3ds     |Nintendo 3DS
// wiiu     |Nintendo Wii U

#pragma multi_compile           ...
#pragma multi_compile_local     ...
#pragma shader_feature          ...
#pragma shader_feature_local    ...
#include 

ENDHLSL

Built-in(Legacy)

// Built-in(Legacy) 볼필요없는것.

CGINCLUDE
ENDCG

Pass
{
    Lighting On | Off
    Material { Material Block }
    SeparateSpecular On | Off
    Color Color-value
    ColorMaterial AmbientAndDiffuse | Emission

    Fog { Fog Block }

    AlphaTest (Less | Greater | LEqual | GEqual | Equal | NotEqual | Always) CutoffValue

    SetTexture textureProperty { combine options }

    GrabPass { } // _GrabTexture 
    GrabPass { "TextureName" } 

    CGPROGRAM
    #pragma surface surfaceFunction lightModel [optionalparams]


    // https://docs.unity3d.com/Manual/SL-ShaderPrograms.html
    // The following compilation directives don’t do anything and can be safely removed:
    #pragma glsl
    #pragma glsl_no_auto_normalization
    #pragma profileoption
    #pragma fragmentoption

    ENDCG
}

URP (Universal Render Pipeline)

• 기존 Built-in(Legacy) 쉐이더의 include 경로 및 함수명등 바뀜

• SBR Batcher 사용가능하게 바뀜.

• 1패스 1라이트방식 => 1패스 16개 라이트 지원

• Upscale

  • UniversalRenderPipelineAsset > Quality > Render Scale

https://unity.com/kr/resources/introduction-universal-render-pipeline-for-advanced-unity-creators-2022lts

sample

Varyings OUT;
ZERO_INITIALIZE(Varyings, OUT);

OUT.positionCS    = TransformObjectToHClip(IN.positionOS.xyz);
OUT.positionWS    = TransformObjectToWorld(IN.positionOS.xyz);
OUT.N             = TransformObjectToWorldNormal(IN.normal);
OUT.uv            = TRANSFORM_TEX(IN.uv, _MainTex);
OUT.fogCoord      = ComputeFogFactor(IN.positionOS.z);          // float
OUT.shadowCoord   = TransformWorldToShadowCoord(OUT.positionWS);// float4

VertexPositionInputs vertexInput = GetVertexPositionInputs(v.vertex.xyz);
OUT.shadowCoord = GetShadowCoord(vertexInput);

Light mainLight = GetMainLight();
Light mainLight = GetMainLight(shadowCoord);

half3 ambient = SampleSH(IN.normal);

half3 cameraWS = GetCameraPositionWS();

// GPU instancing
#pragma multi_compile_instancing

// Fog
#pragma multi_compile_fog

// Light & Shadow
#pragma multi_compile _ _MAIN_LIGHT_SHADOWS
#pragma multi_compile _ _MAIN_LIGHT_SHADOWS_CASCADE
#pragma multi_compile _ _ADDITIONAL_LIGHTS
#pragma multi_compile _ _ADDITIONAL_LIGHTS_CASCADE
#pragma multi_compile _ _SHADOWS_SOFT

// LightMap
#pragma multi_compile _ DIRLIGHTMAP_COMBINED
#pragma multi_compile _ LIGHTMAP_ON

SBR Batcher / GPU인스턴싱

SRP Batcher가 추가됨으로써, 동적오브젝트가 많아져도 좋은 퍼포먼스 유지하는게

• https://docs.unity3d.com/Manual/GPUInstancing.html

// For SRP Batcher

CBUFFER_START(UnityPerMaterial)
...
CBUFFER_END

// for GPU instancing

struct Attributes
{
    UNITY_VERTEX_INPUT_INSTANCE_ID
};

struct Varyings
{
    UNITY_VERTEX_INPUT_INSTANCE_ID
    UNITY_VERTEX_OUTPUT_STEREO // for VR
};

Varyings vert(Attributes IN)
{
    Varyings OUT;
    UNITY_SETUP_INSTANCE_ID(IN);
    UNITY_TRANSFER_INSTANCE_ID(IN, OUT); 
};

half4 frag(Varyings IN) : SV_Target
{
    UNITY_SETUP_INSTANCE_ID(IN); 
}

hlsl

com.unity.render-pipelines.core/ShaderLibrary

Common.hlsl

// com.unity.render-pipelines.core/ShaderLibrary/Common.hlsl

#elif defined(SHADER_API_D3D11)
#include "Packages/com.unity.render-pipelines.core/ShaderLibrary/API/D3D11.hlsl"

Macros.hlsl

// com.unity.render-pipelines.core/ShaderLibrary/Macros.hlsl

#define PI          3.14159265358979323846 // PI
#define TWO_PI      6.28318530717958647693 // 2 * PI
#define FOUR_PI     12.5663706143591729538 // 4 * PI
#define INV_PI      0.31830988618379067154 // 1 / PI
#define INV_TWO_PI  0.15915494309189533577 // 1 / (2 * PI)
#define INV_FOUR_PI 0.07957747154594766788 // 1 / (4 * PI)
#define HALF_PI     1.57079632679489661923 // PI / 2
#define INV_HALF_PI 0.63661977236758134308 // 2 / PI
#define LOG2_E      1.44269504088896340736 // log2e
#define INV_SQRT2   0.70710678118654752440 // 1 / sqrt(2)
#define PI_DIV_FOUR 0.78539816339744830961 // PI / 4

#define TRANSFORM_TEX(tex, name) ((tex.xy) * name##_ST.xy + name##_ST.zw)
#define GET_TEXELSIZE_NAME(name) (name##_TexelSize)

(U ,V)

V
(0,1)       (1,1)
   +----+----+
   |    |    |
   +----+----+
   |    |    |
   +----+----+
(0,0)       (1,0) U

API/(renderer).hlsl

// com.unity.render-pipelines.core/ShaderLibrary/API/D3D11.hlsl

#define CBUFFER_START(name) cbuffer name {
#define CBUFFER_END };

#define ZERO_INITIALIZE(type, name) name = (type)0;


#define TEXTURE2D(textureName)                Texture2D textureName
#define TEXTURE2D_ARRAY(textureName)          Texture2DArray textureName
#define TEXTURECUBE(textureName)              TextureCube textureName
#define SAMPLER(samplerName)                  SamplerState samplerName

#define SAMPLE_TEXTURE2D(textureName, samplerName, coord2)                               textureName.Sample(samplerName, coord2)
#define SAMPLE_TEXTURE2D_LOD(textureName, samplerName, coord2, lod)                      textureName.SampleLevel(samplerName, coord2, lod)

#define SAMPLE_TEXTURE2D_ARRAY(textureName, samplerName, coord2, index)                  textureName.Sample(samplerName, float3(coord2, index))
#define SAMPLE_TEXTURE2D_ARRAY_LOD(textureName, samplerName, coord2, index, lod)         textureName.SampleLevel(samplerName, float3(coord2, index), lod)

#define SAMPLE_TEXTURECUBE(textureName, samplerName, coord3)                             textureName.Sample(samplerName, coord3)
#define SAMPLE_TEXTURECUBE_LOD(textureName, samplerName, coord3, lod)                    textureName.SampleLevel(samplerName, coord3, lod)

#define SAMPLE_DEPTH_TEXTURE(textureName, samplerName, coord2)          SAMPLE_TEXTURE2D(textureName, samplerName, coord2).r
#define SAMPLE_DEPTH_TEXTURE_LOD(textureName, samplerName, coord2, lod) SAMPLE_TEXTURE2D_LOD(textureName, samplerName, coord2, lod).r

Packing.hlsl

// com.unity.render-pipelines.core/ShaderLibrary/Packing.hlsl
real3 UnpackNormal(real4 packedNormal)

com.unity.render-pipelines.universal/ShaderLibrary/

universal/ShaderLibrary/Core.hlsl

#include "Packages/com.unity.render-pipelines.core/ShaderLibrary/Packing.hlsl"

struct VertexPositionInputs
{
    float3 positionWS; // World space position
    float3 positionVS; // View space position
    float4 positionCS; // Homogeneous clip space position
    float4 positionNDC;// Homogeneous normalized device coordinates
};


struct VertexNormalInputs
{
    real3 tangentWS;
    real3 bitangentWS;
    float3 normalWS;
};

#include "Packages/com.unity.render-pipelines.universal/ShaderLibrary/ShaderVariablesFunctions.hlsl"

ShaderVariablesFunctions.hlsl

// com.unity.render-pipelines.universal/ShaderLibrary/ShaderVariablesFunctions.hlsl

VertexPositionInputs GetVertexPositionInputs(float3 positionOS)
{
    VertexPositionInputs input;
    input.positionWS = TransformObjectToWorld(positionOS);
    input.positionVS = TransformWorldToView(input.positionWS);
    input.positionCS = TransformWorldToHClip(input.positionWS);

    float4 ndc = input.positionCS * 0.5f;
    input.positionNDC.xy = float2(ndc.x, ndc.y * _ProjectionParams.x) + ndc.w;
    input.positionNDC.zw = input.positionCS.zw;

    return input;
}

VertexNormalInputs GetVertexNormalInputs(float3 normalOS)
VertexNormalInputs GetVertexNormalInputs(float3 normalOS, float4 tangentOS)

float3 GetCameraPositionWS()

float3 GetWorldSpaceViewDir(float3 positionWS)

real ComputeFogFactor(float z)

half3 MixFog(half3 fragColor, half fogFactor)

half LinearDepthToEyeDepth(half rawDepth)

SpaceTransforms.hlsl

// com.unity.render-pipelines.core/ShaderLibrary/SpaceTransforms.hlsl

float3 TransformObjectToWorld(float3 positionOS) // OS > WS
float4 TransformObjectToHClip(float3 positionOS) // OS > HCS

float3 TransformWorldToView(float3 positionWS)   // WS > VS
float4 TransformWViewToHClip(float3 positionVS)  // VS > HCS

float3 TransformWorldToObject(float3 positionWS) // WS > OS

float3 TransformObjectToWorldDir(float3 dirOS, bool doNormalize = true) // normalOS > normalWS

Varaiable

CheatSheet

• based on [built-in] UnityShadersCheatSheet.shader

// Cheetsheet for URP

// ref: https://docs.unity3d.com/Manual/SL-Shader.html
Shader "Name"
{
    HLSLINCLUDE
    // ...
    ENDHLSL

    // ref: https://docs.unity3d.com/Manual/SL-Properties.html
    Properties
    {
        _Name ("display name", Float)               = number
        _Name ("display name", Int)                 = number
        _Name ("display name", Range (min, max))    = number
        _Name ("display name", Color)               = (number,number,number,number)
        _Name ("display name", Vector)              = (number,number,number,number)

        _Name ("display name", 2D)      = "default-ColorString" {} // power of 2
        _Name ("display name", Rect)    = "default-ColorString" {} // non-power of 2
        _Name ("display name", Cube)    = "default-ColorString" {}
        _Name ("display name", 3D)      = "default-ColorString" {}

        // ## Property Attribute
        // ref: https://docs.unity3d.com/ScriptReference/MaterialPropertyDrawer.html
        // |                                                                          |                                                                                                                                                                                                     |
        // | ------------------------------------------------------------------------ | --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
        // | [HideInInspector]                                                        | 머테리얼 인스펙터에서 표시안함                                                                                                                                                                      |
        // | [NoScaleOffset]                                                          | 텍스쳐의 tiling/offset 필드를 표시안함                                                                                                                                                              |
        // | [Normal]                                                                 | 텍스쳐 설정 normal아니면 경고                                                                                                                                                                       |
        // | [HDR]                                                                    | 텍스쳐 설정 HDR 아니면 경고                                                                                                                                                                         |
        // | [Gamma]                                                                  | indicates that a float/vector property is specified as sRGB value in the UI (just like colors are), and possibly needs conversion according to color space used. See Properties in Shader Programs. |
        // | [PerRendererData]                                                        | indicates that a texture property will be coming from per-renderer data in the form of a MaterialPropertyBlock. Material inspector changes the texture slot UI for these properties.                |
        // | [Toggle]                                                                 |                                                                                                                                                                                                     |
        // | [Toggle(ENABLE_FANCY)] _Fancy ("Fancy?", Float) = 0                      | Will set "ENABLE_FANCY" shader keyword when set.                                                                                                                                                    |
        // | [ToggleOff]                                                              |                                                                                                                                                                                                     |
        // | [ToggleOff(DISABLE_EXAMPLE_FEATURE)]                                     |                                                                                                                                                                                                     |
        // | [Enum(UnityEngine.Rendering.BlendMode)] _Blend ("Blend mode", Float) = 1 | blend modes selection.                                                                                                                                                                              |
        // | [Enum(UnityEngine.Rendering.CompareFunction)] _ZTest("ZTest", Float) = 0 |                                                                                                                                                                                                     |
        // | [Enum(UnityEngine.Rendering.CullMode)] _CullMode("Cull Mode", Int) = 0   |                                                                                                                                                                                                     |
        // | [KeywordEnum(None, Add, Multiply)] _Overlay ("Overlay mode", Float) = 0  | Display a popup with None,Add,Multiply choices. Each option will set _OVERLAY_NONE, _OVERLAY_ADD, _OVERLAY_MULTIPLY shader keywords.                                                                |
        // | [MainTexture]                                                            |                                                                                                                                                                                                     |
        // | [MainColor]                                                              |                                                                                                                                                                                                     |
        // | [Header(A group of things)]                                              |                                                                                                                                                                                                     |
        // | [PowerSlider(3.0)]                                                       |                                                                                                                                                                                                     |
        // | [IntRange]                                                               |                                                                                                                                                                                                     |
        
        // Later on in the shader’s fixed function parts, property values can be accessed using property name in square brackets: [name] e.g. Blend [_SrcBlend] [_DstBlend].

        // ## ColorString
        // "red"
        // "black"         ""
        // "white"
        // "gray"          "grey"
        // "linearGray"    "linearGrey"
        // "grayscaleRamp" "greyscaleRamp"
        // "bump"
        // "blackCube"
        // "lightmap"
        // "unity_Lightmap"
        // "unity_LightmapInd"
        // "unity_ShadowMask"
        // "unity_DynamicLightmap"
        // "unity_DynamicDirectionality"
        // "unity_DynamicNormal"
        // "unity_DitherMask"
        // "_DitherMaskLOD"
        // "_DitherMaskLOD2D"
        // "unity_RandomRotation16"
        // "unity_NHxRoughness"
        // "unity_SpecCube0"
        // "unity_SpecCube1"
    }

    // ref: https://docs.unity3d.com/Manual/SL-SubShader.html
    SubShader
    {
        // ref: https://docs.unity3d.com/Manual/SL-SubShaderTags.html
        Tags
        {
            "TagName1" = "Value1"
            "TagName2" = "Value2"
        }

        // Queue tag: 렌더링 순서 지정. `Geometry+1`, `Geometry-1` 과 같이 가중치 적용가능
        // | Queue       | [min,  max]   | default | order                | etc                      |
        // | ----------- | ------------- | ------- | -------------------- | ------------------------ |
        // | Background  | [0    , 1499] | 100     | render first -> back |                          |
        // | <Geometry>  | [1500 , 2399] | 2000    |                      | Opaque는 이쪽에          |
        // | AlphaTest   | [2400 , 2699] | 2450    |                      | AlphaTest는 이쪽에       |
        // | Transparent | [2700 , 3599] | 3000    | render back -> front | AlphaBlend는 이쪽에      |
        // | Overlay     | [3600 , 5000] | 4000    | render last -> front |                          |

        // RenderType tag : 그룹을 짓는것. 해당 그룹의 셰이더를 바꿔 랜더링 할 수 있음.
        
        // | RenderType        |                                            |
        // | ----------------- | ------------------------------------------ |
        // | Background        | Skybox 쉐이더                              |
        // | Opaque            | 대부분의 쉐이더                            |
        // | TransparentCutout | 마스킹 된 투명 쉐이더(2pass 식물쉐이더 등) |
        // | Transparent       | 투명한 쉐이더                              |
        // | Overlay           | 후광(Halo), 플레어(Flare)                  |

        // 기타 SubShader의 Tags
        // "DisableBatching"      = "(True | <False> | LODFading)"
        // "ForceNoShadowCasting" = "(True | <False>)"
        // "CanUseSpriteAtlas"    = "(<True> | False)"
        // "PreviewType"          = "(<Sphere> | Plane | Skybox)"

        // ref: https://docs.unity3d.com/Manual/SL-ShaderLOD.html
        LOD <lod-number>

        // ref: https://docs.unity3d.com/Manual/SL-UsePass.html
        // 주어진 이름의 셰이더의 (첫번째 SubShader의) 모든 패스들이 삽입됨.
        UsePass "Shader/Name"

        // ref: https://docs.unity3d.com/Manual/SL-Pass.html
        Pass
        {
            Name "Pass Name"

            // ref: https://docs.unity3d.com/Manual/SL-PassTags.html
            // 주의. `Pass의 Tag`는 `SubShader의 Tag`랑 다름
            Tags
            {
                "TagName1" = "Value1"
                "TagName2" = "Value2"
            }

            // LightMode tag : 
            // | LightMode            |                                               |
            // | -------------------- | --------------------------------------------- |
            // | <SRPDefaultUnlit>    | draw an extra Pass  (ex. Outline)    |
            // | UniversalForward     | Forward Rendering                             |
            // | UniversalGBuffer     | Deferred Rendering                            |
            // | UniversalForwardOnly | Forward & Deferred Rendering                  |
            // | Universal2D          | for 2D light                                  |
            // | ShadowCaster         | depth from the perspective of lights          |
            // | DepthOnly            | depth from the perspective of a Camera        |
            // | Meta                 | executes this Pass only when baking lightmaps |

            // ref: https://docs.unity3d.com/Manual/SL-Stencil.html
            Stencil
            {
            }

            // ## Render 설정 (기본값 <>)
            // ref: https://docs.unity3d.com/Manual/SL-CullAndDepth.html
            // ref: https://docs.unity3d.com/Manual/SL-Blend.html
            Cull      (<Back> | Front | Off)
            ZTest     (Less | Greater | <LEqual> | GEqual | Equal | NotEqual | Always)
            ZWrite    (<On> | Off)
            Blend     SourceBlendMode DestBlendMode
            Blend     SourceBlendMode DestBlendMode, AlphaSourceBlendMode AlphaDestBlendMode
            ColorMask (RGB | A | 0 | any combination of R, G, B, A)
            Offset    OffsetFactor, OffsetUnits

            HLSLPROGRAM
            #pragma vertex   name // compile function name as the vertex shader.
            #pragma fragment name // compile function name as the fragment shader.
            #pragma geometry name // compile function name as DX10 geometry shader. Having this option automatically turns on #pragma target 4.0, described below.
            #pragma hull     name // compile function name as DX11 hull shader. Having this option automatically turns on #pragma target 5.0, described below.
            #pragma domain   name // compile function name as DX11 domain shader. Having this option automatically turns on #pragma target 5.0, described below.

            // Other compilation directives:
            // #pragma target            name                  - which shader target to compile to. See Shader Compilation Targets page for details.
            // #pragma only_renderers    space separated names - compile shader only for given renderers. By default shaders are compiled for all renderers. See Renderers below for details.
            // #pragma exclude_renderers space separated names - do not compile shader for given renderers. By default shaders are compiled for all renderers. See Renderers below for details.
            // #pragma enable_d3d11_debug_symbols              - generate debug information for shaders compiled for DirectX 11, this will allow you to debug shaders via Visual Studio 2012 (or higher) Graphics debugger.
            // #pragma multi_compile_instancing
            // #pragma multi_compile_fog
            // #pragma multi_compile                           - for working with multiple shader variants.
            //         multi_compile_local
            //         multi_compile_vertex
            //         multi_compile_vertex_local
            //         multi_compile_fragment
            //         multi_compile_fragment_local
            // #pragma shader_feature                          - for working with multiple shader variants. (unused variants of shader_feature shaders will not be included into game build)
            //         shader_feature_local
            //         shader_feature_vertex
            //         shader_feature_vertex_local
            //         shader_feature_fragment
            //         shader_feature_fragment_local
            
          
            // | Property | Variable                                    |
            // | -------- | ------------------------------------------- |
            // | Float    | float _Name;                                |
            // | Int      | float _Name;                                |
            // | Range    | float _Name;                                |
            // | Color    | float _Name;                                |
            // | Vector   | float _Name;                                |
            // | 2D       | TEXTURE2D(_Name);    SAMPLER(sampler_Name); |
            // | Rect     | TEXTURE2D(_Name);    SAMPLER(sampler_Name); |
            // | Cube     | TEXTURECUBE(_Name);  SAMPLER(sampler_Name); |
            // | 3D       | TEXTURE3D(_Name);    SAMPLER(sampler_Name); |

            // 간단한 Vertex/Fragment Shader 예제.
            TEXTURE2D(_MainTex);    SAMPLER(sampler_MainTex);

            struct APPtoVS
            {
                float4 positionOS   : POSITION;
                float3 normalOS     : NORMAL;
                float4 colorVertex  : COLOR;
                float2 uv           : TEXCOORD0;
            };

            struct VStoFS
            {
                float4 positionCS   : SV_POSITION;
                float2 uv           : TEXCOORD0;
            };

            VStoFS vert(APPtoVS IN)
            {
                VStoFS OUT;
                ZERO_INITIALIZE(VStoFS, OUT);

                OUT.positionCS = TransformObjectToHClip(IN.positionOS.xyz);
                OUT.uv = TRANSFORM_TEX(IN.uv, _MainTex);

                return OUT;
            }

            half4 frag(VStoFS IN) : SV_Target
            {
                half4 mainTex = SAMPLE_TEXTURE2D(_MainTex, sampler_MainTex, IN.uv);
                return mainTex;
            }

            // 멀티타겟용 (Deferred)
            void frag(    VStoFS IN,
                      out half4  outDiffuse        : SV_Target0,
                      out half4  outSpecSmoothness : SV_Target1,
                      out half4  outNormal         : SV_Target2,
                      out half4  outEmission       : SV_Target3)
            {
                // ...
            }
            ENDHLSL
        }
    }

    // ref : https://docs.unity3d.com/Manual/SL-Fallback.html
    Fallback "Diffuse"

    // ref: https://docs.unity3d.com/Manual/SL-CustomEditor.html
    // ref: https://docs.unity3d.com/Manual/SL-CustomMaterialEditors.html
    CustomEditor <custom-editor-class-name>
}

Etc

#define UNITY_BRANCH        [branch]
#define UNITY_FLATTEN       [flatten]
#define UNITY_UNROLL        [unroll]
#define UNITY_UNROLLX(_x)   [unroll(_x)]
#define UNITY_LOOP          [loop]

URP and Builtin

• Teo Dutra's From Built-in To URP

URP

• https://github.com/Unity-Technologies/UniversalRenderingExamples
• https://github.com/phi-lira/UniversalShaderExamples

include

Variant Keyword

• https://blogs.unity3d.com/2018/05/14/stripping-scriptable-shader-variants/

Macro

Shadow

Packages/com.unity.render-pipelines.universal/ShaderLibrary/Shadows.hlsl

Fog

com.unity.render-pipelines.universal/ShaderLibrary/ShaderVariablesFunctions.hlsl

Texture/Sampler Declaration Macros

Important to note that SCREENSPACE_TEXTURE has become TEXTURE2D_X. If you are working on some screen space effect for VR in Single Pass Instanced or Multi-view modes, you must declare the textures used with TEXTURE2D_X. This macro will handle for you the correct texture (array or not) declaration. You also have to sample the textures using SAMPLE_TEXTURE2D_X and use UnityStereoTransformScreenSpaceTex for the uv.

Helper

decodeInstructions is used as half4(LIGHTMAP_HDR_MULTIPLIER, LIGHTMAP_HDR_EXPONENT, 0.0h, 0.0h) by URP

Lighting

URP -_MainLightPosition.xyz 
URP - half3 VertexLighting(float3 positionWS, half3 normalWS) - com.unity.render-pipelines.universal/ShaderLibrary/Lighting.hlsl
If you want to loop over all additional lights using GetAdditionalLight(...), you can query the additional lights count by using GetAdditionalLightsCount().

Built-in	URP	 
_LightColor0	_MainLightColor	Include “Packages/com.unity.render-pipelines.universal/ShaderLibrary/Input.hlsl”
_WorldSpaceLightPos0	_MainLightPosition	Include “Packages/com.unity.render-pipelines.universal/ShaderLibrary/Input.hlsl”
_LightMatrix0	Gone ? Cookies are not supported yet	 
unity_4LightPosX0, unity_4LightPosY0, unity_4LightPosZ0	In URP, additional lights are stored in an array/buffer (depending on platform). Retrieve light information using Light GetAdditionalLight(uint i, float3 positionWS)	Include “Packages/com.unity.render-pipelines.universal/ShaderLibrary/Lighting.hlsl”
unity_4LightAtten0	In URP, additional lights are stored in an array/buffer (depending on platform). Retrieve light information using Light GetAdditionalLight(uint i, float3 positionWS)	Include “Packages/com.unity.render-pipelines.universal/ShaderLibrary/Lighting.hlsl”
unity_LightColor	In URP, additional lights are stored in an array/buffer (depending on platform). Retrieve light information using Light GetAdditionalLight(uint i, float3 positionWS)	Include “Packages/com.unity.render-pipelines.universal/ShaderLibrary/Lighting.hlsl”
unity_WorldToShadow	float4x4 _MainLightWorldToShadow[MAX_SHADOW_CASCADES + 1] or _AdditionalLightsWorldToShadow[MAX_VISIBLE_LIGHTS]	Include “Packages/com.unity.render-pipelines.universal/ShaderLibrary/Shadows.hlsl

| LIGHTING_COORDS| x| |

Vertex-lit Helper Functions

A bunch of utilities can be found in “Packages/com.unity.render-pipelines.core/ShaderLibrary/Common.hlsl”.

Screen-space Helper Functions

• ComputeScreenPos deprecated - https://github.com/Unity-Technologies/Graphics/pull/2529
• GetVertexPositionInputs().positionNDC

Camera Texture

fixed depth = SAMPLE_DEPTH_TEXTURE(texture, sampler, uv);

GrabPass
{
    "_GrabPass"
}
sampler2D _GrabPass
fixed4 fGrapPass = tex2Dproj(_GrabPass, i.screenPosition + 0.5f);

Depth

• Built-in URP
• LinearEyeDepth(sceneZ) LinearEyeDepth(sceneZ, _ZBufferParams) Include “Packages/com.unity.render-pipelines.core/ShaderLibrary/Common.hlsl”
• Linear01Depth(sceneZ) Linear01Depth(sceneZ, _ZBufferParams) Include “Packages/com.unity.render-pipelines.core/ShaderLibrary/Common.hlsl”

To use the camera depth texture, include “Packages/com.unity.render-pipelines.universal/ShaderLibrary/DeclareDepthTexture.hlsl” and the _CameraDepthTexture will be declared for you as well as helper the functions SampleSceneDepth(...) and LoadSceneDepth(...).

etc

• ShadeSH9(normal) SampleSH(normal) Include “Packages/com.unity.render-pipelines.universal/ShaderLibrary/Lighting.hlsl”
• unity_ColorSpaceLuminance Gone. Use Luminance() Include “Packages/com.unity.render-pipelines.core/ShaderLibrary/Color.hlsl”

여깃는건 쓰지말것 https://leegoonz.blog/ https://www.alanzucconi.com/tutorials/

• https://alexanderameye.github.io/outlineshader https://roystan.net/articles/toon-water.html - 물웅덩이 https://darkcatgame.tistory.com/84 https://www.cyanilux.com/recent/2/

https://github.com/hebory?before=Y3Vyc29yOnYyOpK5MjAyMC0xMS0wOVQyMjo0MToyOCswOTowMM4Oqhfn&tab=stars - https://blog.naver.com/cra2yboy/222236607952

• 툰쉐이더

  • https://musoucrow.github.io/2020/07/05/urp_outline/
  • https://kink3d.github.io/blog/2017/10/04/Physically-Based-Toon-Shading-In-Unity
    • https://github.com/Kink3d/kShading/blob/master/Shaders/ToonLit.shader

• 반사

  • https://github.com/Kink3d/kMirrors

• 모션블러

  • https://github.com/Kink3d/kMotion

• 월드 스페이스노말

  • https://github.com/Kink3d/kNormals

• Fast Subsurface Scattering in unity

  • https://www.alanzucconi.com/2017/08/30/fast-subsurface-scattering-2/

• 헤어 그림자

  • https://zhuanlan.zhihu.com/p/232450616

Bulit-in URP

GammaToLinearSpace Gamma22ToLinear com.unity.render-pipelines.core/ShaderLibrary/Color.hlsl

quick sheet

context.cmd.SetRenderTarget(data.TmpCopyTexHandle); // Output이라 보면 됨 Blitter.BlitTexture(nativeCmd, textureHandle, scaleBias, mipLevel: 0, bilinear: false); Blitter.BlitTexture(nativeCmd, textureHandle, scaleBias, material, pass); // => Blitter.BlitTexture함수를 쓰면 inputTextureHandle이 shader에서 _BlitTexture로 바인딩됨.

RenderTextureDescriptor rtdesc2 = new RenderTextureDescriptor(w, h, GraphicsFormat.R16G16_SFloat, 0) { autoGenerateMips = true, useMipMap = true }; RenderingUtils.ReAllocateHandleIfNeeded(ref _tmpCurrMipmapRT, rtdesc2, FilterMode.Bilinear, TextureWrapMode.Clamp, name: "_tmpCurrMipmapRT");

CommandBuffer nativeCmd = CommandBufferHelpers.GetNativeCommandBuffer(context.cmd);

migration RenderGraph API

• RenderingUtils.ReAllocateIfNeeded => RenderingUtils.ReAllocateHandleIfNeeded

• RTHandle 보다는 TextureHandle

textureHandle = renderGraph.ImportTexture(rtHandle);

• ScriptableRendererFeature를 상속받아쓰는 SetupRenderPasses 더 이상 안쓰임.

// before
public override void Configure(CommandBuffer cmd, RenderTextureDescriptor cameraTextureDescriptor)
{
    ConfigureInput(ScriptableRenderPassInput.Normal);
}

// after
using (IUnsafeRenderGraphBuilder builder = renderGraph.AddUnsafePass(passName, out PassData passData))
{
    ConfigureInput(ScriptableRenderPassInput.Normal);
}

활성 color 텍스쳐를 새로운 텍스쳐로 복사하는 예

public class CopyRenderFeature : ScriptableRendererFeature
{
    CopyRenderPass m_CopyRenderPass;

    public override void Create()
    {
        m_CopyRenderPass = new CopyRenderPass();
        m_CopyRenderPass.renderPassEvent = RenderPassEvent.AfterRenderingOpaques;
    }

    public override void AddRenderPasses(ScriptableRenderer renderer, ref RenderingData renderingData)
    {
        renderer.EnqueuePass(m_CopyRenderPass);
    }
}

class CopyRenderPass : ScriptableRenderPass
{
    public CopyRenderPass()
    {
        requiresIntermediateTexture = true;
    }

    public override void RecordRenderGraph(RenderGraph renderGraph, ContextContainer frameData)
    {
        const string passName = "Copy To or From Temp Texture";

        UniversalResourceData resourceData = frameData.Get<UniversalResourceData>();

        TextureHandle srcHandle = resourceData.activeColorTexture;
        TextureHandle dstHandle;
        {
            TextureDesc destinationDesc = renderGraph.GetTextureDesc(srcHandle);
            destinationDesc.name = $"CameraColor-{passName}";
            destinationDesc.clearBuffer = false;
            dstHandle = renderGraph.CreateTexture(destinationDesc);
        }

        if (RenderGraphUtils.CanAddCopyPassMSAA())
        {
            renderGraph.AddCopyPass(srcHandle, dstHandle, passName: passName);
            renderGraph.AddCopyPass(dstHandle, srcHandle, passName: passName);
        }
        else
        {
            Debug.Log("Can't add the copy pass due to MSAA");
        }
    }
}

requiresIntermediateTexture

• RenderFeature가 아니라 ScriptableRenderPass에서 셋팅하는게 good practice

UniversalResourceData

• 텍스쳐 핸들을 포함하고 있음.
  • active color / depth texture 등등.
• isActiveTargetBackBuffer
  • pass의 requiresIntermediateTexture가 true면 resourceData의 isActiveTargetBackBuffer는 무조껀 false

renderGraph

renderGraph.AddCopyPass(srcHandle, dstHandle, passName: passName);

RenderGraphUtils

RenderGraphUtils.CanAddCopyPassMSAA()

renderGraph.AddBlitPass를 이용 후 resourceData.cameraColor로 셋팅

public override void RecordRenderGraph(RenderGraph renderGraph, ContextContainer frameData)
{
    UniversalResourceData resourceData = frameData.Get<UniversalResourceData>();

    if (resourceData.isActiveTargetBackBuffer)
    {
        Debug.LogError($"Skipping render pass. BlitAndSwapColorRendererFeature requires an intermediate ColorTexture, we can't use the BackBuffer as a texture input.");
        return;
    }

    TextureHandle srcHandle = resourceData.activeColorTexture;
    TextureHandle dstHandle;
    {
        TextureDesc destinationDesc = renderGraph.GetTextureDesc(srcHandle);
        destinationDesc.name = $"CameraColor-{m_PassName}";
        destinationDesc.clearBuffer = false;
        dstHandle = renderGraph.CreateTexture(destinationDesc);
    }


    RenderGraphUtils.BlitMaterialParameters blitParam = new RenderGraphUtils.BlitMaterialParameters(srcHandle, dstHandle, m_BlitMaterial, shaderPass: 0);
    renderGraph.AddBlitPass(blitParam, passName: m_PassName);


    resourceData.cameraColor = dstHandle;
}

renderGraph

RenderGraphUtils.BlitMaterialParameters blitParam = new RenderGraphUtils.BlitMaterialParameters(srcHandle, dstHandle, m_BlitMaterial, shaderPass: 0); renderGraph.AddBlitPass(blitParam, passName: m_PassName);

• BlitAndSwapColorRendererFeature.cs 예제와의 차이는
  • requiresIntermediateTexture 를 키지 않았고
  • BlitData를 이용.
  • pass를 3단계
    1. BlitStartRenderPass : 버퍼만들고 activeColorTexture를 m_Texture에 저장
    2. BlitRenderPass : 버퍼 사용해서 m_TextureHandleFront / m_TextureHandleBack 을 blit.
    3. BlitEndRenderPass : 버퍼 사용해서 m_Texture를 activeColorTexture에 저장

class BlitStartRenderPass : ScriptableRenderPass
{
    public override void RecordRenderGraph(RenderGraph renderGraph, ContextContainer frameData)
    {
        BlitData blitTextureData = frameData.Create<BlitData>();
        blitTextureData.RecordBlitColor(renderGraph, frameData);
    }
}

class BlitEndRenderPass : ScriptableRenderPass
{
    public override void RecordRenderGraph(RenderGraph renderGraph, ContextContainer frameData)
    {
        BlitData blitTextureData = frameData.Get<BlitData>();
        blitTextureData.RecordBlitBackToColor(renderGraph, frameData);
    }
}

class BlitRenderPass : ScriptableRenderPass
{
    public override void RecordRenderGraph(RenderGraph renderGraph, ContextContainer frameData)
    {
        BlitData blitTextureData = frameData.Get<BlitData>();

        foreach (Material material in m_Materials)
        {
            if (material == null)
            {
                continue;
            }

            blitTextureData.RecordFullScreenPass(renderGraph, $"Blit {material.name} Pass", material);
        }
    }
}

public class BlitData : ContextItem, IDisposable
{
    // ContextItem.Reset()
    // 텍스쳐 핸들이 1프레임에만 살아있기에 invalid texture handles에 의한 릭을 피하기 위해 매 프레임 후에 texture handle을 리셋시켜야함.
    public override void Reset()
    {
        m_TextureHandleFront = TextureHandle.nullHandle;
        m_TextureHandleBack = TextureHandle.nullHandle;
        m_Texture = TextureHandle.nullHandle;
        m_IsFront = true;
    }

    // IDisposable.Dispose()
    public void Dispose()
    {
        if (m_TextureFront != null)
        {
            m_TextureFront.Release();
        }
        
        if (m_TextureFront != null)
        {
            m_TextureFront.Release();
        }
    }


    // BlitStartRenderPass => RecordBlitColor
    // BlitRenderPass      => RecordFullScreenPass
    // BlitEndRenderPass   => RecordBlitBackToColor
}

// BlitStartRenderPass => RecordBlitColor
public void RecordBlitColor(RenderGraph renderGraph, ContextContainer frameData)
{
    if (!m_Texture.IsValid())
    {
        UniversalCameraData cameraData = frameData.Get<UniversalCameraData>();

        RenderTextureDescriptor descriptor = cameraData.cameraTargetDescriptor;
        descriptor.msaaSamples = 1;
        descriptor.depthStencilFormat = UnityEngine.Experimental.Rendering.GraphicsFormat.None;

        string texName = "_BlitTextureData";

        RenderingUtils.ReAllocateHandleIfNeeded(ref m_TextureFront, descriptor, FilterMode.Bilinear, TextureWrapMode.Clamp, name: $"{texName}_Front");
        RenderingUtils.ReAllocateHandleIfNeeded(ref m_TextureBack, descriptor, FilterMode.Bilinear, TextureWrapMode.Clamp, name: $"{texName}_Back");

        m_TextureHandleFront = renderGraph.ImportTexture(m_TextureFront);
        m_TextureHandleBack = renderGraph.ImportTexture(m_TextureBack);
        m_Texture = m_TextureHandleFront;
    }

    using (IRasterRenderGraphBuilder builder = renderGraph.AddRasterRenderPass("BlitColorPass", out PassData passData))
    {
        UniversalResourceData resourceData = frameData.Get<UniversalResourceData>();

        passData.source = resourceData.activeColorTexture;
        passData.destination = m_Texture;
        passData.material = null;

        builder.UseTexture(passData.source);
        builder.SetRenderAttachment(passData.destination, index: 0);
        builder.SetRenderFunc((PassData passData, RasterGraphContext rgContext) => ExecutePass(passData, rgContext));
    }
}

// BlitRenderPass => RecordFullScreenPass
public void RecordFullScreenPass(RenderGraph renderGraph, string passName, Material material)
{
    if (!m_Texture.IsValid())
    {
        Debug.LogWarning("Invalid input texture handle, will skip fullscreen pass.");
        return;
    }

    if (material == null)
    {
        return;
    }

    using (IRasterRenderGraphBuilder builder = renderGraph.AddRasterRenderPass(passName, out PassData passData))
    {
        m_IsFront = !m_IsFront;

        TextureHandle src = m_Texture;
        TextureHandle dst;
        if (m_IsFront)
        {
            dst = m_TextureHandleFront;
        }
        else
        {
            dst = m_TextureHandleBack;
        }

        m_Texture = dst;

        passData.source = src;
        passData.destination = dst;
        passData.material = material;

        builder.UseTexture(passData.source);
        builder.SetRenderAttachment(passData.destination, index: 0);
        builder.SetRenderFunc((PassData passData, RasterGraphContext rgContext) => ExecutePass(passData, rgContext));
    }
}

// BlitEndRenderPass => RecordBlitBackToColor

public void RecordBlitBackToColor(RenderGraph renderGraph, ContextContainer frameData)
{
    if (!m_Texture.IsValid())
    {
        return;
    }

    using (IRasterRenderGraphBuilder builder = renderGraph.AddRasterRenderPass($"BlitBackToColorPass", out PassData passData))
    {
        UniversalResourceData resourceData = frameData.Get<UniversalResourceData>();

        passData.source = m_Texture;
        passData.destination = resourceData.activeColorTexture;
        passData.material = null;

        builder.UseTexture(passData.source);
        builder.SetRenderAttachment(passData.destination, index: 0);
        builder.SetRenderFunc((PassData passData, RasterGraphContext rgContext) => ExecutePass(passData, rgContext));
    }
}

static void ExecutePass(PassData data, RasterGraphContext rgContext)
{
    if (data.material == null)
    {
        Blitter.BlitTexture(rgContext.cmd, data.source, scaleBias, mipLevel: 0, bilinear: false);
        return;
    }

    Blitter.BlitTexture(rgContext.cmd, data.source, scaleBias, data.material, pass: 0);
}

ContextContainer frameData

BlitData blitTextureData = frameData.Create<BlitData>();



BlitData blitTextureData = frameData.Get<BlitData>();

IRasterRenderGraphBuilder builder

builder.UseTexture(passData.source);                     // set input
builder.SetRenderAttachment(passData.destination, 0);    // set output
builder.SetRenderFunc((PassData passData, RasterGraphContext rgContext) => ExecutePass(passData, rgContext));



using (IRasterRenderGraphBuilder builder = renderGraph.AddRasterRenderPass(passName, out PassData passData))
using (IComputeRenderGraphBuilder builder = renderGraph.AddComputePass(passName, out PassData passData))
using (IUnsafeRenderGraphBuilder builder = renderGraph.AddUnsafePass(passName, out PassData passData))

Etc

// https://github.com/Unity-Technologies/Graphics/blob/master/Packages/com.unity.render-pipelines.core/Runtime/Textures/MSAASamples.cs
public enum MSAASamples
{
    /// <summary>No MSAA.</summary>
    None = 1,
    /// <summary>MSAA 2X.</summary>
    MSAA2x = 2,
    /// <summary>MSAA 4X.</summary>
    MSAA4x = 4,
    /// <summary>MSAA 8X.</summary>
    MSAA8x = 8
}

BlitRendererFeature

ContextContainer frameData 동작방식

• 2 ScriptableRenderPass
  • UpdateRefPass
  • CopyBackRefPass
• ContextItem를 상속받는 TexRefData를 이용
  • UpdateRefPass에서 resourceData.activeColorTexture => texRef.texture
    • UpdateRefPass에서 foreach (Material mat in m_DisplayMaterials) 하므로 매 프레임 첫 루프 단계에서는 frameData.Contains()가 false, 그 후로는 true.
  • CopyBackRefPass에서 texRef.texture => resourceData.activeColorTexture

public class TexRefData : ContextItem
{
    public TextureHandle texture = TextureHandle.nullHandle;

    public override void Reset()
    {
        texture = TextureHandle.nullHandle;
    }
}

//     class UpdateRefPass : ScriptableRenderPass

public override void RecordRenderGraph(RenderGraph renderGraph, ContextContainer frameData)
{
    foreach (Material mat in m_DisplayMaterials)
    {
        if (mat == null)
        {
            Debug.LogWarning($"Skipping render pass for unassigned material.");
            continue;
        }

        using (IRasterRenderGraphBuilder builder = renderGraph.AddRasterRenderPass($"UpdateRefPass_{mat.name}", out PassData passData))
        {

            TexRefData texRef;
            if (frameData.Contains<TexRefData>())
            {
                texRef = frameData.Get<TexRefData>();
            }
            else
            {
                UniversalResourceData resourceData = frameData.Get<UniversalResourceData>();
                texRef = frameData.Create<TexRefData>();
                texRef.texture = resourceData.activeColorTexture;
            }

            TextureDesc descriptor = texRef.texture.GetDescriptor(renderGraph);
            descriptor.msaaSamples = MSAASamples.None;
            descriptor.name = $"BlitMaterialRefTex_{mat.name}";
            descriptor.clearBuffer = false;

            TextureHandle srcHandle = texRef.texture;
            TextureHandle dstHandle = renderGraph.CreateTexture(descriptor);
            texRef.texture = dstHandle;

            passData.source = srcHandle;
            passData.destination = dstHandle;
            passData.material = mat;


            builder.UseTexture(input: passData.source);
            builder.SetRenderAttachment(tex: passData.destination, index: 0);
            builder.SetRenderFunc((PassData data, RasterGraphContext rgContext) => ExecutePass(data, rgContext));
        }
    }
}

static void ExecutePass(PassData data, RasterGraphContext rgContext)
{
    Blitter.BlitTexture(rgContext.cmd, data.source, scaleBias, data.material, pass: 0);
}

// class CopyBackRefPass : ScriptableRenderPass

public override void RecordRenderGraph(RenderGraph renderGraph, ContextContainer frameData)
{
    if (!frameData.Contains<TexRefData>())
    {
        return;
    }

    TexRefData texRef = frameData.Get<TexRefData>();
    TextureHandle srcHandle = texRef.texture;

    UniversalResourceData resourceData = frameData.Get<UniversalResourceData>();
    TextureHandle dstHandle = resourceData.activeColorTexture;

    renderGraph.AddBlitPass(srcHandle, dstHandle, scale: Vector2.one, offset: Vector2.zero, passName: "Blit Back Pass");
}

ContextContainer frameData

bool isExist_TexRefData = frameData.Contains<TexRefData>();
texRef = frameData.GetOrCreate<TexRefData>();
texRef = frameData.Create<TexRefData>();
texRef = frameData.Get<TexRefData>();

scaleBias

Vector4 scaleBias(scaleX, scaleY, offsetX, offsetY); Blitter.BlitTexture(rgContext.cmd, data.source, scaleBias, data.material, pass: 0);

[Flags]
public enum ScriptableRenderPassInput
{
    None   = 0,
    Depth  = 1 << 0,
    Normal = 1 << 1,
    Color  = 1 << 2,
    Motion = 1 << 3,
}

EnqueuePass로 pass를 넣기전 pass의 ConfigureInput(ScriptableRenderPassInput passInput)를 호출

// TextureHandle
//     resourceData.cameraOpaqueTexture
//     resourceData.cameraDepthTexture
//     resourceData.cameraNormalsTexture
//     resourceData.motionVectorColor
//     TextureHandle.nullHandle

public override void RecordRenderGraph(RenderGraph renderGraph, ContextContainer frameData)
{
    UniversalResourceData resourceData = frameData.Get<UniversalResourceData>();

    TextureHandle source = GetTextureHandleFromType(resourceData, m_TextureType);
    if (!source.IsValid())
    {
        Debug.Log("Input texture is not created. Likely the pass event is before the creation of the resource. Skipping OutputTexturePass.");
        return;
    }

    RenderGraphUtils.BlitMaterialParameters para = new RenderGraphUtils.BlitMaterialParameters(source, resourceData.activeColorTexture, m_Material, shaderPass: 0);
    param.sourceTexturePropertyID = Shader.PropertyToID(m_TextureName);
    renderGraph.AddBlitPass(param, passName: "Blit Selected Resource");
}

resourceData.activeColorTexture 를 1/4배까지 다운샘플링하고 다시 업샘플링하여 원복시킴.

private class PassData
{
    internal TextureHandle source;             // resourceData.activeColorTexture
    internal TextureHandle destination;        // 1
    internal TextureHandle destinationHalf;    // 1/2
    internal TextureHandle destinationQuarter; // 1/4
}

public override void RecordRenderGraph(RenderGraph renderGraph, ContextContainer frameData)
{
    string passName = "Unsafe Pass";

    using (IUnsafeRenderGraphBuilder builder = renderGraph.AddUnsafePass(passName, out PassData passData))
    {
        UniversalResourceData resourceData = frameData.Get<UniversalResourceData>();

        TextureDesc descriptor;
        {
            descriptor = resourceData.activeColorTexture.GetDescriptor(renderGraph);
            descriptor.msaaSamples = MSAASamples.None;
            descriptor.clearBuffer = false;
        }

        TextureHandle destination;
        {
            descriptor.name = "UnsafeTexture";
            destination = renderGraph.CreateTexture(descriptor);
        }

        TextureHandle destinationHalf;
        {
            descriptor.width /= 2;
            descriptor.height /= 2;
            descriptor.name = "UnsafeTexture2";
            destinationHalf = renderGraph.CreateTexture(descriptor);
        }

        TextureHandle destinationQuarter;
        {
            descriptor.width /= 2;
            descriptor.height /= 2;
            descriptor.name = "UnsafeTexture3";
            destinationQuarter = renderGraph.CreateTexture(descriptor);
        }

        passData.source = resourceData.activeColorTexture;
        passData.destination = destination;
        passData.destinationHalf = destinationHalf;
        passData.destinationQuarter = destinationQuarter;

        builder.UseTexture(passData.source);
        builder.UseTexture(passData.destination, AccessFlags.WriteAll);
        builder.UseTexture(passData.destinationHalf, AccessFlags.WriteAll);
        builder.UseTexture(passData.destinationQuarter, AccessFlags.WriteAll);

        builder.AllowPassCulling(value: false);

        builder.SetRenderFunc((PassData data, UnsafeGraphContext context) => ExecutePass(data, context));
    }
}

static void ExecutePass(PassData data, UnsafeGraphContext context)
{
    CommandBuffer unsafeCmd = CommandBufferHelpers.GetNativeCommandBuffer(context.cmd);

    // 1 => 1
    context.cmd.SetRenderTarget(data.destination);
    Blitter.BlitTexture(unsafeCmd, data.source, new Vector4(1, 1, 0, 0), 0, false);

    // 1 => 1/2
    context.cmd.SetRenderTarget(data.destinationHalf);
    Blitter.BlitTexture(unsafeCmd, data.destination, new Vector4(1, 1, 0, 0), 0, false);

    // 1/2 => 1/4
    context.cmd.SetRenderTarget(data.destinationQuarter);
    Blitter.BlitTexture(unsafeCmd, data.destinationHalf, new Vector4(1, 1, 0, 0), 0, false);

    // 1/4 => 1/2
    context.cmd.SetRenderTarget(data.destinationHalf);
    Blitter.BlitTexture(unsafeCmd, data.destinationQuarter, new Vector4(1, 1, 0, 0), 0, false);

    // 1/2 => 1
    context.cmd.SetRenderTarget(data.destination);
    Blitter.BlitTexture(unsafeCmd, data.destinationHalf, new Vector4(1, 1, 0, 0), 0, false);
}

• unsafe에서는 UseTexture를 써야하고 다음은 안됨
  • builder.UseTextureFragment
  • builder.UseTextureFragmentDepth

• 프레임버퍼 페치는비디오 메모리 대신 GPU의 온칩 메모리에서 프레임버퍼에 액세스할 수 있는 렌더 패스를 허용하는 프로세스
  • 모바일 기기에서 타일 기반 지연 렌더링(TBDR)을 사용하는 모바일 기기에서 렌더링 속도가 향상

https://docs.unity3d.com/6000.0/Documentation/Manual/urp/render-graph-framebuffer-fetch.html

FBF( Frame Buffer Fetch)

SetInputAttachment

• support API
  • Vulkan
  • Metal

Input
    FRAMEBUFFER_INPUT_X_HALF
    FRAMEBUFFER_INPUT_X_FLOAT
    FRAMEBUFFER_INPUT_X_INT
    FRAMEBUFFER_INPUT_X_UINT

Load
    LOAD_FRAMEBUFFER_INPUT(idx)
    LOAD_FRAMEBUFFER_INPUT_MS(idx,sampleIdx)  // 멀티샘플링(MSAA)이 활성화된 텍스처를 처리하기 위해 사용됨.

builder.SetInputAttachment(sourceTextureHandle, 0, AccessFlags.Read);

public override void RecordRenderGraph(RenderGraph renderGraph, ContextContainer frameData)
{
    UniversalResourceData resourceData = frameData.Get<UniversalResourceData>();

    TextureHandle source = resourceData.activeColorTexture;

    TextureDesc destinationDesc = renderGraph.GetTextureDesc(source);
    destinationDesc.name = "FBFetchDestTexture";
    destinationDesc.clearBuffer = false;

    if (destinationDesc.msaaSamples == MSAASamples.None || RenderGraphUtils.CanAddCopyPassMSAA())
    {
        bool isUseMSAA = destinationDesc.msaaSamples != MSAASamples.None;

        TextureHandle fbFetchDestination = renderGraph.CreateTexture(destinationDesc);

        FBFetchPass(renderGraph, frameData, srcHandle: source, dstHandle: fbFetchDestination, isUseMSAA);

        renderGraph.AddCopyPass(fbFetchDestination, source, passName: "Copy Back FF Destination (also using FBF)");
    }
    else
    {
        Debug.Log("Can't add the FBF pass and the copy pass due to MSAA");
    }
}

private void FBFetchPass(RenderGraph renderGraph, ContextContainer frameData, TextureHandle srcHandle, TextureHandle dstHandle, bool isUseMSAA)
{
    string passName = "FrameBufferFetchPass";

    using (IRasterRenderGraphBuilder builder = renderGraph.AddRasterRenderPass(passName, out PassData passData))
    {
        passData.material = m_FBFetchMaterial;
        passData.isUseMSAA = isUseMSAA;

        builder.SetInputAttachment(tex: srcHandle, index: 0, AccessFlags.Read);
        builder.SetRenderAttachment(tex: dstHandle, index: 0); // Output
        builder.AllowPassCulling(false);
        builder.SetRenderFunc((PassData data, RasterGraphContext context) => ExecuteFBFetchPass(data, context));
    }
}

static void ExecuteFBFetchPass(PassData data, RasterGraphContext context)
{
    context.cmd.DrawProcedural(Matrix4x4.identity, data.material, data.useMSAA ? 1 : 0, MeshTopology.Triangles, 3, 1, null);
}

Shader "FrameBufferFetch"
{
   SubShader
   {
       Tags { "RenderType"="Opaque" "RenderPipeline" = "UniversalPipeline"}
       ZWrite Off Cull Off
       Pass
       {
           Name "FrameBufferFetch"

           HLSLPROGRAM
           #include "Packages/com.unity.render-pipelines.universal/ShaderLibrary/Core.hlsl"
           #include "Packages/com.unity.render-pipelines.core/Runtime/Utilities/Blit.hlsl"

           #pragma vertex Vert
           #pragma fragment Frag

           FRAMEBUFFER_INPUT_HALF(0);

           // Out frag function takes as input a struct that contains the screen space coordinate we are going to use to sample our texture. It also writes to SV_Target0, this has to match the index set in the UseTextureFragment(sourceTexture, 0, …) we defined in our render pass script.   
           float4 Frag(Varyings input) : SV_Target0
           {
               UNITY_SETUP_STEREO_EYE_INDEX_POST_VERTEX(input); // for XR platform

               float2 uv = input.texcoord.xy;
               half4 color = LOAD_FRAMEBUFFER_INPUT(0, input.positionCS.xy);

               return half4(0,0,1,1) * color;
           }

           ENDHLSL
       }

       Tags { "RenderType"="Opaque" "RenderPipeline" = "UniversalPipeline"}
       ZWrite Off Cull Off
       Pass
       {
           Name "FrameBufferFetchMS"

           HLSLPROGRAM
           #include "Packages/com.unity.render-pipelines.universal/ShaderLibrary/Core.hlsl"
           #include "Packages/com.unity.render-pipelines.core/Runtime/Utilities/Blit.hlsl"

           #pragma vertex Vert
           #pragma fragment Frag
           #pragma target 4.5
           #pragma require msaatex

           FRAMEBUFFER_INPUT_HALF_MS(0);

           // Out frag function takes as input a struct that contains the screen space coordinate we are going to use to sample our texture. It also writes to SV_Target0, this has to match the index set in the UseTextureFragment(sourceTexture, 0, …) we defined in our render pass script.   
           float4 Frag(Varyings input, uint sampleID : SV_SampleIndex) : SV_Target0
           {
               UNITY_SETUP_STEREO_EYE_INDEX_POST_VERTEX(input); // for XR platform

               float2 uv = input.texcoord.xy;
               half4 color = LOAD_FRAMEBUFFER_INPUT_MS(0, sampleID, input.positionCS.xy);
               
               // Modify the sampled color
               return half4(0,0,1,1) * color;
           }

           ENDHLSL
       }
   }
}

지정된 LayerMask에 있는 오브젝트만 그리도록 RenderList 사용.

renderGraph.CreateRendererList()로 RendererListHandle을 만듬.

람다대신 그냥 넣는게 나을듯?

builder.SetRenderFunc((PassData data, RasterGraphContext context) => ExecutePass(data, context)); builder.SetRenderFunc(ExecutePass);

public override void RecordRenderGraph(RenderGraph renderGraph, ContextContainer frameData)
{
    string passName = "RenderList Render Pass";

    using (IRasterRenderGraphBuilder builder = renderGraph.AddRasterRenderPass(passName, out PassData passData))
    {
        UniversalResourceData resourceData = frameData.Get<UniversalResourceData>();

        InitRendererLists(renderGraph, frameData, ref passData);

        if (!passData.rendererListHandle.IsValid())
        {
            return;
        }

        builder.UseRendererList(passData.rendererListHandle);
        builder.SetRenderAttachment(resourceData.activeColorTexture, index: 0);
        builder.SetRenderAttachmentDepth(resourceData.activeDepthTexture, AccessFlags.Write);
        builder.SetRenderFunc((PassData data, RasterGraphContext context) => ExecutePass(data, context));
    }
}


static void ExecutePass(PassData passData, RasterGraphContext context)
{
    context.cmd.ClearRenderTarget(clearFlags: RTClearFlags.Color, backgroundColor: Color.green, depth: 1, stencil: 0);
    context.cmd.DrawRendererList(passData.rendererListHandle);
}

private class PassData
{
    public RendererListHandle rendererListHandle;
}

private void InitRendererLists(RenderGraph renderGraph, ContextContainer frameData, ref PassData passData)
{
    UniversalRenderingData universalRenderingData = frameData.Get<UniversalRenderingData>();

    RendererListParams param;
    {
        DrawingSettings drawSettings;
        {
            m_ShaderTagIdList.Clear();
            ShaderTagId[] forwardOnlyShaderTagIds = new ShaderTagId[]
                {
                new ShaderTagId("UniversalForwardOnly"),
                new ShaderTagId("UniversalForward"),
                new ShaderTagId("SRPDefaultUnlit"), // Legacy shaders (do not have a gbuffer pass) are considered forward-only for backward compatibility
                new ShaderTagId("LightweightForward") // Legacy shaders (do not have a gbuffer pass) are considered forward-only for backward compatibility
                };
            m_ShaderTagIdList.AddRange(forwardOnlyShaderTagIds);
            UniversalCameraData cameraData = frameData.Get<UniversalCameraData>();
            UniversalLightData lightData = frameData.Get<UniversalLightData>();
            SortingCriteria sortFlags = cameraData.defaultOpaqueSortFlags;

            drawSettings = RenderingUtils.CreateDrawingSettings(m_ShaderTagIdList, universalRenderingData, cameraData, lightData, sortFlags);
        }
        RenderQueueRange renderQueueRange = RenderQueueRange.opaque;
        FilteringSettings filterSettings = new FilteringSettings(renderQueueRange, m_LayerMask);

        param = new RendererListParams(universalRenderingData.cullResults, drawSettings, filterSettings);
    }
    passData.rendererListHandle = renderGraph.CreateRendererList(param);
}

• RendererAsset에서 Rendering > Deferred로 변경
• RenderPassEvent.AfterRenderingDeferredLights
• activeColorTexture == _GBuffer3 (Lighting)

public override void RecordRenderGraph(RenderGraph renderGraph, ContextContainer frameData)
{
    if (m_Material == null)
    {
        return;
    }

    UniversalRenderingData universalRenderingData = frameData.Get<UniversalRenderingData>();
    if (universalRenderingData.renderingMode != RenderingMode.Deferred)
    {
        return;
    }

    UniversalResourceData resourceData = frameData.Get<UniversalResourceData>();
    TextureHandle[] gBuffer = resourceData.gBuffer;

    using (IRasterRenderGraphBuilder builder = renderGraph.AddRasterRenderPass(m_PassName, out PassData passData))
    {
        passData.material = m_Material;
        passData.gBuffer = gBuffer;

        builder.SetRenderAttachment(resourceData.activeColorTexture, index: 0, flags: AccessFlags.Write); // resourceData.activeColorTexture == _GBuffer3 (Lighting)
        for (int i = 0; i < resourceData.gBuffer.Length; i++)
        {
            if (i == GbufferLightingIndex)
            {
                continue;
            }

            builder.UseTexture(resourceData.gBuffer[i]);
        }

        builder.SetRenderFunc((PassData data, RasterGraphContext context) => ExecutePass(data, context));
    }
}

static void ExecutePass(PassData data, RasterGraphContext context)
{
    for (int i = 0; i < data.gBuffer.Length; i++)
    {
        data.material.SetTexture(s_GBufferShaderPropertyIDs[i], data.gBuffer[i]);
    }
    context.cmd.DrawProcedural(Matrix4x4.identity, data.material, shaderPass: 0, MeshTopology.Triangles, vertexCount: 3, instanceCount: 1);
}

UniversalRenderingData

UniversalRenderingData universalRenderingData = frameData.Get<UniversalRenderingData>();
universalRenderingData.renderingMode

public enum RenderingMode
{
    Forward = 0,
    /// <summary>Render all objects and lighting in one pass using a clustered data structure to access lighting data.</summary>
    [InspectorName("Forward+")]
    ForwardPlus = 2,
    Deferred = 1
};

shader

#include "Packages/com.unity.render-pipelines.core/ShaderLibrary/GlobalSamplers.hlsl"

TEXTURE2D_X(_GBuffer2);

struct Attributes
{
    uint vertexID : SV_VertexID;
    UNITY_VERTEX_INPUT_INSTANCE_ID
};

struct Varyings
{
    float4 positionCS : SV_POSITION;
    float2 texcoord   : TEXCOORD0;

    UNITY_VERTEX_INPUT_INSTANCE_ID
    UNITY_VERTEX_OUTPUT_STEREO
};

Varyings GBufferVisPassVertex(Attributes input)
{
    Varyings output;
    UNITY_SETUP_INSTANCE_ID(input);
    UNITY_TRANSFER_INSTANCE_ID(input, output);
    UNITY_INITIALIZE_VERTEX_OUTPUT_STEREO(output);

    float4 pos = GetFullScreenTriangleVertexPosition(input.vertexID);
    float2 uv  = GetFullScreenTriangleTexCoord(input.vertexID);

    output.positionCS = pos;
    output.texcoord   = uv;

    return output;
}

void GBufferVisPassFragment(Varyings input, out half4 outColor : SV_Target0)
{
    UNITY_SETUP_INSTANCE_ID(input);
    UNITY_SETUP_STEREO_EYE_INDEX_POST_VERTEX(input);

    float2 uv = input.texcoord;
    #ifndef UNITY_UV_STARTS_AT_TOP
        uv.y = 1.0 - uv.y;
    #endif

    outColor = SAMPLE_TEXTURE2D_X_LOD(_GBuffer2, sampler_PointClamp, uv, 0);
}

_GBuffer

//  Make sure this is consistent with DeferredLights.cs
private static readonly int s_GbufferLightingIndex = 3; // _GBuffer3 is the activeColorTexture

private static readonly int[] s_GBufferShaderPropertyIDs = new int[]
{
    Shader.PropertyToID("_GBuffer0"), // Albedo
    Shader.PropertyToID("_GBuffer1"), // Specular Metallic
    Shader.PropertyToID("_GBuffer2"), // Normals and Smoothness      ( _CameraNormalsTexture )
    Shader.PropertyToID("_GBuffer3"), // Lighting
    Shader.PropertyToID("_GBuffer4"), // (optional) Depth            ( _CameraDepthTexture )
    Shader.PropertyToID("_GBuffer5"), // (optional) Rendering Layers ( _CameraRenderingLayersTexture )
    Shader.PropertyToID("_GBuffer6"), // (optional) ShadowMask (Rendering Layers가 없으면 이게 _GBuffer5가 된다)
};

셰이더에서 텍스쳐 사용시 builder.UseTexture를 사용해야 하는데, builder.SetGlobalTextureAfterPass로 미리 등록시키면 builder.UseAllGlobalTextures(enable: true)로 등록된 텍스쳐들을 사용할 수 있다.

public override void RecordRenderGraph(RenderGraph renderGraph, ContextContainer frameData)
{
    UniversalRenderingData universalRenderingData = frameData.Get<UniversalRenderingData>();
    if (universalRenderingData.renderingMode != RenderingMode.Deferred)
    {
        return;
    }

    UniversalResourceData resourceData = frameData.Get<UniversalResourceData>();
    TextureHandle[] gBuffer = resourceData.gBuffer;

    using (IRasterRenderGraphBuilder builder = renderGraph.AddRasterRenderPass(m_PassName, out PassData passData))
    {
        builder.AllowPassCulling(value: false);
        SetGlobalGBufferTextures(builder, gBuffer);
        builder.SetRenderFunc((PassData data, RasterGraphContext context) => { /* nothing to be rendered */ });
    }
}

private void SetGlobalGBufferTextures(IRasterRenderGraphBuilder builder, TextureHandle[] gBuffer)
{
    for (int i = 0; i < gBuffer.Length; i++)
    {
        if (i != GbufferLightingIndex && gBuffer[i].IsValid())
        {
            builder.SetGlobalTextureAfterPass(gBuffer[i], s_GBufferShaderPropertyIDs[i]);
        }
    }

    builder.UseAllGlobalTextures(enable: true);

    if (gBuffer[GBufferNormalSmoothnessIndex].IsValid())
    {
        builder.SetGlobalTextureAfterPass(gBuffer[GBufferNormalSmoothnessIndex], Shader.PropertyToID("_CameraNormalsTexture"));
    }

    if (gBuffer[GbufferDepthIndex].IsValid())
    {
        builder.SetGlobalTextureAfterPass(gBuffer[GbufferDepthIndex], Shader.PropertyToID("_CameraDepthTexture"));
    }

    if (GBufferRenderingLayersIndex < gBuffer.Length && gBuffer[GBufferRenderingLayersIndex].IsValid())
    {
        builder.SetGlobalTextureAfterPass(gBuffer[GBufferRenderingLayersIndex], Shader.PropertyToID("_CameraRenderingLayersTexture"));
    }
}

IRasterRenderGraphBuilder builder

builder.AllowPassCulling(value: false);

builder.SetGlobalTextureAfterPass(in TextureHandle input, int propertyId);

[SerializeField] ComputeShader computeShader;

#pragma kernel CSMain

StructuredBuffer<int>   inputData;  // readonly
RWStructuredBuffer<int> outputData; // read/write

[numthreads(20,1,1)]
void CSMain (uint3 id : SV_DispatchThreadID)
{
    outputData[id.x] = 2 * inputData[id.x];
}

GraphicsBuffer _inputBuffer;
GraphicsBuffer _outputBuffer;

public ComputePass()
{
    List<int> list = Enumerable.Range(start: 0, count: 20).ToList();

    _inputBuffer = new GraphicsBuffer(GraphicsBuffer.Target.Structured, 20, sizeof(int));
    _inputBuffer.SetData(list);
    _outputBuffer = new GraphicsBuffer(GraphicsBuffer.Target.Structured, 20, sizeof(int));
    _outputBuffer.SetData(list);
}



class PassData
{
    public ComputeShader cs;
    public BufferHandle input;
    public BufferHandle output;
}



ComputeShader _cs;
int[] _outputData = new int[20];

public override void RecordRenderGraph(RenderGraph renderGraph, ContextContainer frameData)
{
    _outputBuffer.GetData(_outputData);
    Debug.Log($"Output from compute shader: {string.Join(", ", _outputData)}");

    BufferHandle inputHandle = renderGraph.ImportBuffer(_inputBuffer);
    BufferHandle outputHandle = renderGraph.ImportBuffer(_outputBuffer);

    using (IComputeRenderGraphBuilder builder = renderGraph.AddComputePass("ComputePass", out PassData passData))
    {
        passData.cs = _cs;
        passData.input = inputHandle;
        passData.output = outputHandle;

        builder.UseBuffer(passData.input);
        builder.UseBuffer(passData.output, flags: AccessFlags.Write);
        builder.SetRenderFunc((PassData data, ComputeGraphContext cgContext) => ExecutePass(data, cgContext));
    }
}

static void ExecutePass(PassData data, ComputeGraphContext cgContext)
{
    cgContext.cmd.SetComputeBufferParam(data.cs, data.cs.FindKernel("CSMain"), "inputData", data.input);
    cgContext.cmd.SetComputeBufferParam(data.cs, data.cs.FindKernel("CSMain"), "outputData", data.output);
    cgContext.cmd.DispatchCompute(data.cs, data.cs.FindKernel("CSMain"), threadGroupsX: 1, threadGroupsY: 1, threadGroupsZ: 1);
}

MRT: Multiple Render Targets

public override void RecordRenderGraph(RenderGraph renderGraph, ContextContainer frameData)
{
    TextureHandle[] handles = new TextureHandle[3];
    for (int i = 0; i < 3; i++)
    {
        handles[i] = renderGraph.ImportTexture(m_RTs[i], m_RTInfos[i]);
    }

    using (IRasterRenderGraphBuilder builder = renderGraph.AddRasterRenderPass("MRT Pass", out PassData passData))
    {
        UniversalResourceData resourceData = frameData.Get<UniversalResourceData>();

        passData.colorTextureHandle = resourceData.activeColorTexture;
        passData.texName = m_texName;
        passData.material = m_Material;


        builder.UseTexture(passData.colorTextureHandle);
        for (int i = 0; i < 3; i++)
        {
            builder.SetRenderAttachment(handles[i], i);
        }
        builder.SetRenderFunc((PassData data, RasterGraphContext rgContext) => ExecutePass(data, rgContext));
    }
}

static void ExecutePass(PassData data, RasterGraphContext rgContext)
{
    data.material.SetTexture(name: data.texName, value: data.colorTextureHandle);
    rgContext.cmd.DrawProcedural(Matrix4x4.identity, data.material, shaderPass: 0, MeshTopology.Triangles, vertexCount: 3);
}

RTHandle[] m_RTs = new RTHandle[3];
RenderTargetInfo[] m_RTInfos = new RenderTargetInfo[3];

public void Setup(string texName, Material material, RenderTexture[] renderTextures)
{
    m_Material = material;
    m_texName = string.IsNullOrEmpty(texName) ? "_ColorTexture" : texName;

    for (int i = 0; i < 3; i++)
    {
        RenderTexture rednerTexture = renderTextures[i];
        if (m_RTs[i] == null || m_RTs[i].rt != rednerTexture)
        {
            if (m_RTs[i] != null)
            {
                m_RTs[i].Release();
            }

            m_RTs[i] = RTHandles.Alloc(rednerTexture, name: $"ChannelTexture[{i}]");
            m_RTInfos[i] = new RenderTargetInfo
            {
                format = rednerTexture.graphicsFormat,
                height = rednerTexture.height,
                width = rednerTexture.width,
                bindMS = rednerTexture.bindTextureMS,
                volumeDepth = rednerTexture.volumeDepth,
                msaaSamples = 1,
            };
        }
    }
}

Shader "Hidden/MrtColor"
{
    Properties
    {
        _ColorTexture("ColorTexture", 2D) = "white" {}
    }
    SubShader
    {
        Cull Off
        ZWrite Off

        Pass
        {
            HLSLPROGRAM
            #pragma vertex Vert
            #pragma fragment frag

            #include "Packages/com.unity.render-pipelines.universal/ShaderLibrary/Core.hlsl"
            #include "Packages/com.unity.render-pipelines.core/ShaderLibrary/GlobalSamplers.hlsl"

            TEXTURE2D_X(_ColorTexture);

            struct Attributes
            {
                uint vertexID : SV_VertexID;
                UNITY_VERTEX_INPUT_INSTANCE_ID
            };

            struct Varyings
            {
                float4 positionCS : SV_POSITION;
                float2 texcoord   : TEXCOORD0;
                UNITY_VERTEX_OUTPUT_STEREO
            };

            Varyings Vert(Attributes input)
            {
                Varyings output;

                output.positionCS = GetFullScreenTriangleVertexPosition(input.vertexID);
                output.texcoord = GetFullScreenTriangleTexCoord(input.vertexID);

                return output;
            }
    

            // MRT shader
            struct FragmentOutput
            {
                half4 dest0 : SV_Target0;
                half4 dest1 : SV_Target1;
                half4 dest2 : SV_Target2;
            };

            FragmentOutput frag(Varyings input) : SV_Target
            {
                half4 color = SAMPLE_TEXTURE2D_X(_ColorTexture, sampler_LinearRepeat, input.texcoord); 
                FragmentOutput output;
                output.dest0 = half4(color.r, 0.0, 0.0, 1.0);
                output.dest1 = half4(0.0, color.g, 0.0, 1.0);
                output.dest2 = half4(0.0, 0.0, color.b, 1.0);
                return output;
            }
            ENDHLSL
        }
    }
}

Full screen quad vs full screen triangle

• 사각형의 전체화면 그리기 위해
  • (full screen quad) 삼각형 2개로 사각형을 만드는것이 아닌
  • (full screen triangle) 삼각형을 크게 그려서 삐저나온 부분을 잘라 사각형 그림.

rgContext.cmd.DrawProcedural(Matrix4x4.identity, data.material, shaderPass: 0, MeshTopology.Triangles, vertexCount: 3);

• GPU는 한 번에 하나씩 픽셀을 렌더링하지 않고 2x2에서 8x8 픽셀 사이의 그룹으로 렌더링합니다.

• full screen quad로 그리면 삼각형이 맞닿은 대각선이 겹침

• 겹치는 부분만큼 오버드로우가 생김.

• https://knarkowicz.wordpress.com/2010/10/26/shader-optimizations/

• https://www.saschawillems.de/blog/2016/08/13/vulkan-tutorial-on-rendering-a-fullscreen-quad-without-buffers/

PostProcess

• Compute 셰이더가 좀 더 빠르지만, 여기선 일단 픽셀 셰이더로 구현하는 걸로.

목록

• Bloom
• LightStreak
• EyeAdaptation
• SSAO
• SSGI
• _LightShaft_postprocess
• _SSR

영역 총합 테이블 (summed area table)

• 1984 - Frank
• GDC2003 - Simon

Ref

• https://github.com/QianMo/X-PostProcessing-Library
• https://github.com/UWA-MakeItSimple/Course-PostProcessingEffect

Filter / Blur

• 포스트프로세스니 ComputeShader와, 어차피 흐려지는 것이니 RenderPipeline을 이용하여 다운샘플링 된걸가지고 하는걸 추천.

// 조심
// ref: https://docs.unity3d.com/Manual/SL-PlatformDifferences.html
// Flip sampling of the Texture: 
// The main Texture
// texel size will have negative Y).

#if UNITY_UV_STARTS_AT_TOP
if (_MainTex_TexelSize.y < 0)
{
    uv.y = 1-uv.y;
}
#endif

Average / 평균값

• Mean Filter라 불리기도함
  • Median Filter도 있는데, 이는 평균값에 너무 벗어난 값은 포함하지 않음.
• N * N 블럭
• 주변 픽셀들을 더하고 평균값으로 칠함.

// -1, -1 | 0, -1 | +1, -1
// -1,  0 | 0,  0 | +1,  0
// -1, +1 | 0, +1 | +1, +1

// weight
// 1 | 1 | 1
// 1 | 1 | 1
// 1 | 1 | 1

16 박스

• 바이너리필터를 이용 4개의 샘플링으로 16개의 텍셀의 평균값을 구함

/// 16박스
//   .   | 0,-1 |   .  
// -1, 0 |  .   |  1, 0  
//   .   | 0, 1 |   .  

// weight
// 0 | 1 | 0
// 1 | 0 | 1
// 0 | 1 | 0


/// 64박스
// -2,-2 | -1,-2 | 0,-2 | 1,-2
// -2,-1 | -1,-1 | 0,-1 | 1,-1
// -2, 0 | -1, 0 | 0, 0 | 1, 0
// -2, 1 | -1, 1 | 0, 1 | 1, 1

// weight
// 1 | 1 | 1 | 1
// 1 | 1 | 1 | 1
// 1 | 1 | 1 | 1

9콘 - tent blur

• 바이너리필터를 이용 4개의 샘플링으로 각기다른 가중치를 지닌 9개 텍셀을 얻어옴

// 0,  0 | +1,  0
// 0, +1 | +1, +1

// weight
// 1 | 1
// 1 | 1


// 샘플링하면 다음과 같은 9개의 텍셀 가중치를 지니게 된다
// 1 | 2 | 1
// 2 | 4 | 2
// 1 | 2 | 1

Gaussian / 가우스

• N * N 블럭
• 보다 원본 이미지가 잘 살도록, 중심부에 가중치를 더 준다.

// -1, -1 | 0, -1 | +1, -1
// -1,  0 | 0,  0 | +1,  0
// -1, +1 | 0, +1 | +1, +1

// weight
// Sigma: 1.0 | Kernel Size : 3
// 0.077847 |  0.123317 | 0.077847
// 0.123317 |  0.195346 | 0.123317
// 0.077847 |  0.123317 | 0.077847

• Gaussian Kernel Calculator
• https://www.sysnet.pe.kr/2/0/11623
• https://rastergrid.com/blog/2010/09/efficient-gaussian-blur-with-linear-sampling/

two-pass Gaussian blur

• https://learnopengl.com/Advanced-Lighting/Bloom
• 가로로 한번 블러먹이고
• 세로로 한번 블러먹인다

ex)

5x5 = 0.38774 , 0.24477  , 0.06136
9x9 = 0.227027, 0.1945946, 0.1216216, 0.054054, 0.016216

Bilateral / 쌍방

• 엣지를 보존하면서 노이즈를 제거

Kawase / 카와세

• 대각선을 샘플.

// -1, -1 |   -   | +1, -1
//    -   | 0,  0 |    -   
// -1, +1 |   -   | +1, +1

// weight
// 1/8 |  -  | 1/8
//  -  | 1/2 |  - 
// 1/8 |  -  | 1/8

• https://github.com/JujuAdams/Kawase
• CDEDC2009 - IMAGIRE DAY : 続・レンダリスト養成講座
• GDC2003 - Frame Buffer Postprocessing Effects in DOUBLE-S.T.E.A.L (Wreckless)
• GDC2003 - Real-Time 3D Scene Post-processing
• https://github.com/nobnak/KawaseLightStreakUnity

Dual Filter

// DownSampler pos
// -0.5, -0.5 |   -   | +0.5, -0.5
//    -       | 0,  0 |    -   
// -0.5, +0.5 |   -   | +0.5, +0.5

// DownSampler weight
//   1/8  |   -   |   1/8
//    -   |  1/2  |    - 
//   1/8  |   -   |   1/8

// UpSampler pos
//     -   |     -      |   0, -1   |    -       |    -
//     -   | -0.5, -0.5 |     -     | +0.5, -0.5 |    -
//   -1, 0 |     -      |     -     |    -       |  +1, 0
//     -   | -0.5, -0.5 |     -     | +0.5, -0.5 |    -
//     -   |     -      |   0, +1   |    -       |    -

// UpSampler weight
//     -   |     -      |   1/12    |    -       |    -
//     -   |    1/6     |     -     |    1/6     |    -
//   1/12  |     -      |     -     |    -       |   1/12
//     -   |    1/6     |     -     |    1/6     |    -
//     -   |     -      |   1/12    |    -       |    -

• SIGGRAPH2015 - Bandwidth-Efficient Rendering
• https://github.com/MarcusXie3D/FastBloomForMobiles
  • https://zhuanlan.zhihu.com/p/126351976
• https://developer.arm.com/documentation/102259/0100/Bloom
• Unite Berlin 2018 - Attaining Post Processing Effects on Mid Range Smartphones
  • slide

Radial / 방사형

• 중심에서 원형으로 뻗혀나가는 방사형 블러.
• 줌(Zoom)되는 효과
• https://forum.unity.com/threads/radial-blur.31970/#post-209514
• https://blog.naver.com/mnpshino/221478999495

Bloom

- 원본
- 축소 (w/4, h/4)
  - 밝은부분
    - 블러적용
- 원본과 블러적용 더하기

• 참조 : Filter.md
• [GPU Gems 1] Chapter 21. Real-Time Glow

Bloom Curve

• https://graphtoy.com/
• https://catlikecoding.com/unity/tutorials/custom-srp/post-processing/#2.7

Color Grading LUT

색 보정(Color Grading)에는 여러 수치 보정 수식들이 들어가게 되는데, 이걸 실시간 계산이 아닌 텍스쳐에 구어 연산 부하를 낮출 수 있다.

• unity: 32x32 가 옆으로 32개 => 1024 × 32
  • NeutralLdrLut.png
  • ldr(Low Dynamic Range)

// (r,g,b)
// (0,1,0) +----+ (1,1,0)          (0,1,n) +----+ (1,1,n)
//         |    |           ......         |    |        
// (0,0,0) +----+ (1,0,0)          (0,0,n) +----+ (1,0,n)

• http://ttmayrin.tistory.com/34

// 16 기준

float3 CalcLUT2D( sampler InLUT, float3 InColor )
{
    // requires a volume texture 16x16x16 unwrapped in a 2d texture 256x16
    // can be optimized by using a volume texture
    float2 Offset = float2(0.5f / 256.0f, 0.5f / 16.0f);
    float Scale = 15.0f / 16.0f; 

    // Also consider blur value in the blur buffer written by translucency
    float IntB = floor(InColor.b * 14.9999f) / 16.0f;
    float FracB = InColor.b * 15.0f - IntB * 16.0f;

    float U = IntB + InColor.r * Scale / 16.0f;
    float V = InColor.g * Scale;

    float3 RG0 = tex2D( InLUT, Offset + float2(U               , V) ).rgb;
    float3 RG1 = tex2D( InLUT, Offset + float2(U + 1.0f / 16.0f, V) ).rgb;

    return lerp( RG0, RG1, FracB );
}

float3 CalcLUT3D( sampler InLUT, float3 InColor )
{
    return tex3D( InLUT, InColor * 15.f / 16.f + 0.5f / 16.f ).rgb;
}

색보정 방법

채도/대비/샤픈 마젠타/사이언/그린

• 영화같은 색보정을 위한 가장 중요한 이것
• 하이비드 HYVID STUDIO
  • 눈대중으로 색보정 그만! 기준을 갖고 하는 색보정 (feat 피부톤)
  • 매드맥스 DI 작업자 에릭의 헐리웃 색보정 팁
  • 존윅 4 개봉!! 시선을 사로잡는 존윅의 색보정 비밀! / DI 작업자 질 보그다노비치 색보정 작업방식
• https://en.wikipedia.org/wiki/Zone_System

Ref

• DirectX Shader LUT 필터 코드 구현
• [Unite Seoul 2019] 최재영 류재성 - 일곱개의 대죄 : "애니메이션의 감성을 그대로＂와 “개발 최적화"
• [GPU Gems 2] Chapter 24. Using Lookup Tables to Accelerate Color Transformations

Color Space

Color Wheel

• RGB
• HSL (for hue, saturation, lightness) and HSV (for hue, saturation, value; also known as HSB, for hue, saturation, brightness)
• HCL (Hue-Chroma-Luminance)
• YUV
• ..

ICtCp : ICtCp has near constant luminance, which improves chroma subsampling versus YCBCR YCgCo : 색평면 사이에 상관성이 매우 낮음

CIE RGB CIE XYZ CIE Lab

• ITU1990

Simple

const static half3x3 MAT_RGB_TO_XYZ = {
    0.4124, 0.3576, 0.1805,
    0.2126, 0.7152, 0.0722,
    0.0193, 0.1192, 0.9505
};

const static half3x3 MAT_XYZ_TO_RGB = {
    +3.2405, -1.5371, -0.4985,
    -0.9693, +1.8760, +0.0416,
    +0.0556, -0.2040, +1.0572
};

// ======================================
/// XYZ => Yxy
float SUM_XYZ = dot(float3(1.0, 1.0, 1.0), XYZ);
Yxy.r  = XYZ.g;
Yxy.gb = XYZ.rg / SUM_XYZ;

// ======================================
/// Yxy => XYZ
XYZ.r = Yxy.r * Yxy.g / Yxy. b;
XYZ.g = Yxy.r;
XYZ.b = Yxy.r * (1 - Yxy.g - Yxy.b) / Yxy.b;

CIE Yxy CIE Lab

// https://www.ronja-tutorials.com/post/041-hsv-colorspace/

float3 hue2rgb(float hue)
{
    hue = frac(hue); //only use fractional part
    float r = abs(hue * 6 - 3) - 1; //red
    float g = 2 - abs(hue * 6 - 2); //green
    float b = 2 - abs(hue * 6 - 4); //blue
    float3 rgb = float3(r,g,b); //combine components
    rgb = saturate(rgb); //clamp between 0 and 1
    return rgb;
}

float3 hsv2rgb(float3 hsv)
{
    float3 rgb = hue2rgb(hsv.x); //apply hue
    rgb = lerp(1, rgb, hsv.y); //apply saturation
    rgb = rgb * hsv.z; //apply value
    return rgb;
}

float3 rgb2hsv(float3 rgb)
{
    float maxComponent = max(rgb.r, max(rgb.g, rgb.b));
    float minComponent = min(rgb.r, min(rgb.g, rgb.b));
    float diff = maxComponent - minComponent;
    float hue = 0;
    if(maxComponent == rgb.r)
    {
        hue = 0+(rgb.g-rgb.b)/diff;
    }
    else if(maxComponent == rgb.g)
    {
        hue = 2+(rgb.b-rgb.r)/diff;
    }
    else if(maxComponent == rgb.b)
    {
        hue = 4+(rgb.r-rgb.g)/diff;
    }
    hue = frac(hue / 6);
    float saturation = diff / maxComponent;
    float value = maxComponent;
    return float3(hue, saturation, value);
}

GrayScale / Monochrome

// YUV로 변환후, 밝기만 취하기.
const static half3x3 MAT_RGB_TO_YUV = {
  +0.299, +0.587, +0.114, // 밝기
  -0.147, -0.289, +0.436,
  +0.615, -0.515, -0.100
};

const static half3x3 MAT_YUV_TO_RGB = {
  +1.0, +0.000, +1.140,
  +1.0, -0.396, -0.581,
  +1.0, +2.029, +0.000
};

half YUV_y = mul(MAT_RGB_TO_YUV[0], color.rgb);

Sepia

• MS문서에 나온 SepiaMatrix 이용.
• YIQ나 YUV 이용.

half3x3 MAT_TO_SEPIA = {
    0.393, 0.769, 0.189,   // tRed
    0.349, 0.686, 0.168,   // tGreen
    0.272, 0.534, 0.131    // tBlue
};

half3 sepiaResult = mul(MAT_TO_SEPIA, color.rgb);

// ref: http://www.aforgenet.com/framework/docs/html/10a0f824-445b-dcae-02ef-349d4057da45.htm
// I = 51
// Q = 0

half3x3 MAT_RGB_TO_YIQ = {
    +0.299, +0.587, +0.114,
    +0.596, -0.274, -0.322,
    +0.212, -0.523, +0.311
};

half3x3 MAT_YIQ_TO_RGB = {
    +1.0, +0.956, +0.621,
    +1.0, -0.272, -0.647,
    +1.0, -1.105, +1.702
};

// Cb : -0.2
// Cr : 0.1

const static half3x3 MAT_RGB_TO_YUV = {
  +0.299, +0.587, +0.114, // 밝기
  -0.147, -0.289, +0.436,
  +0.615, -0.515, -0.100
};

const static half3x3 MAT_YUV_TO_RGB = {
  +1.0, +0.000, +1.140,
  +1.0, -0.396, -0.581,
  +1.0, +2.029, +0.000
};

sRGB

• standard RGB
• Rec. 709(다른 이름: Rec.709, BT.709, ITU 709)
  • Rec: Recommendation
  • BT : Broadcast Television
  • ITU : International Telecommunication Union

sRGB Adobe RGB DCI-P3 Rec.2020

하늘과 관련된 모든 조명 파란색을 사용했습니다 . R: 0.4 / G: 0.7 / B: 1. 태양 R: 1 / G: 0.8 / B: 0.6

색온도는 일반적으로 절대 온도의 측정 단위인 K 기호를 사용하여 켈빈으로 표현합니다 주황색(낮음) => 파랑색(높음)

Ref

• https://raphlinus.github.io/color/2021/01/18/oklab-critique.html
• https://bottosson.github.io/posts/oklab/
• https://chrisbrejon.com/cg-cinematography/chapter-2-color-theory/
• The Dimensions of Colour
  • http://www.huevaluechroma.com/index.php
  • https://sites.google.com/site/djcbriggs/1-basic-colour-attributes#h.p_QpXoaFjHTppK

Eye Adaptation / Luminance Adaptation / Auto Exposure

• 광적응
  • 현재밝기와 이전밝기를 이용하여 적절한밝기를 구하고, 이를 원본 이미지에 적용한다.

Overview

1. 현재밝기(lumaAverageCurr)
   • 현재 화면의 평균밝기 저장(사이즈를 줄여가며 1x1텍스쳐로)
   • 휘도 전용, POT(Power of Two), Mipmap 적용(1x1용), R16 색이면 충분.
2. 적절한밝기(lumaAdaptCurr)
   • 이전화면 평균밝기와 비교해서 적절한 밝기 얻기
3. 적절한밝기 적용
   • 앞서구한 적절한 밝기를 원본 이미지에 적용
4. 이전밝기(lumaAdaptPrev)
   • 적절한밝기를 이전밝기에 저장

예제코드

L_w : 해당 픽셀의 밝기 L_avg : 씬 평균 밝기 L_min : 최소밝기. 이 값 아래로 가면강제로 0이됨. L_white(WhitePoint) : 최대밝기. 이 값 위로가면 강제로 1로됨. Logarithmic / Exponential - 로그/지수 성질 이용.

// https://en.wikipedia.org/wiki/Relative_luminance
// https://github.com/Unity-Technologies/Graphics/blob/master/com.unity.render-pipelines.core/ShaderLibrary/Color.hlsl
real Luminance(real3 linearRgb)
{
    return dot(linearRgb, real3(0.2126729, 0.7151522, 0.0721750));
}

간상체T_robs  약 0.2
추상체T_cones 약 0.4

r = p * T_robs + (1 - p) * T_cones

float SensitivityOfRods(float y)
{
   return 0.04 / (0.04 + y);
}

half middleGray = 1.03 - (2 / (2 + log10(lumaAverageCurr + 1)));
half lumaScaled = (lumCurr * middleGrey) / lumaAverageCurr;

half lumaAdaptCurr = lumaAdaptedPrev + (lumaAverageCurr - lumaAdaptedPrev) * (1 - exp(- (dt) / AdaptationConstatnt));
// half sensitivity = SensitivityOfRod(luma)
half lumaAdaptCurr = lumaAdaptedPrev + (lumaAverageCurr - lumaAdaptedPrev) * (1 - exp( -(dt) / sensitivity * _FrameSpeed));

///  ref: Programming Vertex Geometry and Pixel Shaders : High-Dynamic Range Rendering
float lumaScaled = Yxy.r * MiddleGray / (AdaptedLum.x + 0.001f);
Yxy.r = (lumaScaled * (1.0f + lumaScaled / WhitePoint))/(1.0f + lumaScaled);

// HSV로 변경후 V(명도)를 이용하는 방법도 있음. // HSV를 이용하는 방식은 LUT를 이용해서 성능을 올려야함. // Yxy를 이용하는 방식은 충분히 빠름.

AutoKey = saturate(1.5 - (1.5 / (lumaAverageCurr * 0.1 + 1))) + 0.1; 


Color *= Key / (lumaAdaptCurr + 0.0001f);
Color = ToneMap(Color);

/// ref: 2007 Realtime HDR Rendering - Christian Luksch = 13.4. Adaptive Logarithmic Mapping
// _B = 0.5 and 1 
Color = Key
       / log10(lumaAverageCurr + 1)
       * log(Color + 1)
       / log(2 + pow((Color / lumaAverageCurr), log(_B) / log(0.5)) * 8);

/// ref: Perceptual Eects in Real-time Tone Mapping
lumaToned = ToneMap(luma);
rgbL = rgb * (lumaToned * (1 - s)) / luma + (1.05, 0.97, 1.27) * lumaToned * s;

Ref

• GDC2006 - Hdr Meets Black And White 2
• HDRToneMappingCS11
• Reverse engineering the rendering of The Witcher 3, part 2 - eye adaptation
• https://github.com/przemyslawzaworski/Unity3D-CG-programming/tree/master/HDR
• http://developer.download.nvidia.com/SDK/9.5/Samples/samples.html
  • http://developer.download.nvidia.com/SDK/9.5/Samples/DEMOS/Direct3D9/DeferredShading.zip
• High dynamic range : https://www.slideshare.net/cagetu/high-dynamic-range
• Programming Vertex Geometry and Pixel Shaders : Range Mapping, Light Adaptation
• https://www.cg.tuwien.ac.at/research/publications/2007/Luksch_2007_RHR/
• https://knarkowicz.wordpress.com/2016/01/09/automatic-exposure/
• http://resources.mpi-inf.mpg.de/tmo/logmap/
• 블로그
  • canny708 - Eye Adaptation (Automatic Exposure)
  • SuperRealizm - HDR + Bloom
• https://placeholderart.wordpress.com/2014/11/21/implementing-a-physically-based-camera-manual-exposure/
• ShaderX3 Advanced Rendering with DirectX and OpenGL
  • 4.2 Adaptive Glare

FXAA (Fast Approximate Anti-Aliasing)

• 밝기를 분석하여 바뀌는 곳을 기반, 외곽선을 추출하여 블러시켜줌.
• 나중에 AA를 적용할 것을 감안하여, SSR같은 곳에서 적절히 디더링만 시켜주고 블러는 안해주는 식으로 흐림 효과를 AA에 맡길 수 있다.

- Luminance Conversion
  - 픽셀당 8방향의 Luma가 필요하니 미리 계산해서 alpha채널에 넣거나, 그냥 green그대로 쓰기(명암차는 green에 민감하니)
- Local Contrast Check
  - 4방(상하좌우) Luma 차이 계산
  - 차이가 미미하면 AA미적용.
    - early exit
- Vertical/Horizontal Edge Test
  - 8방중 나머지 4방(대각선)의 luma역시 구해서 수평, 수직으로 외곽선 검출하여(Sobel필터 비슷하게) 외곽선의 방향을 얻어낸다.
- End-of-edge Search
  - 앞서구한 방향(수평 혹은 수직)으로 외곽선이 끝나는 양쪽 지점 검출
  - 양쪽 지점에 대한 평균 luma계산
- Blending
  - 외곽선이 끝나는 점 사이에 있는 기준점에 대한 blend값 구함
  - blend값을 이용하여 기준점에서 어느정도 떨어진 픽셀값을 반환
    - RenderTexture(이하 RT)는 Bilinear filtering된것을 이용

// https://en.wikipedia.org/wiki/Relative_luminance
// https://github.com/Unity-Technologies/Graphics/blob/master/com.unity.render-pipelines.core/ShaderLibrary/Color.hlsl
real Luminance(real3 linearRgb)
{
    return dot(linearRgb, real3(0.2126729, 0.7151522, 0.0721750));
}

// 블루값은 거의 영향이 없기에 무시(연산 아낌)
// ref: [FXAA WhitePaper - Nvidia - Timothy Lottes](https://developer.download.nvidia.com/assets/gamedev/files/sdk/11/FXAA_WhitePaper.pdf)
float FxaaLuma(float3 rgb)
{
    return rgb.g * (0.587/0.299) + rgb.r;
}

Ref

• FXAA WhitePaper - Nvidia - Timothy Lottes
• https://catlikecoding.com/unity/tutorials/advanced-rendering/fxaa/
• https://github.com/Unity-Technologies/Graphics/blob/master/com.unity.render-pipelines.high-definition/Runtime/PostProcessing/Shaders/FXAA.hlsl
• assetstore: FXAA Fast Approximate Anti-Aliasing
• SIGGRAPH2011 - FXAA 3.11
  • http://iryoku.com/aacourse/
• Implementing FXAA - Simon Rodriguez
  • [번역] Implementing FXAA
• https://www.slideshare.net/JinWooLee2/anti-aliasing

LightStreak

• aka. CrossFilter
• Streak 자국, 흔적

- 원본
- 축소
  - 밝은부분
    - 블러
    - 카메라 기준 6방향 늘이기
    - 늘린것 합치기
- 원본과 합쳐진 늘려진것 더하기

                source,             _TmpCopy : 복사
               _TmpCopy,          _ScaledTex : 축소
            _ScaledTex,           _BrightTex : 밝기 추출
            _BrightTex, _BaseStarBlurredTex1 : 블러
  _BaseStarBlurredTex1, _BaseStarBlurredTex2 : 블러

  _BaseStarBlurredTex2,            _StarTex0 : 늘리기
             _StarTex0,            _StarTex1 : 늘리기2
             _StarTex1,            _StarTex2 : 광선 임시저장
             ...
           _StarTex2~7,            _StarTex0 : 최종적으로 광선별 합치기
  _TmpCopy + _StarTex0,               source : 원본 텍스쳐에 적용

cropW = srcW - srcW / Scale
cropH = srcH - srcH / Scale
scaledW = cropW / Scale
scaledH = cropH / Scale
brightW = scaledW + 2;
brightH = scaledH + 2;

• Scale을 4로 축소시키면서 축소버퍼의 복사로 픽셀과 텍셀이 1:1 대응함으로 무리없이 복사가능.
• 휘도버퍼에서 패딩은(+2)
  • 블러를 먹이거나 광선을 늘일 경우, 화면을 초과한 픽셀을 검은색으로 처리하기 위한 방법
  • 안하면 텍스쳐 끝부분이 밝으면, 바깥쪽이 모두 밝은것으로 계산되어 필요 이상으로 빛나게됨.
  • 예전 GPU에서는 NPOT인경우 성능하락이 있었음. 최신은 아님.
    • https://forum.unity.com/threads/disadvantages-of-non-power-of-two-textures.814542/#post-5403687

public static int ToPow2RoundUp(int x)
{
  if(x == 0)
  {
    return 0;
  }
  return MakeMSB(x - 1) + 1;
}

public static int MakeMSB(int x)
{
  // 0b_0000_0000_0000_0000_1000_0000_1000_1011
  // MakeMSB(0b_1000) => 0b_1111 //  8 => 15
  // MakeMSB(0b_1111) => 0b_1111 // 15 => 15

  x |= x >> 1;
  x |= x >> 2;
  x |= x >> 4;
  x |= x >> 8;
  x |= x >> 16;
  return x;
}

int topBloomWidth = width >> Properties.DownSampleLevel;
int topBloomHeight = height >> Properties.DownSampleLevel;

w = TextureUtil.ToPow2RoundUp( topBloomWidth), 
h = TextureUtil.ToPow2RoundUp( topBloomHeight), 

brightnessOffsetX = (w - topBloomWidth) / 2;
brightnessOffsetY = (h - topBloomHeight) / 2;

Ref

• 2004 - Directx 9 셰이더 프로그래밍 - 타카시 이마기레
• GDC2003 - Frame Buffer Postprocessing Effects in DOUBLE-S.T.E.A.L (Wreckless)
• A Steerable Streak Filter - Christopher Oat
  • ShaderX3 Advanced Rendering with DirectX and OpenGL
    • 4.1 A Steerable Streak Filter

SSAO

• SSAO / Screen Space Ambient Occlusion
• 주변물체에 의해서 가려지는 곳이 어둡게 되어, 장면의 깊이감을 더함.
• Ambient : 주변
• Occlusion : 차폐, 가려짐

역사

• 2001 : 영화 진주만Pearl Harbor에서 Ambient Occlusion 사용
• 2007 : 게임 Crytek의 Crysis에서 SSAO(Screen Space Ambient Occlusion)등장
  • 실제 지오메트리 데이터를 이용하는게 아니라 화면 공간 뎁스를 이용.
• 2008 : SIGGRAPH2008에서 Nvidia 기존 SSAO개선판 HBAO(Horizon Based Ambient Occulusion)발표.
  • 노멀등 추가정보를 이용.
• 2011 : MSSAO(Multi-Scale Screen Space Ambient Occlusion)
• 2016 : SIGGRAPH2016에서 GTAO(Ground True Ambient Occulusion) 소개.
• ...

샘플링 : 구? 반구?

• Sphere - Crysis Method
  • 특정 점의 구형(Sphere) 주변의 뎁스값을 수집 및 계산 => 구형속 뎁스값이 많아지면 어두워짐
  • 구형 주변의 샘플 갯수가 적으면, 정밀도가 줄어들고 소위 (줄무늬가 드리워지는) banding이라는 현상 발생.

• Normal-oriented Hemisphere
  • 특정 점의 노말방향으로 반구형(Hemisphere) 주변의 뎁스값을 수집 및 계산

구현

• 반구 주변의 점으로 가려짐 정도(Occlusion factor) 계산
  • 성능상 샘플링 갯수를 줄이는게...
• 계산된 가려짐 정도를 블러(Blur)로 적당히 흐려지게 만들기
• 원본 텍스쳐에 적용

// 1. 가려짐 정도(Occlusion factor) 계산
// 2. 계산된 가려짐 정도를 블러(Blur)로 적당히 흐려지게 만들기
// 3. 원본 텍스쳐에 적용
half4 mainTex = SAMPLE_TEXTURE2D(_MainTex, sampler_MainTex, IN.uv);
half ambientOcclusionTex = SAMPLE_TEXTURE2D(_AmbientOcclusionTex, sampler_AmbientOcclusionTex, IN.uv).r;
col.rgb *= ambientOcclusionTex;
return col;

// https://babytook.tistory.com/171
꽁수 1 : 2x downscaled depth buffer 사용
꽁수 2 : 랜덤 ray 벡터를 랜덤 벡터로 reflect (회전 변환보다 싸다)
꽁수 3 : view normal 값을 사용할 수 있다면 self occlusion 을 피하기 위해 sign(dot(ray, n)) 을 곱한다
꽁수 4 : 카메라와 가까이 있는 픽셀의 ray 는 작게 scale 하여 어색함을 줄임
꽁수 5 : 깊이값 비교 함수를 적절히 조절 ^^;

예)

| rt                   | 넓이, 높이 |
| -------------------- | ---------- |
| source               | w, h       |
| _TmpCopy             | w, h       |
| _AmbientOcclusionTex | w/4, h/4   |
| _BlurTex0            | w/8, h/8   |
| _BlurTex1            | w/16, h/16 |

              source,             _TmpCopy : 복사
            _TmpCopy, _AmbientOcclusionTex : Occlusion factor 계산
_AmbientOcclusionTex,            _BlurTex0 : 블러 - DownSampling
           _BlurTex0,            _BlurTex1 : 블러 - DownSampling
           _BlurTex1,            _BlurTex0 : 블러 - UpSampling
           _BlurTex0, _AmbientOcclusionTex : 블러 - UpSampling
            _TmpCopy,               source : 원본 텍스쳐에 적용

흠 어차피 블러시킬것이고 채널 하나밖에 안쓰니, (r 혼은 rgb 채널 쓰는) Bloom효과에서 안쓰는 채널(a) 이용하면 괜춘할듯.

case

// [assetstore - Fast SSAO ( Mobile , URP , VR , AR , LWRP )](https://assetstore.unity.com/packages/vfx/shaders/fullscreen-camera-effects/fast-ssao-mobile-urp-vr-ar-lwrp-169024)
// 깊이텍스쳐를 이용해서 노말을 구하고, 노말방향으로 레이를 쏴서 AO를 구함.
// 빠름은 1번, 보통은 3번의 occlusion을 구하도록 코드가 작성됨.

float depthDiff = srcDepth - dstDepth;
float occlusion = step(0.02h, depthDiff) * (1.0h - smoothstep(0.02h, _Area, depthDiff));

/// [canny - SSAO (Screen Space Ambient Occlusion)](https://blog.naver.com/canny708/221878564749)
// 중앙점 픽셀로부터 일정 범위 내에 있는 랜덤 위치의 샘플 좌표의 위치 값과 노멀 벡터를 얻어온 후,
// 중앙점의 노멀과 비교하여 각도가 가파를수록, 위치가 가까울수록 차폐의 영향이 크다고 판단.
// depth버퍼로 positionWS랑 normal을 구할 수 있다.

inline float Random(in float2 uv)
{
    // 렌덤텍스쳐 이용하는 방법
    // return SAMPLE_TEXTURE2D(_RandTex, sampler_RandTex, uv).r;

    // 그냥 계산하는 방법
    return frac(sin(dot(uv, float2(12.9898, 78.233))) * 43758.5453);
}

float3 GetWorldSpacePosition(in float2 uv)
{
    float sceneRawDepth = SampleSceneDepth(uv);
    return ComputeWorldSpacePosition(uv, sceneRawDepth, UNITY_MATRIX_I_VP);
}

inline float3 GetNormal(in float2 uv)
{
    return SampleSceneNormals(uv);
}

float3 srcPos = GetWorldSpacePosition(IN.uv);
float3 srcNormal = GetNormal(IN.uv);

const int SAMPLE_COUNT = 32;
float AO = 0;

// 매 계산마다 depth를 불러오니 => 32번 depth를 불러온다
for (int i = 0; i < SAMPLE_COUNT; ++i)
{
    float2 dstUV = IN.uv + (float2(Random(IN.uv.xy + i), Random(IN.uv.yx + i)) * 2 - 1) / _ScreenParams.xy * _Radius;
    float3 dstPos = GetWorldSpacePosition(dstUV);

    float3 distance = dstPos - srcPos;
    float3 direction = normalize(distance);
    float delta = length(distance) * _Scale;

    AO += max(0, dot(srcNormal, direction) - _Bias) * (1 / (1 + delta)) * _Amount;
}

AO /= SAMPLE_COUNT;
AO = 1 - AO;

return half4(AO.xxx, 1);

// [(X) SSAO (Screen Space Ambient Occlusion) 처리 기법(소스포함)](http://eppengine.com/zbxe/programmig/2982)
// - backup : https://babytook.tistory.com/158
// 장점 : 따로 blur pass 를 실행할 필요가 없다. ( 랜덤맵 텍스쳐를 활용하기에 가능한 부분)
//        시각적 품질이 나쁘지 않다.
//        노멀 버퍼없이도 깔끔하다.
// 단점 : 전통적 ssao 보다는 좀 느리다.
//        역시 노멀 버퍼를 사용하지 않는 것 으로 인한 약간의 시각적 어색함이 존재.. 약간...
// 

uniform sampler2D som;  // Depth texture 
uniform sampler2D rand; // Random texture
uniform vec2 camerarange = vec2(1.0, 1024.0);

float pw = 1.0/800.0*0.5;
float ph = 1.0/600.0*0.5; 

float readDepth(in vec2 coord) 
{ 
    if (coord.x<0||coord.y<0) return 1.0;
    float nearZ = camerarange.x; 
    float farZ =camerarange.y; 
    float posZ = texture2D(som, coord).x;  
    return (2.0 * nearZ) / (nearZ + farZ - posZ * (farZ - nearZ)); 
}  

float compareDepths(in float depth1, in float depth2,inout int far) 
{ 
    float diff = (depth1 - depth2)*100; //depth difference (0-100)
    float gdisplace = 0.2; //gauss bell center
    float garea = 2.0; //gauss bell width 2

    //reduce left bell width to avoid self-shadowing
    if (diff < gdisplace)
    {
        garea = 0.1;
    }
    else
    {
        far = 1;
    }
    float gauss = pow(2.7182,-2*(diff-gdisplace)*(diff-gdisplace)/(garea*garea));

    return gauss;
} 

float calAO(float depth,float dw, float dh) 
{ 
    float temp = 0;
    float temp2 = 0;
    float coordw = gl_TexCoord[0].x + dw/depth;
    float coordh = gl_TexCoord[0].y + dh/depth;
    float coordw2 = gl_TexCoord[0].x - dw/depth;
    float coordh2 = gl_TexCoord[0].y - dh/depth;

    if (coordw  < 1.0 && coordw  > 0.0 && coordh < 1.0 && coordh  > 0.0)
    {
        vec2 coord = vec2(coordw , coordh);
        vec2 coord2 = vec2(coordw2, coordh2);
        int far = 0;
        temp = compareDepths(depth, readDepth(coord),far);
        //DEPTH EXTRAPOLATION:
        if (far > 0)
        {
            temp2 = compareDepths(readDepth(coord2),depth,far);
            temp += (1.0-temp)*temp2;
        }
    }
    return temp; 
}  

void main(void) 
{ 
    //randomization texture:
    vec2 fres = vec2(20,20);
    vec3 random = texture2D(rand, gl_TexCoord[0].st*fres.xy);
    random = random*2.0-vec3(1.0);

    //initialize stuff:
    float depth = readDepth(gl_TexCoord[0]); 
    float ao = 0.0;

    for(int i=0; i<4; ++i)
    { 
        //calculate color bleeding and ao:
        ao+=calAO(depth,  pw, ph); 
        ao+=calAO(depth,  pw, -ph); 
        ao+=calAO(depth,  -pw, ph); 
        ao+=calAO(depth,  -pw, -ph);

        ao+=calAO(depth,  pw*1.2, 0); 
        ao+=calAO(depth,  -pw*1.2, 0); 
        ao+=calAO(depth,  0, ph*1.2); 
        ao+=calAO(depth,  0, -ph*1.2);

        //sample jittering:
        pw += random.x*0.0007;
        ph += random.y*0.0007;

        //increase sampling area:
        pw *= 1.7;
        ph *= 1.7;
    }        

    //final values, some adjusting:
    vec3 finalAO = vec3(1.0-(ao/32.0));


    gl_FragColor = vec4(0.3+finalAO*0.7,1.0); 
}

Ref

• Game Engine Gems, Volume 1 - 18. Fast Screen-space Ambient Occlusion and Indirect Lighting
• https://babytook.tistory.com/171
• https://github.com/sebastianhein/urp-ssao
• https://learnopengl.com/Advanced-Lighting/SSAO
• https://ogldev.org/www/tutorial45/tutorial45.html
  • https://ogldev.org/www/tutorial46/tutorial46.html
• https://github.com/Unity-Technologies/Graphics/blob/master/com.unity.render-pipelines.universal/ShaderLibrary/SSAO.hlsl
• SIGGRAPH2007 - Finding Next Gen - CryEngine 2
• GDC2008 - Real-Time Depth Buffer Based Ambient Occlusion
• SIGGRAPH2008 - Image-Space Horizon-Based Ambient Occlusion
• SIGGRAPH2009 - Multi-LayerDual-ResolutionScreen-SpaceAmbient Occlusion
• SIGGRAPH2010 - Rendering techniques in Toy Story 3
• SIGGRAPH2016 - Practical Real-Time Strategies for Accurate Indirect Occlusion
• 일어
  • http://maverickproj.web.fc2.com/d3d11_24.html
  • https://marina.sys.wakayama-u.ac.jp/~tokoi/?date=20101122

ToneMapping

• 컴퓨터 모니터와 같은 LDR 매체에서 볼 수 있지만 HDR 이미지의 선명도와 톤 범위를 갖는 결과 이미지를 생성

• Tone-mapping -> 디스플레이에 출력가능한 값
• 평균 Luminance
• Luminance 중심으로 0~1 : Tone Mapping
• RGB -> CIE XYZ -> CIE Yxy -> y : Luminance

Reinhard

float3 Reinhard(float3 v)
{
    return v / (1.0f + v);
}

float3 Reinhard_extended(float3 v, float max_white)
{
    float3 numerator = v * (1.0f + (v / float3(max_white * max_white)));
    return numerator / (1.0f + v);
}

Flimic

• Jim Hejl and Richard Burgess-Dawson
  • color_Linear => flmic => result_Gamma

half3 TonemapFilmic(half3 color_Linear)
{
    // optimized formula by Jim Hejl and Richard Burgess-Dawson
    half3 X = max(color_Linear - 0.004, 0.0);
    half3 result_Gamma = (X * (6.2 * X + 0.5)) / (X * (6.2 * X + 1.7) + 0.06);
    return pow(result_Gamma, 2.2); // convert Linear Color
}

Uncharted2

• John Hable

float3 Uncharted2_tonemap_partial(float3 x)
{
    const float A = 0.15f;
    const float B = 0.50f;
    const float C = 0.10f;
    const float D = 0.20f;
    const float E = 0.02f;
    const float F = 0.30f;
    return ((x * (A * x + C * B) + D * E) / (x * (A * x + B) + D * F)) - (E / F);
}

float3 Uncharted2_filmic(float3 v)
{
    float exposure_bias = 2.0f;
    float3 curr = Uncharted2_tonemap_partial(v * exposure_bias);

    float3 W = 11.2f;
    float3 white_scale = 1.0 / Uncharted2_tonemap_partial(W);
    return curr * white_scale;
}

Flimic_Hejl2015

half3 TonemapFilmic_Hejl2015(half3 hdr, half whitePoint)
{
    half4 vh = half4(hdr, whitePoint);
    half4 va = (1.425 * vh) + 0.05;
    half4 vf = ((vh * va + 0.004) / ((vh * (va + 0.55) + 0.0491))) - 0.0821;
    return vf.rgb / vf.aaa;
}

Unreal3

• Used in Unreal Engine 3 up to 4.14. Adapted to be close to ACES, with similar range.

float3 Unreal3(float3 color_Linear)
{
    float3 result_Gamma = color_Linear / (color_Linear + 0.155) * 1.019;
    return pow(result_Gamma, 2.2); // convert Linear Color
}

ACES

• based on ‘ACES Filmic Tone Mapping Cuve‘ by Narkowicz in 2015.
• unreal 4.8

// sRGB => XYZ => D65_2_D60 => AP1 => RRT_SAT
static const float3x3 ACESInputMat =
{
    {0.59719, 0.35458, 0.04823},
    {0.07600, 0.90834, 0.01566},
    {0.02840, 0.13383, 0.83777}
};

// ODT_SAT => XYZ => D60_2_D65 => sRGB
static const float3x3 ACESOutputMat =
{
    { 1.60475, -0.53108, -0.07367},
    {-0.10208,  1.10813, -0.00605},
    {-0.00327, -0.07276,  1.07602}
};

float3 RRTAndODTFit(float3 v)
{
    float3 a = v * (v + 0.0245786f) - 0.000090537f;
    float3 b = v * (0.983729f * v + 0.4329510f) + 0.238081f;
    return a / b;
}

float3 ACES_Fitted(float3 color)
{
    color = mul(ACESInputMat, color);

    // Apply RRT and ODT
    color = RRTAndODTFit(color);

    color = mul(ACESOutputMat, color);

    // Clamp to [0, 1]
    color = saturate(color);

    return color;
}

Uchimura

• Practical HDR and Wide Color Techniques inGran TurismoSPORT
• linear to gamma
• CEDEC2017 "HDR, Theory and Practice"
• https://www.slideshare.net/nikuque/hdr-theory-and-practicce-jp

Others

• kawase
  • http://cedec.cesa.or.jp/2016/session/ENG/13532.html
  • CEDEC2016 - HDR Output Theory and Practice by Silicon Studio
• LottesVDR
  • GDC2016 - Advanced Techniques and Optimization of HDR VDR Color Pipelines

Ref

• https://danielilett.com/2019-05-01-tut1-1-smo-greyscale/
• https://en.wikipedia.org/wiki/Grayscale
• https://blog.ggaman.com/965
• https://docs.microsoft.com/en-us/archive/msdn-magazine/2005/january/net-matters-sepia-tone-stringlogicalcomparer-and-more
• https://www.gdcvault.com/play/1012351/Uncharted-2-HDR
  • https://www.slideshare.net/naughty_dog/lighting-shading-by-john-hable
  • http://filmicworlds.com/blog/filmic-tonemapping-operators/
  • https://github.com/dmnsgn/glsl-tone-map
• https://twitter.com/jimhejl/status/633777619998130176
• https://github.com/tizian/tonemapper
• https://64.github.io/tonemapping/
• https://www.slideshare.net/nikuque/hdr-theory-and-practicce-jp
• Color Control : https://www.slideshare.net/cagetu/928501785227148871
• SIGGRAPH2016 - Technical Art of Uncharted 4
  • https://www.slideshare.net/DaeHyekKIM/naughty-dog-techartfinalpart1tonemappingamphdrcolorgrading
• GDC2017 - High Dynamic Range Color Grading and Display in Frostbite
• https://mynameismjp.wordpress.com/2010/04/30/a-closer-look-at-tone-mapping/
• http://renderwonk.com/publications/s2010-color-course/
• shadertoy - Tone mapping
• GDC2004 - Advanced Depth of Field
• GDC2009 - Star Ocean 4 - Flexible Shader Managment and Post-processing

Chromatic Aberration

• 빛의 파장에 따라 다른 초점 거리를 가짐.
• Chromatic : 색채
• Aberration : 일탈 / 변형

• 종 색수차(Longitudinal/Axial Chromatic Aberration)
• 횡 색수차(Lateral/Yanal Chromatic Aberration)

// ref: https://github.com/Unity-Technologies/Graphics/blob/master/com.unity.render-pipelines.universal/Shaders/PostProcessing/UberPost.shader

#define DistCenter              _Distortion_Params1.xy
#define DistAxis                _Distortion_Params1.zw
#define DistTheta               _Distortion_Params2.x
#define DistSigma               _Distortion_Params2.y
#define DistScale               _Distortion_Params2.z
#define DistIntensity           _Distortion_Params2.w

#define ChromaAmount            _Chroma_Params.x

float2 DistortUV(float2 uv)
{
    // Note: this variant should never be set with XR
    #if _DISTORTION
    {
        uv = (uv - 0.5) * DistScale + 0.5;
        float2 ruv = DistAxis * (uv - 0.5 - DistCenter);
        float ru = length(float2(ruv));

        UNITY_BRANCH
        if (DistIntensity > 0.0)
        {
            float wu = ru * DistTheta;
            ru = tan(wu) * (rcp(ru * DistSigma));
            uv = uv + ruv * (ru - 1.0);
        }
        else
        {
            ru = rcp(ru) * DistTheta * atan(ru * DistSigma);
            uv = uv + ruv * (ru - 1.0);
        }
    }
    #endif

    return uv;
}

float2 uvDistorted = DistortUV(uv);
half3 color = (0.0).xxx;

float2 coords = 2.0 * uv - 1.0;
float2 end = uv - coords * dot(coords, coords) * ChromaAmount;
float2 delta = (end - uv) / 3.0;

half r = SAMPLE_TEXTURE2D_X(_SourceTex, sampler_LinearClamp, uvDistorted                ).x;
half g = SAMPLE_TEXTURE2D_X(_SourceTex, sampler_LinearClamp, DistortUV(delta + uv)      ).y;
half b = SAMPLE_TEXTURE2D_X(_SourceTex, sampler_LinearClamp, DistortUV(delta * 2.0 + uv)).z;

color = half3(r, g, b);

// ref: [2019 - [Unite Seoul 2019] 최재영 류재성 - 일곱개의 대죄 : "애니메이션의 감성을 그대로"와 "개발 최적화."](https://youtu.be/0LwlNVS3FJo?t=1988)
half3 mainTex = SAMPLE_TEXTURE2D(_MainTex, sampler_MainTex, IN.uv).rgb;
half k = _ParamK;
half kcube = _ParamKcube;

float2 centerUV = (IN.uv - 0.5);
half r2 = dot(centerUV, centerUV);
half f = 0;
if (kcube == 0)
{
    f = 1 + r2 * k;
}
else
{
    f = 1 + r2 * (k + kcube * sqrt(r2));
}

float2 chromaticUV = 0.5 + centerUV * f;
half3 final_chromatic = half3(SAMPLE_TEXTURE2D(_MainTex, sampler_MainTex, chromaticUV).rg, mainTex.b);

Ref

• The Witcher 3 - Chromatic Aberration
• Chromatic Aberration 1 (색수차)

Depth of Field

• DOF: Depth Of Field

• CoC: Circle of Confusion

• Bokeh

• 초점을 맞춘 이 위치의 물체를 제외하고는 전경과 배경이 모두 흐려짐.

• 흐릿함의 정도는 초점에서 멀어질수록(비선형적으로) 커짐.

CoC(Circle of Confusion)

착란원

// SIGGRAPH2018 A Life of a Bokeh

A         = Apertune diameter         // 센서크기
F         = Focal length              // 피사체와의 거리
P         = Focus distance            // 초점거리
MaxBgdCoc = (A * F)/(P - F)           // 배경의 최대 착란원의 반경
Coc(z)    = (1 - (P / z)) * MaxBgdCoc // 혼란원(확산원)

float z = LinearEyeDepth(SAMPLE_DEPTH_TEXTURE(_CameraDepthTexture,i.uv));
float CoC = (1 - (_FocusDistance / z));
CoC = clamp(CoC, -1, 1) * _MaxBgdCoc;

REF

• https://catlikecoding.com/unity/tutorials/advanced-rendering/depth-of-field/
• slideshare.net/youpyo/d2-depth-of-field
• https://blog.naver.com/eryners/220152909470
• ShaderX3 Advanced Rendering with DirectX and OpenGL
  • 4.4 Improved Depth-of-Field Rendering
• https://qiita.com/UWATechnology/items/2742e5726f3b8a01796d
• SIGGRAPH2018 A Life of a Bokeh
  • A Life Of A Bokeh Advances in real-time rendering SIGGRAPH 2018 UE4
• https://www.slideshare.net/ohyecloudy/gpu-gems3-chapter-28-practical-post-process-depth-of-field

HDR

High Dynamic Range

• 컴퓨터 모니터와 같은 LDR 매체에서 볼 수 있지만 HDR 이미지의 선명도와 톤 범위를 갖는 결과 이미지를 생성
• Bloom(빛발산)/Eye Adaptation(광적응)/ToneMapping/감마보정

Bloom

1. 일정 밝기의 한계점(Threshold)을 넘는 부분을 저장
2. 블러
3. 원본이미지와 합치기

Eye Adaptation

• 참고: EyeAdaptation

ToneMapping

• 참고: ToneMapping

감마보정

• linear color를 마지막에 gamma적용시킴
• pow(linearColor, 2.2);

RGBM과 LogLUV

TODO

• http://egloos.zum.com/cagetu/v/5697725
• https://kblog.popekim.com/2013/11/blendable-rgbm.html

통합셰이더 예제

• CHROMATIC_ABERRATION 색수차 - 렌즈 무지개현상
• BLOOM(dirt w/o)
• Vignette 카메라 가장자리 어둡게 하는 효과
• ApplyColorGrading
• FlimGrain 필름 표면 미세한 입자(노이즈)
• DITHERING

Ref

• [Ndc12] 누구나 알기쉬운 hdr과 톤맵핑 박민근
• Ndc11 이창희_hdr
• HDR - 김정희, 최유표
• KGC2014 - 울프나이츠 엔진 프로그래밍 기록
• https://www.ea.com/frostbite/news/high-dynamic-range-color-grading-and-display-in-frostbite
• https://research.activision.com/publications/archives/hdr-in-call-of-duty
• http://diaryofagraphicsprogrammer.blogspot.com/2013/09/call-for-new-post-processing-pipeline.html

렌즈 플레어

• [0821 박민근] 렌즈 플레어(lens flare)
• http://cg.skku.edu/pub/2013-lee-egsr-matrixflare
• https://github.com/modanhan/Unity-Lens-Flare-2019
• https://opengameart.org/content/lens-flares-and-particles?page=1
• http://john-chapman-graphics.blogspot.com/2013/02/pseudo-lens-flare.html
• https://developer.arm.com/documentation/102259/0100/Dirty-lens-effect

LightShaft_Postprocess

• Shaft : 한 줄기 광선, 전광
• wiki : aka. Sun beam / Sun Shafts/ Crepuscular Ray / God Ray

Frustum

• Frustum : 절두체

• Camera.stereoActiveEye
  • Camera.MonoOrStereoscopicEye.Mono : 파랑
  • Camera.MonoOrStereoscopicEye.Left : 초록
  • Camera.MonoOrStereoscopicEye.Right : 빨강

_materia_LightShaft.SetVector("_CameraPositionWS", camera.transform.position);
_materia_LightShaft.SetMatrix("_Matrix_CameraFrustum", FrustumCorners(camera));


private Matrix4x4 FrustumCorners(Camera cam)
{
  // ref: http://hventura.com/unity-post-process-v2-raymarching.html

    Transform camtr = cam.transform;

    // frustumCorners
    //    1  +----+ 2
    //       |    |
    //    0  +----+ 3
    Vector3[] frustumCorners = new Vector3[4];
    cam.CalculateFrustumCorners(
        new Rect(0, 0, 1, 1),  // viewport
        cam.farClipPlane,      // z
        cam.stereoActiveEye,   // eye
        frustumCorners         // outCorners
    );

    // frustumVectorsArray
    //    3  +----+ 2
    //       |    |
    //    0  +----+ 1
    Matrix4x4 frustumVectorsArray = Matrix4x4.identity;
    frustumVectorsArray.SetRow(0, camtr.TransformVector(frustumCorners[0]));
    frustumVectorsArray.SetRow(1, camtr.TransformVector(frustumCorners[3]));
    frustumVectorsArray.SetRow(2, camtr.TransformVector(frustumCorners[1]));
    frustumVectorsArray.SetRow(3, camtr.TransformVector(frustumCorners[2]));

    // in Shader
    // IN.uv
    // (0,1) +----+ (1,1)
    //       |    |
    // (0,0) +----+ (1,0)
    //
    // int frustumIndex = (int)(IN.uv.x +  2 * IN.uv.y);
    //    2  +----+ 3
    //       |    |
    //    0  +----+ 1
    return frustumVectorsArray;
}

struct Ray
{
    float3 origin;
    float3 direction;
    float3 energy;
};

Ray CreateRay(in float3 origin, in float3 direction)
{
    Ray ray;
    ray.origin = origin;
    ray.direction = direction;
    ray.energy = float3(1, 1, 1);
    return ray;
}

Ray CreateCameraRay(in float2 uv)
{
    // 원점이 화면 중앙이 되도록 uv [0, 1]을 [-1, 1]로 변경.
    float2 screenUV = uv * 2.0f - 1.0f;

    // -1이 있는 이유는 unity_CameraToWorld의 구성요소가 camera.cameraToWorldMatrix와는 다르기 때문이다.
    float4x4 negativeMat = float4x4(
        1,  0,  0,  0,
        0,  1,  0,  0,
        0,  0, -1,  0,
        0,  0,  0,  1
    );
    float4x4 cameraToWorldMatrix = mul(unity_CameraToWorld, negativeMat);

    // 카메라 공간 (0, 0, 0)으로 cameraPositionWS 좌표를 얻어와서
    float3 cameraPositionWS = mul(cameraToWorldMatrix, float4(0.0f, 0.0f, 0.0f, 1.0f)).xyz;
    
    // 프로젝션 (curr.x, curr.y, 1)에서 currPositionWS를 얻어와서 ray의 방향을 구한다.
    float3 currPositionCamera = mul(unity_CameraInvProjection, float4(screenUV, 1.0f, 1.0f)).xyz;
    float3 currPositionWS = mul(cameraToWorldMatrix, float4(currPositionCamera, 0.0f)).xyz;
    float3 rayDirection = normalize(currPositionWS);

    return CreateRay(cameraPositionWS, rayDirection);
}

float4x4 _Matrix_CameraFrustum;

// IN.uv
// (0,1) +----+ (1,1)
//       |    |
// (0,0) +----+ (1,0)

// frustumIndex
//    2  +----+ 3
//       |    |
//    0  +----+ 1
int   frustumIndex          = (int)(IN.uv.x +  2 * IN.uv.y);
half3 frustumPositionCamera = _Matrix_CameraFrustum[frustumIndex].xyz;

frustumPositionWS와 normalize(frustumPositionWS)

예

// 1. shadow맵과 depth 맵을 이용하여 마스크 맵을 만든다
#include "Packages/com.unity.render-pipelines.universal/ShaderLibrary/Shadows.hlsl"
#include "Packages/com.unity.render-pipelines.universal/ShaderLibrary/DeclareDepthTexture.hlsl"

// 카메라의 Frustum의 각 꼭지점을 구하고, normalize하여 해당 방향으로 ray를 쏜다

half stepDistance = _MinDistance + (step * Random(rayDirWS.xy, _Time.y * 100));

// 2. 합한다
color = lerp(mainTex, _MainLightColor.rgb, lightShaftMask);

관련자료

• Shadow Volume Algorithm (Modified) [ MAX 1986 ]
• Slice-based volume-rendering technique [ Dobashi & Nishta & Yamamoto 2002 ]
• Hardware Shadow Map [ Mitchell 2004 ]
• Polygonal Volume [ James 2003 Based On Radomir Mech 2001]
• Volumetric Light Scattering [ Hoffman & Preetham 2003 ]

Ref

• GDC2008 - Crysis Next-Gen Effects
• https://developer.nvidia.com/gpugems/gpugems3/part-ii-light-and-shadows/chapter-13-volumetric-light-scattering-post-process
  • https://zhuanlan.zhihu.com/p/405975067
  • https://blog.csdn.net/qq_38275140/article/details/91049023
  • https://github.com/togucchi/urp-postprocessing-examples/tree/main/Assets/LightShaft
• https://blog.naver.com/sorkelf/40152690614
• https://blog.naver.com/eryners/110127353463
• http://maverickproj.web.fc2.com/pg65.html
• ShaderX3 Advanced Rendering with DirectX and OpenGL
  • 8.1 Light Shaft Rendering

MotionBlur

• Frame Blur

• Position Reconstruction

• Velocity Buffer

• https://blog.naver.com/sorkelf/40158984437

• https://medium.com/spaceapetech/motion-blur-for-mobile-devices-in-unity-656c8047508

2023 - [GDC2023] Stupid OpenGL Shader Tricks by Simon Green, NVIDIA Image space (2.5D) motion blur

• 3 stages:
  • Render scene to texture
    • At current time
  • Calculate velocity at each pixel
    • Using vertex shader
    • Calculate current position – previous position
  • Render motion blurred scene
    • Using fragment shader
    • Look up into scene texture

// Calculate velocity at each pixel
struct a2v {
    float4 coord;
    float4 prevCoord;
    float3 normal;
    float2 texture;
};
struct v2f {
    float4 hpos : HPOS;
    float3 velocity : TEX0;
};

v2f main(a2v in,
    uniform float4x4 modelView,
    uniform float4x4 prevModelView,
    uniform float4x4 modelViewProj,
    uniform float4x4 prevModelViewProj,
    uniform float3 halfWinSize,
)
{
    v2f out;
    
    // transform previous and current pos to eye space
    float4 P = mul(modelView, in.coord);
    float4 Pprev = mul(prevModelView, in.prevCoord);

    // transform normal to eye space
    float3 N = vecMul(modelView, in.normal);
    
    // calculate eye space motion vector
    float3 motionVector = P.xyz - Pprev.xyz;
    
    // calculate clip space motion vector
    P = mul(modelViewProj, in.coord);
    Pprev = mul(prevModelViewProj, in.prevCoord);
    
    // choose previous or current position based
    // on dot product between motion vector and normal
    float flag = dot(motionVector, N) > 0;
    float4 Pstretch = flag ? P : Pprev;
    out.hpos = Pstretch;
    
    // do divide by W -> NDC coordinates
    P.xyz = P.xyz / P.w;
    Pprev.xyz = Pprev.xyz / Pprev.w;
    Pstretch.xyz = Pstretch.xyz / Pstretch.w;
    
    // calculate window space velocity
    float3 dP = halfWinSize.xyz * (P.xyz - Pprev.xyz);
    out.velocity = dP;
    return v2f;
}

// Motion Blur Shader Code
struct v2f {
    float4 wpos : WPOS;
    float3 velocity : TEX0;
};
struct f2f {
    float4 col;
};

f2fConnector main(
    v2f in,
    uniform samplerRECT sceneTex,
    uniform float blurScale = 1.0
)
{
    f2f out;
    
    // read velocity from texture coordinate
    half2 velocity = v2f.velocity.xy * blurScale;
    
    // sample scene texture along direction of motion
    const float samples = SAMPLES;
    const float w = 1.0 / samples; // sample weight
    fixed4 a = 0; // accumulator
    float i;
    for(i = 0; i < samples; i += 1)
    {
        float t = i / (samples-1);
        a = a + x4texRECT(sceneTex, in.wpos + velocity*t) * w;
    }
    out.col = a;
}

SSGI

• SSGI / Screen Space Global Illumination
• Illumination : 조명

구현

• 반구 주변의 점으로 가려짐 정도(Occlusion factor) 계산
  • 성능상 샘플링 갯수를 줄이는게...
• 계산된 가려짐 정도를 블러(Blur)로 적당히 흐려지게 만들기
• 원본 텍스쳐에 적용

Case

// [(X) SSGI 관련 정리 (소스 포함)](http://eppengine.com/zbxe/programmig/2985)
// - backup article: https://babytook.tistory.com/157

uniform sampler2D som;  // Depth texture 
uniform sampler2D rand; // Random texture
uniform sampler2D color; // Color texture

uniform vec2 camerarange = vec2(1.0, 1024.0);

float pw = 1.0/800.0*0.5;
float ph = 1.0/600.0*0.5; 

float readDepth(in vec2 coord) 
{ 
    if (coord.x<0||coord.y<0) return 1.0;
    float nearZ = camerarange.x; 
    float farZ =camerarange.y; 
    float posZ = texture2D(som, coord).x;  
    return (2.0 * nearZ) / (nearZ + farZ - posZ * (farZ - nearZ)); 
}  

vec3 readColor(in vec2 coord) 
{ 
    return texture2D(color, coord).xyz; 
}

float compareDepths(in float depth1, in float depth2) 
{ 
    float gauss = 0.0;
    float diff = (depth1 - depth2)*100.0; //depth difference (0-100)
    float gdisplace = 0.2; //gauss bell center
    float garea = 3.0; //gauss bell width

    //reduce left bell width to avoid self-shadowing
    if (diff<gdisplace) garea = 0.2;

    gauss = pow(2.7182,-2*(diff-gdisplace)*(diff-gdisplace)/(garea*garea));

    return max(0.2,gauss); 
} 

vec3 calAO(float depth,float dw, float dh, inout float ao) 
{ 
    float temp = 0;
    vec3 bleed = vec3(0.0,0.0,0.0);
    float coordw = gl_TexCoord[0].x + dw/depth;
    float coordh = gl_TexCoord[0].y + dh/depth;

    if (coordw  < 1.0 && coordw  > 0.0 && coordh < 1.0 && coordh  > 0.0)
    {
        vec2 coord = vec2(coordw , coordh);
        temp = compareDepths(depth, readDepth(coord));
        bleed = readColor(coord);
    }
    ao += temp;
    return temp*bleed; 
}  

void main(void) 
{ 
    //randomization texture:
    vec2 fres = vec2(20,20);
    vec3 random = texture2D(rand, gl_TexCoord[0].st*fres.xy);
    random = random*2.0-vec3(1.0);

    //initialize stuff:
    float depth = readDepth(gl_TexCoord[0]);
    vec3 gi = vec3(0.0,0.0,0.0); 
    float ao = 0.0;

    for(int i=0; i<8; ++i)
    { 
        //calculate color bleeding and ao:
        gi += calAO(depth,  pw, ph,ao); 
        gi += calAO(depth,  pw, -ph,ao); 
        gi += calAO(depth,  -pw, ph,ao); 
        gi += calAO(depth,  -pw, -ph,ao);

        //sample jittering:
        pw += random.x*0.0005;
        ph += random.y*0.0005;

        //increase sampling area:
        pw *= 1.4; 
        ph *= 1.4;   
    }        

    //final values, some adjusting:
    vec3 finalAO = vec3(1.0-(ao/32.0));
    vec3 finalGI = (gi/32)*0.6;

    gl_FragColor = vec4(readColor(gl_TexCoord[0])*finalAO+finalGI,1.0); 
}  

Ref

• https://www.slideshare.net/jangho/real-time-global-illumination-techniques
• https://forum.unity.com/threads/my-ssao-ssgi-prototype.78566/#post-521710
• https://github.com/demonixis/StylisticFog-URP/tree/master/Assets/StylisticFog-URP
• https://people.mpi-inf.mpg.de/~ritschel/SSDO/index.html
• 用Unity SRP实现SSGI，和手机端的效果测试（一）

SSR

• SSR(Screen Space Reflection)

주의점

• 화면 공간이므로, 당연히 화면밖이나 가려져 있는 것을 반사시키진 못한다
  • 화면 바깥과 가까우면 fadeout
  • 어느정도 구께일때만 반사적용
• 깊이버퍼를 이용함으로, 깊이버퍼를 안쓰는 오브젝트는 반사가 안됨
• 3d ray marching 언더샘플링 오버샘플링
• 모션블러 감소

TODO

• SSR 준비물
  • 색상
  • 깊이(위치를 얻기 위해)
  • 노말
  • 반사 마스크

반사 레이를 쏘는 방식

• http://casual-effects.blogspot.com/2014/08/screen-space-ray-tracing.html

코드 예

_material_SSR.SetMatrix( "_MATRIX_InverseCameraProjection", _camera.projectionMatrix.inverse);

// vert =====================
// 카메라레이
float4 cameraRay = float4(IN.uv * 2 - 1, 1, 1);
cameraRay = mul(_MATRIX_InverseCameraProjection, cameraRay);
OUT.cameraRay = cameraRay.xyz / cameraRay.w;



// frag =====================
half reflectMask = SAMPLE_TEXTURE2D(_ReflectMask_, sampler_ReflectMask, IN.uv).r;
clip(reflectMask - 0.1);

// 반사레이 시작점
half sceneRawDepth = SampleSceneDepth(IN.uv);
half scene01Depth  = Linear01Depth(sceneRawDepth, _ZBufferParams);
half3 reflectRayStartPositionWS = IN.cameraRay * sceen01Depth;

// 입사벡터
half3 incidentVec = normalize(reflectRayStartPositionWS);

// 반사레이벡터
half3 sceneNormal = SampleSceneNormals(IN.uv);
half3 reflectRayDirWS = normalize(reflect(incidentVec, sceneNormal));

// 레이 처리
half step = _MaxDistance / _MaxIteration;
half stepDistance = _MinDistance + step;
half availableThickness = _MaxThickness / _MaxIteration;
int iteratorCount = min(64, _MaxIteration);
half3 reflectionColor = 0;

UNITY_UNROLL
for (int i = 0; i < iteratorCount; ++i)
{
    // 반사레이 도착점 
    half3 reflectRayEndPositionWS = reflectRayStartPositionWS + (reflectRayDirWS * stepDistance);
    float4 reflectRayEndPositionCS = TransformWorldToHClip(reflectRayEndPositionWS);
    float2 reflectRayEndUV = reflectRayEndPositionCS.xy / reflectRayEndPositionCS.w * 0.5 + 0.5;

    bool isValidUV = max(abs(reflectRayEndUV.x - 0.5), abs(reflectRayEndUV.y - 0.5)) <= 0.5;
    if (!isValidUV)
    {
        break;
    }

    half reflectRayEndDepth = ComputeDepth(reflectRayEndPositionCS);
    half sceneReflectRayEndDepth = SampleSceneDepth(reflectRayEndUV);
    half depthDiff = reflectRayEndDepth - sceneReflectRayEndDepth;
    if (0 < depthDiff && depthDiff < availableThickness)
    {
        // 반사색
        reflectionColor = SAMPLE_TEXTURE2D(_MainTex, sampler_MainTex, reflectRayEndUV).rgb;
        break;
    }

    stepDistance += step;
}
return half4(reflectionColor, 1);


// etc ==========================
float ComputeDepth(float4 positionCS)
{
#if defined(SHADER_TARGET_GLSL) || defined(SHADER_API_GLES) || defined(SHADER_API_GLES3)
    return (positionCS.z / positionCS.w) * 0.5 + 0.5;
#else
    return (positionCS.z / positionCS.w);
#endif
}

???
    float sampleDepth = tex_depth.read(tid).x;
    float4 samplePosInCS =  float4(((float2(tid)+0.5)/sceneInfo.ViewSize)*2-1.0f, sampleDepth, 1);
    samplePosInCS.y *= -1;

incidentVec = normalize(rayStartPositionWS)
N

half3 reflectionColor = 0;
if (reflectMask > 0)
{
  reflectionColor = 
}

half3 scaledR = _RayStepScale * R;
for (_MaxRayStep)
{

}

Hi-Z Buffer

• Hierarchical-Z buffer
• 기존 Z buffer를 축소시키며 계층을 만들며(밉맵)
  • 셀은 최소 3×3 셀
• 기존 Z Buffer보다 비교적 적은 샘플 횟수로 교차점을 얻을 수 있다.

• https://zhuanlan.zhihu.com/p/278793984

Ref

• https://www.slideshare.net/xtozero/screen-space-reflection
• http://www.kode80.com/blog/2015/03/11/screen-space-reflections-in-unity-5/
  • https://github.com/kode80/kode80SSR
• Screen Space Reflections : Implementation and optimization
  • src, pt1, pt2
  • 번역1, 번역2
• GPU Pro 5: Hi-Z Screen-Space Cone-Traced Reflections by Yasin Uludag
• https://github.com/Unity-Technologies/Graphics/blob/master/com.unity.postprocessing/PostProcessing/Shaders/Builtins/ScreenSpaceReflections.hlsl
• Unity で Screen Space Reflection の実装をしてみた
  • https://github.com/hecomi/UnityScreenSpaceReflection
• GDC2016 - Low Complexity, High Fidelity: The Rendering of INSIDE
• GDC2016 - Temporal Reprojection Anti-Aliasing in INSIDE
  • https://github.com/playdeadgames/temporal
• GPU Pro 6: Advanced Rendering Techniques - II: RENDERING - 1.2.3 Screen-Space Reflections
• Approximating ray traced reflections using screen-space data by MATTIAS JOHNSSON
• https://dataprocess.tistory.com/315
• https://sakibsaikia.github.io/graphics/2016/12/26/Screen-Space-Reflection-in-Killing-Floor-2.html
• http://roar11.com/2015/07/screen-space-glossy-reflections/
• Screen Space Reflections in The Surge
• SIGGRAPH2015 - Stochastic Screen-Space Reflections
• https://lettier.github.io/3d-game-shaders-for-beginners/screen-space-reflection.html
• https://www.gamedeveloper.com/disciplines/screen-space-reflections-in-blightbound

TAA / Temporal Anti-Aliasing

• [번역] Temporal Anti-Aliasing(TAA) Tutorial
• https://github.com/sienaiwun/TAA_Unity_URP/

Cross Fade Shader

• LOD 변환시 블렌딩

// LOD_FADE_CROSSFADE 정의는
// |- LOD Group 컴포넌트에서
// |- Fade Mode : Cross Fade
// |- 그리고 각 LOD에서 Fade Transition Width값이 0이 아닐때 활성화가 된다.

// 현재 LOD가 `1`에서 `0`으로, 다음 LOD가 `-1`에서 `0`로 전환된다.

// 예)
// RootGameObject | FadeMode : CrossFade
//   - Sphere     | LOD : 2, Fade Transaction Width : 0.3
//   - Cube       | LOD : 1, Fade Transaction Width : 0.2

#pragma multi_compile _ LOD_FADE_CROSSFADE

TEXTURE2D(_DitherTex);   SAMPLER(sampler_DitherTex);
float4 _DitherTex_TexelSize;
SAMPLER(unity_DitherMask); // 유니티가 4x4 디더링 마스크를 제공해준다.

#ifdef LOD_FADE_CROSSFADE
    half2 screenUV = IN.positionNDC.xy / IN.positionNDC.w;
    float fade = unity_LODFade.x;

    // ex-1
    // float dither = (IN.positionCS.y % 32) / 32;
    // clip(fade - dither);

    // ex-2
    // half ditherTex = SAMPLE_TEXTURE2D(_DitherTex, sampler_DitherTex, IN.uv).r;
    // clip(fade - ditherTex);

    // ex-3
    //float2 ditherUV = screenUV.xy * _ScreenParams.xy * _DitherTex_TexelSize.xy;
    //half ditherTex = SAMPLE_TEXTURE2D(_DitherTex, sampler_DitherTex, ditherUV).r;
    //clip(fade - CopySign(ditherTex, fade));

    // ex-4
    // float2 fadeMaskSeed = IN.positionCS.xy;
    // LODDitheringTransition(fadeMaskSeed, fade);

    // ex-5
    //float2 ditherUV = screenUV * _ScreenParams.xy;
    //float DITHER_THRESHOLDS[16] =
    //{
    //    1.0 / 17.0,  9.0 / 17.0,  3.0 / 17.0, 11.0 / 17.0,
    //    13.0 / 17.0,  5.0 / 17.0, 15.0 / 17.0,  7.0 / 17.0,
    //    4.0 / 17.0, 12.0 / 17.0,  2.0 / 17.0, 10.0 / 17.0,
    //    16.0 / 17.0,  8.0 / 17.0, 14.0 / 17.0,  6.0 / 17.0
    //};
    //uint index = (uint(ditherUV.x) % 4) * 4 + uint(ditherUV.y) % 4;
    //clip(fade - CopySign(DITHER_THRESHOLDS[index], fade));

    // ex-6
    float2 ditherUV = screenUV.xy * _ScreenParams.xy / 4.0;
    float dither = tex2D(unity_DitherMask, ditherUV).a;
    clip(fade - CopySign(dither, fade));
 #endif

// built-in
// CGIncludes/UnityCG.cginc
#ifdef LOD_FADE_CROSSFADE
    #define UNITY_APPLY_DITHER_CROSSFADE(vpos)  UnityApplyDitherCrossFade(vpos)
    sampler2D unity_DitherMask;
    void UnityApplyDitherCrossFade(float2 vpos)
    {
        vpos /= 4; // the dither mask texture is 4x4
        float mask = tex2D(unity_DitherMask, vpos).a;
        float sgn = unity_LODFade.x > 0 ? 1.0f : -1.0f;
        clip(unity_LODFade.x - mask * sgn);
    }
#else
    #define UNITY_APPLY_DITHER_CROSSFADE(vpos)
#endif

float2 vpos = IN.screenPos.xy / IN.screenPos.w * _ScreenParams.xy;
UnityApplyDitherCrossFade(vpos);


// URP
// com.unity.render-pipelines.core/ShaderLibrary/Common.hlsl
void LODDitheringTransition(uint2 fadeMaskSeed, float ditherFactor)
{
    // Generate a spatially varying pattern.
    // Unfortunately, varying the pattern with time confuses the TAA, increasing the amount of noise.
    float p = GenerateHashedRandomFloat(fadeMaskSeed);

    // This preserves the symmetry s.t. if LOD 0 has f = x, LOD 1 has f = -x.
    float f = ditherFactor - CopySign(p, ditherFactor);
    clip(f);
}

float CopySign(float x, float s, bool ignoreNegZero = true)
{
#if !defined(SHADER_API_GLES)
    if (ignoreNegZero)
    {
        return (s >= 0) ? abs(x) : -abs(x);
    }
    else
    {
        uint negZero = 0x80000000u;
        uint signBit = negZero & asuint(s);
        return asfloat(BitFieldInsert(negZero, signBit, asuint(x)));
    }
#else
    return (s >= 0) ? abs(x) : -abs(x);
#endif
}

float fade = unity_LODFade.x;
float2 fadeMaskSeed = IN.positionCS.xy;
LODDitheringTransition(fadeMaskSeed, fade);

Ref

• https://docs.unity3d.com/Manual/class-LODGroup.html
• https://github.com/keijiro/CrossFadingLod
• https://blog.naver.com/daehuck/221562275385
• NDC2018 - 'AxE' 클라이언트 최적화와 문제해결-MMORPG를 Unity로 만들 때 겪을 수 있는 이야기
• https://assetstore.unity.com/packages/add-ons/lcfs-lod-cross-fading-shaders-for-urp-177298?locale=ko-KR#content
• LOD asset
  • Ultimate LOD System MT - Automatic LOD Generator, Mesh Simplifier & More
  • Poly Few | Mesh Simplifier and Auto LOD Generator

Cracked Ice

• https://blog.naver.com/daehuck/222228360615
• Unity Shadergraph Tutorial - Cracked Ice
• https://80.lv/articles/how-to-build-cracked-ice-in-material-editor/

Parallax쓰면 조금 들어간것처럼 보임. - 이걸 겹겹히 쌓으면 더 깊이 들어간것처럼 보임

baseColor = blend(mainTex, parallax(heightMap), 0.5)

blendNormal(Normal1, strength(Normal2(uv * 0.25), 0.25))

void Unity_NormalStrength_float(float3 In, float Strength, out float3 Out)
{
    Out = {precision}3(In.rg * Strength, lerp(1, In.b, saturate(Strength)));
}

float2 ParallaxSampling(half3 viewDirTS, half scale, float2 uv)
{
    half h = 0.217637;// pow(0.5, 2.2);
    float2 offset = ParallaxOffset1Step(h, scale, viewDirTS);
    return offset;
}

half2 ParallaxMappingUV(TEXTURE2D_PARAM(heightMap, sampler_heightMap), half2 uv, half3 V_TS, half amplitude)
{
    // 높이 맵에서 높이를 구하고,
    half height = SAMPLE_TEXTURE2D(heightMap, sampler_heightMap, uv).r;
    height = height * amplitude - amplitude / 2.0;

    // 시선에 대한 offset을 구한다.
    // 시선은 반대방향임으로 부호는 마이너스(-) 붙여준다.
    // TS.xyz == TS.tbn

    // TS.n에 0.42를 더해주어서 0에 수렴하지 않도록(E가 너무 커지지 않도록) 조정.
    half2 E = -(V_TS.xy / (V_TS.z + 0.42));

    // 근사값이기에 적절한 strength를 곱해주자.
    return uv + E * height;
}

void ParallaxMapping_float(in float amplitude, in float numSteps, in float4 UV, in float3 viewDir, out float4 Out)
{
    float one = 1;
    float4 ParallaxedTexture = (0, 0, 0, 0);
    float4 UV2 = (0, 0, 0, 0);
    for (float d = 0.0; d < amplitude; d += amplitude / numSteps)
    {
        one = one - (1 / numSteps);
        UV2.xy = UV.xy + ParallaxSampling(viewDir, d * 0.01, UV);
        ParallaxedTexture += saturate(SAMPLE_TEXTURE2D_BIAS(_MainTex, SamplerState_Linear_Repeat, UV2.xy, 0)) * (one + (1 / numSteps));
    }
    Out = saturate(ParallaxedTexture);
}

half ParallaxMappingMask(TEXTURE2D_PARAM(maskTex, sampler_maskTex), in half2 uv, in half3 V_TS, in half parallaxOffset, in int iterCount)
{
    half one = 1;
    half parallaxedMask = 0;
    half result = 1;
    half2 parallaxUV;
    half totalOffset = 0.0;
    parallaxOffset = parallaxOffset * -0.001;

    for (int i = 0; i < iterCount; ++i)
    {
        totalOffset += parallaxOffset;
        parallaxUV = uv + half2(V_TS.x * totalOffset, V_TS.y * totalOffset);
        parallaxedMask = SAMPLE_TEXTURE2D(maskTex, sampler_maskTex, parallaxUV).r;
        result *= clamp(parallaxedMask + (i / iterCount), 0, 1);
    }

    return result;
}

Fake Thickness Window

• ID맵과 Parallax(ID맵)의 겹쳐지는 부분(cross)을 이용.

half2 parallaxUV = ParallaxMappingUV(_IdMaskHeightTex, sampler_IdMaskHeightTex, IN.uv, mul(TBN, V), _ParallaxScale * 0.01);

half idMaskTex = SAMPLE_TEXTURE2D(_IdMaskTex, sampler_IdMaskTex, IN.uv).r;
half idMaskParallaxTex = SAMPLE_TEXTURE2D(_IdMaskTex, sampler_IdMaskTex, parallaxUV).r;

half cross = 0;
if (idMaskTex != idMaskParallaxTex)
{
    cross = 1;
}
return half4(cross, cross, cross, 1);

• Parallax Mapping

half2 ParallaxMappingUV(TEXTURE2D_PARAM(heightMap, sampler_heightMap), half2 uv, half3 V_TS, half amplitude)
{
    // 높이 맵에서 높이를 구하고,
    half height = SAMPLE_TEXTURE2D(heightMap, sampler_heightMap, uv).r;
    height = height * amplitude - amplitude / 2.0;

    // 시선에 대한 offset을 구한다.
    // 시선은 반대방향임으로 부호는 마이너스(-) 붙여준다.
    // TS.xyz == TS.tbn

    // TS.n에 0.42를 더해주어서 0에 수렴하지 않도록(E가 너무 커지지 않도록) 조정.
    half2 E = -(V_TS.xy / (V_TS.z + 0.42));

    // 근사값이기에 적절한 strength를 곱해주자.
    return uv + E * height;
}

• 라스트 오브 어스 2에서
  • ID맵 샘플 + Parallax(ID맵) 샘플 => cross section
  • 노말맵 샘플
    • diffuse는 cross section 이용해서
    • specular는 그대로
  • 환경맵 샘플
    • cross section 노말 이용.

Ref

• https://www.naughtydog.com/blog/naughty_dog_at_siggraph_2020
• SIGGRAPH 2020 - Fake Thickness - The Technical Art of The Last of Us Part II by Waylon Brinck and Steven Tang
• daehuck - 라스트 오브 어스2에 쓰인 유리 쉐이더 유니티로 따라해 보기

Screen Space Decal

• Deferred Decals(Jan Krassnigg, 2010)
• Volume Decals (Emil Persson, 2011)
• SSD : SIGGRAPH2012 - ScreenSpaceDecal
• 큐브를 프로젝터처럼 이용, 화면에 데칼을 그린다.
• 뎁스로부터 포지션을 다시 구축하는 것이므로 Reconstructing position from depth라고도 한다.

1. SSD를 제외한 메쉬들을 화면에 그림
2. SSD 상자를 그림(rasterization)
3. 각 픽셀마다 장면깊이(scene depth)를 읽어옴
4. 그 깊이로부터 3D 위치를 계산함
5. 그 3D 위치가 SSD 상자 밖이면 레젝션(rejection)
6. 그렇지 않으면 데칼 텍스처를 그림

ver1. URP

// NDC에서 depth를 이용 역산하여 데칼 위치를 구하는법.

// vert:
OUT.positionNDC = vertexPositionInput.positionNDC;

// frag:
// ============== 1. 씬뎁스 구하기
half2 uv_Screen = IN.positionNDC.xy / IN.positionNDC.w;
half sceneRawDepth = SampleSceneDepth(uv_Screen);
half sceneEyeDepth = LinearEyeDepth(sceneRawDepth, _ZBufferParams);

// ============== 2. 뎁스로부터 3D위치를 구하기
// positionNDC: [-1, 1]
float2 positionNDC = uv_Screen * 2.0 - 1.0;
half4 positionVS_decal;
positionVS_decal.x = (positionNDC.x * sceneEyeDepth) / unity_CameraProjection._11;
positionVS_decal.y = (positionNDC.y * sceneEyeDepth) / unity_CameraProjection._22;
positionVS_decal.z = -sceneEyeDepth;
positionVS_decal.w = 1;

half4x4 I_MV = mul(UNITY_MATRIX_I_M, UNITY_MATRIX_I_V);
// positionOS_decal: [-0.5, 0.5] // clip 으로 잘려질것이기에 
half4 positionOS_decal = mul(I_MV, positionVS_decal);

// ============== 3. SSD상자 밖이면 그리지않기
clip(0.5 - abs(positionOS_decal.xyz));

// ============== 4. 데칼 그리기
// uv_decal: [0, 1]
half2 uv_decal = positionOS_decal.xz + 0.5;
half2 uv_MainTex = TRANSFORM_TEX(uv_decal, _MainTex);
half4 mainTex = SAMPLE_TEXTURE2D(_MainTex, sampler_MainTex, uv_MainTex);

ver2. URP

• Depth에선 WorldSpace상 좌표를 이용해서 Depth로부터 위치를 구했는데, 최적화를 위해 Object Space상에서 구함(한눈에 봐서는 어색하지만 따라가다보면 말이 되긴 한다)

// 오브젝트 공간의 viewRay를 구하고 depth에 맞추어 데칼 위치를 구하는법.

// vert:
float4x4 I_MV = mul(UNITY_MATRIX_I_M, UNITY_MATRIX_I_V);
OUT.positionOS_camera = mul(I_MV, float4(0, 0, 0, 1)).xyz;

OUT.positionOSw_viewRay.xyz = mul((float3x3)I_MV, -vertexPositionInput.positionVS);
OUT.positionOSw_viewRay.w = vertexPositionInput.positionVS.z;

// frag:
// ============== 1. 씬뎁스 구하기
half2 uv_Screen = IN.positionNDC.xy / IN.positionNDC.w;
half sceneRawDepth = SampleSceneDepth(uv_Screen);
half sceneEyeDepth = LinearEyeDepth(sceneRawDepth, _ZBufferParams);

// ============== 2. 뎁스로부터 3D위치를 구하기
// positionOS_decal: [-0.5, 0.5] // clip 으로 잘려질것이기에
half3 positionOS_decal = IN.positionOS_camera + IN.positionOSw_viewRay.xyz / IN.positionOSw_viewRay.w * sceneEyeDepth;

// ============== 3. SSD상자 밖이면 그리지않기
clip(0.5 - abs(positionOS_decal.xyz));

// ============== 4. 데칼 그리기
// uv_decal: [0, 1]
half2 uv_decal = positionOS_decal.xz + 0.5;
half2 uv_MainTex = TRANSFORM_TEX(uv_decal, _MainTex);
half4 mainTex = SAMPLE_TEXTURE2D(_MainTex, sampler_MainTex, uv_MainTex);

ver. Pope

1. 알파블렌딩
   • Blend Factor
2. 움직이는 물체들
   • 움직이면 데칼 적용안함(Stencil)
3. 옆면이 늘어나요.... ㅜ.ㅠ
   • 투영방향 Gbuffer의 법선방향의 각도이용 리젝션.
   • NormalThreashold
     • 60도(적정값)
     • 180도(리젝션안함)
4. 짤린(clipped) 데칼
   • 데칼상자범위가 카메라를 뚷어버리면, 뒷면을 그림(깊이 데스트 방향을 뒤집어서)
   • 회피책
     • 엄청 얇게
     • 엄청 두껍게(성능 떨어짐)

// 3. 옆면이 늘어나요.... ㅜ.ㅠ

// 정점 셰이더에서 데칼 상자의 방위를 구함:
output.Orientation = normalize(WorldView[1].xyz);

gNormalThreashold == cos(각도)

// 픽셀 셰이더에서 GBuffer 법선을 읽어와 리젝션 테스트
float3 normal = DecodeGbufferNormal(tex2D(GNormalMap, depth_uv));
clip(dot(normal, orientation) - gNormalThreshold);

fadeout

• http://ttmayrin.tistory.com/37

수직인 지형에서의 경계면이 잘리는 거 fadeout

// 유니티는 Y가 높이이기에
// #define HALF_Y 0.25f
// OutColor *= (1.f - max((positionOS_decal.y - HALF_Y) / HALF_Y, 0.f));
OutColor *= (1.f - max(4 * positionOS_decal.y - 1, 0.f));

컨택트 섀도(Contact Shadow)

• 전형규, 가성비 좋은 렌더링 테크닉 10선, NDC2012

float4 ContactShadowPSMain(PSInput input) : COLOR
{
  input.screenTC.xyz /= input.screenTC.w;

  float depthSample = tex2D(DepthSampler, input.screenTC.xy).x;
  float sceneDepth = GetClipDistance(depthSample);
  float3 scenePos = FromScreenToView(input.screenTC.xy, sceneDepth);

  float shadow = length(scenePos - input.origin) / (input.attributes.x + 0.001f);
  shadow = pow(saturate(1 - shadow), 2);

  const float radiusInMeter = input.attributes.x;
  float aoIntensity = saturate(4.0f - 2.5 * radiusInMeter);

  shadow *= 0.7f * input.attributes.y;

  return float4(shadow, 0.0f, shadow * aoIntensity, 0);
}

Ref

• KGC2012 - 스크린 스페이스 데칼에 대해 자세히 알아보자(워햄머 40,000: 스페이스 마린)
  • KGC 2011
• GDC2016 - Low Complexity, High Fidelity: The Rendering of INSIDE
• GDC2012 - Effects Techniques Used in Uncharted 3: Drake's Deception
• https://github.com/ColinLeung-NiloCat/UnityURPUnlitScreenSpaceDecalShader
  • https://assetstore.unity.com/packages/vfx/shaders/lux-urp-essentials-150355
• https://blog.theknightsofunity.com/make-it-snow-fast-screen-space-snow-shader/
• https://samdriver.xyz/article/decal-render-intro
• http://www.ozone3d.net/tutorials/glsl_texturing_p08.php#part_8
  • https://diehard98.tistory.com/entry/Projective-Texture-Mapping-with-OpenGL-GLSL
• https://blog.csdn.net/puppet_master/article/details/84310361
• https://mynameismjp.wordpress.com/2009/03/10/reconstructing-position-from-depth/

Rain

• 메쉬 생성해서 텍스쳐 입히고 메쉬 자체를 흘려내리기.
• Geometry Shader이용하기
• PostProcessing으로 처리하기(부하크고 각도 변화 힘듬)
• 파티클로 처리하기
  • 컬링 유의
  • https://blog.unity.com/technology/unitytips-particlesystem-performance-culling

Ref

• https://github.com/tsugi/exampleunityangrybots/tree/master/Assets/Scripts/Rain
• SIGGRAPH2011-FastMobileShaders
• Rain And Snow Effect With Geometry Shaders In Unity
  • https://github.com/tiredamage42/RainSnowGeometryShader
  • https://seblagarde.wordpress.com/2012/12/27/water-drop-2a-dynamic-rain-and-its-effects/
• Artist-directable Real-Time Rain Rendering in City Environments
  • SIGGRAPH2006
  • GDC06
• ShaderX7 Advanced Rendering Techniques
  • 5.1 Dynamic Weather Effects
• rain_streak 이미지들
• https://developer.download.nvidia.com/SDK/10/direct3d/samples.html#rain

Sky

Skybox 메쉬 형태

• Cube, Sphere, HemiSphere, Ring, Plane

• 유니티 Shader별 사용 메쉬 형태

• 렌더독으로 본 유니티의 Sphere Sky Mesh

  • 일반 Sphere 와는 다르게, 버텍스 갯수는 적게 그리고 수평선 부분이 조금 디테일에 힘을 줬다.

  

유니티 Skybox셰이더 작성시 주의점

• 유니티 Skybox 설정 : Window > Rendering > Lighting > Environment
  • unity_SpecCube0가 위에서 설정된 메테리얼로 스카이박스를 렌더링함.(Camera > Background Type과는 상관없음)
• URP 환경이라도 Built-in(legacy)의 기본 Pass의 태그값 "LightMode" = "ForwardBase"로 하여야만 동작한다.

유니티의 Skybox

• 유니티에서는 Skybox의 메쉬를 지정할 수 있는 방법이 없다.(2021.09.23 기준)
• 유니티 Skybox 설정을 안따르면 ReflectionProbe(unity_SpecCube0)를 다루기 껄끄러워진다.
  • Window > Rendering > Lighting > Environment > Environment Reflections > Source > Custom 으로 처리 가능.

Skybox 메쉬 조합

• How to Create Skies for 3D Games?

구름을 표현하기 위해 돔형태의 메쉬, 링형 매쉬, 평면 매쉬를 이용했다.

Case Study

half daytimeGradient            = max(0, L_Sun.y);                   // 낮시간 변화 // max(0, dot(-L, DIR_DOWN));
half skybox_MidTopGradient      = max(0, -V.y);                      // 하늘쪽 변화 // max(0, dot(-V, DIR_UP));
half skybox_HorizBottomGradient = pow(1 - skybox_MidTopGradient, 8); // 바닥 + 수평 변화



// 낮시간 하늘의 3단계 변화
// - 카메라가 스카이박스 안쪽에 있으니 `-V`를 시켜주고, 하늘(Up)쪽으로 변화를 넣는다.
// - 수평선은 역으로해서 역변화를 얻음.
half3 daytimeSkyMiddleColor       = lerp(_SkyColor_Sunset, _SkyColor_DaytimeMiddle, daytimeGradient);
half3 daytimeSkyMiddleBottomColor = lerp(daytimeSkyMiddleColor, _SkyColor_DaytimeBottom, skybox_HorizBottomGradient);
half3 daytimeSkyGradientColor     = lerp(daytimeSkyMiddleBottomColor, _SkyColor_DaytimeTop, skybox_MidTopGradient);



// 밤낮을 표현하기 위해 빛이 땅을 바라볼때 변화량([0, 1]) 이용.
half3 skyNightDayColor = lerp(_SkyColor_Night, daytimeSkyGradientColor, daytimeGradient);



// 빛이 바라보는 반대 방향에 해를 위치 시킨다.
half sunGradient = dot(-L_Sun, V);
half sun = pow(sunGradient, 20);


// 노을의 빛의 퍼짐을 표현하기 위해, 노을색과 붉기를 조절한 빛의 색을 섞는다.
half _SunsetRedness = 0.5; // [0, 1]
half invRedness = 1 - _SunsetRedness;
half3 redishLightColor;
redishLightColor.r = IN.colorLight.r;
redishLightColor.g = IN.colorLight.g * invRedness;
redishLightColor.b = IN.colorLight.b * invRedness * 0.5;
half3 sunsetColor = lerp(_SkyColor_Sunset, redishColor, sun);



// 해 위치를 빛 방향과 같게 조정하면
// 케릭터에 조명효과를 다르게 주고 싶어 불가피하게 빛 방향을 바꿔 버리면,
// 의도치 않도록 해 위치가 바뀌어 버릴 수 있다.
_ControlledDaytime // [0, 1]
#define TWO_PI            6.28318530717958647693  // com.unity.render-pipelines.core/ShaderLibrary/Macros.hlsl
#define SHORT_TWO_PI      6.2831853

half rad =  _ControlledDaytime * SHORT_TWO_PI;
half s;
half c;
sincos(rad, s, c);
OUT.L_Sun.x = -c;
OUT.L_Sun.y = s;
OUT.L_Sun.z = 0;

|                       | 0   | 90   | 180 | 270  | 360    |
| --------------------- | --- | ---- | --- | ---- | ------ |
| _ControlTime          | 0   | 0.25 | 0.5 | 0.75 | 1      |
| _ControlTime x TWO_PI | 0   |      | PI  |      | TWO_PI |
| x (-cos)              | -1  | 0    | 1   | 0    | -1     |
| y (sin)               | 0   | 1    | 0   | -1   | 0      |

기타 코드 조각들

// ref: [mapping texture uvs to sphere for skybox](https://gamedev.stackexchange.com/questions/189357/mapping-texture-uvs-to-sphere-for-skybox)
// ref: [Correcting projection of 360° content onto a sphere - distortion at the poles](https://gamedev.stackexchange.com/questions/148167/correcting-projection-of-360-content-onto-a-sphere-distortion-at-the-poles/148178#148178)

uv.x = (PI + atan2(positionWS.x, positionWS.z)) * INV_TWO_PI;
uv.y = uv.y * 0.5 + 0.5

// ref: [ARM - The Ice Cave demo](https://developer.arm.com/documentation/102259/0100/Procedural-skybox)

half3 _SunPosition;
half3 _SunColor;
half _SunDegree;    // [0.0, 1.0], corresponds to a sun of diameter of 5 degrees: cos(5 degrees) = 0.995

half4 SampleSun(in half3 viewDir, in half alpha)
{
    // 원형 해
    half sunContribution = dot(viewDir,_SunPosition);

    half sunDistanceFade = smoothstep(_SunDegree - (0.025 * alpha), 1.0, sunContribution);
    half sunOcclusionFade = clamp(0.9 - alpha, 0.0, 1.0);
    half3 sunColorResult = sunDistanceFade * sunOcclusionFade * _SunColor;
    return half4(sunColorResult, 1.0);
}

// ['Infinite' sky shader for Unity](https://aras-p.info/blog/2019/02/01/Infinite-sky-shader-for-Unity/)
// 유니티는 "reversed-Z projection"을 이용하지만, "infinite projection"은 아니다

#if defined(UNITY_REVERSED_Z)
// when using reversed-Z, make the Z be just a tiny
// bit above 0.0
OUT.positionCS.z = 1.0e-9f;
#else
// when not using reversed-Z, make Z/W be just a tiny
// bit below 1.0
OUT.positionCS.z = o.positionCS.w - 1.0e-6f;
#endif

TODO

• 구름 그림자
  • https://github.com/EntroPi-Games/Unity-Cloud-Shadows/

Ref

• Unity's built-in Skybox-Procedural.shader
• 마른 하늘에 날구름 넣기
• [TA] 테라에 사용된 렌더링 테크닉 - 임신형 (valhashi)
• Procedural Skybox - Evan edwards
• EasySky: Breakdown of a Procedural Skybox for UE4
• GDC2014 - Moving the Heavens: An Artistic and Technical Look at the Skies of The Last of Us
• Reaching for the stars - Let’s create a procedural skybox shader with Unity’s Shader Graph!
• Unity ShaderGraph Procedural Skybox Tutorial
  • cloud texture
  • pt1
  • pt2
• Rastertek's Terrain Tutorial
  • Tutorial 10: Sky Domes : 원문, 번역
  • Tutorial 11: Bitmap Clouds : 원문, 번역
  • Tutorial 12: Perturbed Clouds : 원문, 번역
• Creating a unique animated sky shader for UT3/UDK - Cr4zys_Cloud_Textures.rar
• Maya - Skydome Techniques
• Unity Assets
  • https://assetstore.unity.com/packages/tools/particles-effects/tenkoku-dynamic-sky-34435
  • https://assetstore.unity.com/packages/2d/textures-materials/sky/procedural-sky-builtin-lwrp-urp-jupiter-159992
  • https://assetstore.unity.com/packages/tools/particles-effects/azure-sky-dynamic-skybox-36050
• https://simul.co/
• https://github.com/shadowlenz/Procedural.SkyBox
• https://github.com/SebLague/Clouds
  • https://www.youtube.com/watch?v=4QOcCGI6xOU
• https://guildofwriters.org/wiki/Adding_Atmosphere
• Volumetric Clouds – 体积云的做法
• 천체리소스 : https://pngtree.com/so/celestial
• ShaderX3 Advanced Rendering with DirectX and OpenGL
  • 8.4 Volumetric Clouds
• ShaderX2 Shader Programming Tips and Tricks with DirectX 9
  • Advanced Sky Dome Rendering

_Grass

• https://roystan.net/articles/grass-shader.html

물

• https://ateliersera.blog.me/220413097549

• https://www.ronja-tutorials.com/2018/11/03/river.html

• https://alexanderameye.github.io/simple-water.html

• https://darkcatgame.tistory.com/30?category=806332

• https://github.com/simplestargame/UniversalOceanShaderGraph

• https://www.gamedeveloper.com/programming/water-interaction-model-for-boats-in-video-games-part-2

• https://www.e2gamedev.com/water-buoyancy

• Acerola - Rendering Water With Sine Waves

• Acerola - I Tried Simulating The Entire Ocean

• Ocean waves simulation with Fast Fourier transform

• Very fast procedural ocean

• http://peerplay.nl/ripple-water-shader/

• rain

  • https://tympanus.net/codrops/2015/11/04/rain-water-effect-experiments/

• 위치

  • sin을 이용한 리니어 방식
    • linear wave : y = y - Amplitude * sin(Frequency * x - Phase)
  • noise를 이용한 방식

• 디테일

  • 노말맵
    • 1장 - 단조로움
    • 2장 - 역방향으로 하고 합친다
      • n = n1 + n2 * 0.5 * (0.5, 0.5, 1)
  • 버텍스칼라 가중치

• 주의해야할 점

  • 너무 단순히 만들면 타일링 패턴으로, 부자연스럽게 보임

바다/강

• 특징 짚기
  • 물결
  • 거품
    • sea form
      • 오브젝트 상호작용은 깊이맵 이용
    • sea spray
  • 거리 가까울때 투명
  • 거리 멀때 거울처럼 반사
    • cubemap ? Planar Reflection
  • 수면 아래 Caustic
    • 카메라 앞에 커스틱 매쉬를 둬서 처리

셰이더 참고 단어

Ref

• gpugems 1

  • Chapter 1. Effective Water Simulation from Physical Models
  • Chapter 2. Rendering Water Caustics
  • Chapter 42. Deformers

• gpugems 2

  • Chapter 18. Using Vertex Texture Displacement for Realistic Water Rendering
    • 번역

• Boat Attack

  • https://github.com/Unity-Technologies/BoatAttack
  • 모바일 플랫폼을 위한 URP에서의 물 표현. 모바일 플랫폼을 위한 물 표현 (1/6)
  • 2 Gerstner wave - 카메라를 따라다니게
  • 3 Planar Reflection
  • 4 Caustic - 카메라를 따라다니게
  • 5 Fresnel

• https://en.wikipedia.org/wiki/Trochoidal_wave

• Water Caustics Shader - Unity

  • https://github.com/z4gon/water-caustics-shader-unity

• Deep-Water Animation and Rendering

  • https://www.semanticscholar.org/paper/Deep-Water-Animation-and-Rendering-Jensen-Goli%C3%A1%C5%A1/b82acf6a543aff6a3581b1e7ad02efb88b501750
  • 번역

• https://catlikecoding.com/unity/tutorials/flow/waves/

• [SIGGRAPH2018] The Technical Art of Sea of Thieves

  • https://history.siggraph.org/learning/the-technical-art-of-sea-of-thieves/

float3 GerstnerWave (
      float4 wave,
      float3 p,
      inout float3 tangent,
      inout float3 binormal)
{
  float steepness = wave.z;
  float wavelength = wave.w;
 
  float  k = 2 * UNITY_PI / wavelength;
  float  c = sqrt(9.8 / k);
  float2 d = normalize(wave.xy);
  float  f = k * (dot(d, p.xz) - c * _Time.y);
  float  a = steepness / k;
  
  float sinf;
  float cosf;
  sincos(f, sinf, cosf);

  tangent += float3(
    -d.x * d.x * (steepness * sinf),
    d.x * (steepness * cosf),
    -d.x * d.y * (steepness * sinf)
  );
  binormal += float3(
    -d.x * d.y * (steepness * sinf),
    d.y * (steepness * cosf),
    -d.y * d.y * (steepness * sinf)
  );
  
  return float3(
    d.x * (a * cosf),
    a * sinf,
    d.y * (a * cosf)
  );
}

퓨리에변환 fourier transform https://en.wikipedia.org/wiki/Fourier_transform

웨블레트 (Wavelet) 웨이블릿이란 0을 중심으로 증가와 감소를 반복하는 진폭을 수반한 파동 같은 진동을 말한다

phillips spectrum https://github.com/Scrawk/Phillips-Ocean

얼음

Unity Shader Graph - Ice Tutorial

ScreenColor add Fresnel

Interior Mapping

• 인테리어 맵핑
• youtube: Overwatch: фейковое окно на King's Row (fake window)

잘 모르겠다...

Ref

• AABB(Axis Aligned Bounding Box) 방식의 레이 충돌공식
• coposuke 설명(일본어)
  • https://coposuke.hateblo.jp/entry/2018/12/11/064755
  • https://coposuke.hateblo.jp/entry/2020/12/24/003734
• https://forum.unity.com/threads/interior-mapping.424676/#post-2751518
• https://forums.ogre3d.org/viewtopic.php?t=30275
• gamasutra
  • Game Tech Deep Dive: A window into Playground Games' latest shader development
  • Interior Mapping: rendering real rooms without geometry
• Shader Showcase Saturday #9: Interior Mapping
• From AAA to Indie
  • https://youtu.be/_x88tvkAGuo?t=59
• assetstore
  • Fake Interiors FREE
  • TOZ Interior Shaders
• http://interiormapping.oogst3d.net/
  • Interior mapping with Godot - Pedro J. Estébanez | GodotCon Brussels 2020
• Interior Mapping shader (.unitypackage)
• andrewgotow(1, 2, 3)
• https://docs.unrealengine.com/udk/Three/DevelopmentKitGemsInteriorMapping.html

Vegetation

• Mesh의 중앙에서 vertex사이의 거리 이용.
• 작은것은 r채널만 이용해서 흔들거려도 될듯

Ref

• https://developer.nvidia.com/gpugems/gpugems3/part-iii-rendering/chapter-16-vegetation-procedural-animation-and-shading-crysis
• https://docs.unity3d.com/Packages/com.unity.polybrush@1.0/manual/modes_color.html
  • u: Polybrush Intro and Tutorial
• https://blogs.unity3d.com/kr/2018/06/29/book-of-the-dead-quixel-wind-scene-building-and-content-optimization-tricks/
• https://blogs.unity3d.com/2018/08/07/shader-graph-updates-and-sample-project/

Billboard

All-Axis / Spherical

• 뷰까지 오브젝트 정보(회전/확대)를 들고 가는게 아니라, 뷰공간에서 오브젝트의 정보를 더해 화면을 바라보게 만든다.

float3 positionVS = TransformWorldToView(UNITY_MATRIX_M._m03_m13_m23);
positionVS += float3(IN.positionOS.xy * _Scale, 0);

OUT.positionCS = TransformWViewToHClip(positionVS);

Y-Axis / Cylindrical

// 개념위주, 장황하게
float3 viewDirWS = -GetWorldSpaceViewDir(UNITY_MATRIX_M._m03_m13_m23);

float toViewAngleY = atan2(viewDirWS.x, viewDirWS.z);
float s = sin(toViewAngleY);
float c = cos(toViewAngleY);
float3x3 ROTATE_Y_AXIS_M = {
    c, 0, s,
    0, 1, 0,
    -s, 0, c
};

float3 positionOS = mul(ROTATE_Y_AXIS_M, IN.positionOS.xyz);

// 간소화
float2 viewDirWS = -normalize(
    GetCameraPositionWS().xz - UNITY_MATRIX_M._m03_m23
);

float2x2 ROTATE_Y_AXIS_M = {
    viewDirWS.y, viewDirWS.x,
    -viewDirWS.x, viewDirWS.y
};

float3 positionOS;
positionOS.xz = mul(ROTATE_Y_AXIS_M, IN.positionOS.xz);
positionOS.y = IN.positionOS.y;

Ref

• https://www.sysnet.pe.kr/2/0/11641
• https://en.wikibooks.org/wiki/Cg_Programming/Unity/Billboards
• https://80.lv/articles/the-shader-approach-to-billboarding/
• http://unity3d.ru/distribution/viewtopic.php?f=35&t=24903
• https://forum.unity.com/threads/billboard-shader-using-vertex-offsets-in-color-or-uv2-data.192652/
• https://gam0022.net/blog/2019/07/23/unity-y-axis-billboard-shader/

Dissolve

• dissolve : 녹다, 용해되다.
• Dissolve텍스쳐를 이용하여, 특정 값 이하일때 표시를 안하면 사라지는 효과를 얻을 수 있다.
• Alpha.md 참조.

Sample

half dissolveTex = SAMPLE_TEXTURE2D(_DissolveTex, sampler_DissolveTex, IN.uv).r;
clip(dissolveTex - _Amount);

// https://developer.download.nvidia.com/cg/clip.html

void clip(float4 x)
{
  if (any(x < 0))
    discard;
}

• 유니티 함수 AlphaDiscard를 쓰는 사람도 있는데, 이 경우 _ALPHATEST_ON를 이용하는지 여부에 따라 결과가 달라짐으로 주의.

// com.unity.render-pipelines.universal/ShaderLibrary/ShaderVariablesFunctions.hlsl

void AlphaDiscard(real alpha, real cutoff, real offset = real(0.0))
{
    #ifdef _ALPHATEST_ON
        clip(alpha - cutoff + offset);
    #endif
}

SpriteFlipBook_FlowMap_MotionVector

SpriteFlipBook

// uv         : [0, 1]
// frame      : time * FramePerSeconds
// imageCount : (ColumnCount, RowCount)
half2 GetSubUV(in half2 uv, in half frame, in uint2 imageCount)
{

    // ex)
    // frame      = 9.9
    // imageCount = (8, 8)

    half2 scale = 1.0 / imageCount;           // scale = (0.125, 0.125)

    // floor : 소수점이하 버림.
    // frac  : 소수점이하 반환.
    half index = floor(frame);                 // index = 9
    // half index = frame - frac(frame);
    
    // fmod  : 나머지 반환.
    // offset.x = 9 % 8                                     => 1
    // offset.y = -1 - floor(9 * 0.125) => -1 -floor(1.125) => -2
    half2 offset = half2(
      fmod(index, imageCount.x),
      -1 - floor(index * scale.x)
    );

    // y가 -1로 시작하는 이유.
    //  - uv좌표     : 좌하단
    //  - 이미지시트 : 좌상단
    // 기준점을 uv(0, 0)을 sheet(0, -1)로 변환해야함.

    return (uv + offset) * scale;
}

half frameNumber = _Time.y * _FramesPerSeconds;
half2 subUV = GetSubUV(IN.uv, frameNumber, uint2(_ColumnsX, _RowsY));

FlowMap

• FlowMapPainter
• mnpshino - 모션벡터와 플로우 맵 (2)
• SIGGRAPH2010 presentation Water Flow in Portal 2
• https://catlikecoding.com/unity/tutorials/flow/texture-distortion/
• http://wiki.polycount.com/wiki/Flow_map
• youtube: Flow Map Material Setup in 13 Minutes
• GDC2012 - Water Technology of Uncharted
• Graphics Runner: Animating Water Using Flow Maps by Kyle Hayward
• http://ttmayrin.tistory.com/40

half2 flowTex = SAMPLE_TEXTURE2D(_FlowTex, sampler_FlowTex, IN.uv).rg;

// flowTex[0, 1] => [-1, 1]
half2 flowUV = flowTex * 2.0 - 1.0;

// or flowTex[0, 1] => [-0.5, 0.5]
// half2 flowUV = flowTex - 0.5;

// [-1, 1] 선형 반복.
// frac  : 소수점이하 반환.
half flowLerp = abs((frac(_Time.x * _FlowSpeed) - 0.5) * 2.0);

half2 uv0 = IN.uv + flowUV * frac(_Time.x * _FlowSpeed);
half2 uv1 = IN.uv + flowUV * frac(_Time.x * _FlowSpeed + 0.5);

half3 mainTex0 = SAMPLE_TEXTURE2D(_MainTex, sampler_MainTex, uv0).rgb;
half3 mainTex1 = SAMPLE_TEXTURE2D(_MainTex, sampler_MainTex, uv1).rgb;

half3 finalColor = lerp(mainTex0, mainTex1, flowLerp);

MotionVector

• http://www.klemenlozar.com/frame-blending-with-motion-vectors/
• GDC2012 - The Tricks Up Our Sleeves: A Walkthrough of the Special FX of Uncharted 3: Drake’s Deception

half frameNumber = _Time.y * _FramesPerSeconds;
uint2 imageCount = uint2(_ColumnsX, _RowsY);

OUT.frameNumber = frameNumber;
OUT.subUV0 = GetSubUV(IN.uv, frameNumber, imageCount);
OUT.subUV1 = GetSubUV(IN.uv, frameNumber + 1, imageCount);

// -------------------
// flowTex[0, 1] => [-1, 1]
half2 flowTex0 = SAMPLE_TEXTURE2D(_FlowTex, sampler_FlowTex, IN.subUV0).rg;
half2 flowTex1 = SAMPLE_TEXTURE2D(_FlowTex, sampler_FlowTex, IN.subUV1).rg;
flowTex0 = flowTex0 * 2.0 - 1.0;
flowTex1 = flowTex1 * 2.0 - 1.0;

half interval = frac(IN.frameNumber);

half2 mainUV0 = IN.subUV0 - (flowTex0 * interval * _DistortionStrength);
half2 mainUV1 = IN.subUV1 + (flowTex1 * (1 - interval) * _DistortionStrength);
half4 mainTex0 = SAMPLE_TEXTURE2D(_MainTex, sampler_MainTex, mainUV0);
half4 mainTex1 = SAMPLE_TEXTURE2D(_MainTex, sampler_MainTex, mainUV1);

half4 finalColor = lerp(mainTex0, mainTex1, interval);

Hatching

• TAM / Tonal Art Maps

Ref

• SIGGRAPH2001 - Real-Time Hatching
• https://github.com/OskarSigvardsson/tonal-art-map
• https://blog.naver.com/pch413/10001173790
• http://kylehalladay.com/blog/tutorial/2017/02/21/Pencil-Sketch-Effect.html
  • https://github.com/khalladay/PencilSketchEffect

MatCap

• MatCap(Material Capture)
  • SEM(Spherical Environment Mapping) / Lit Sphere
• 환경을 텍스쳐에 맵핑하고, 뷰스페이스 노말로 색을 얻어온다.
• 케릭터에 사용할때는 Diffuse용/Reflect용 맵캡을 이용하도록 하자

// vert
float4x4 MATRIX_IT_MV = UNITY_MATRIX_IT_MV;
float3 binormalOS = cross(IN.normalOS, IN.tangentOS.xyz) * IN.tangentOS.w * unity_WorldTransformParams.w;
float3x3 TBN_os = float3x3(IN.tangentOS.xyz, binormalOS, IN.normalOS);

OUT.TtoV0 = mul(TBN_os, MATRIX_IT_MV[0].xyz);
OUT.TtoV1 = mul(TBN_os, MATRIX_IT_MV[1].xyz);

// frag
half2 normalVS;
normalVS.x = dot(IN.tan0, normalTS);
normalVS.y = dot(IN.tan1, normalTS);
half2 uv_Matcap = normalVS * 0.5 + 0.5;

half3 matcapTex = SAMPLE_TEXTURE2D(_MatcapTex, sampler_MatcapTex, uv_Matcap).rgb;

// vert
half4x4 MATRIX_IT_MV = UNITY_MATRIX_IT_MV;

half2 normalVS;
normalVS.x = dot(MATRIX_IT_MV[0].xyz, IN.normalOS);
normalVS.y = dot(MATRIX_IT_MV[1].xyz, IN.normalOS);
OUT.uv_Matcap = normalVS * 0.5 + 0.5;

// vert
half3 normalWS = TransformObjectToWorldNormal(IN.normalOS);
half3 normalVS = normalize(TransformWorldToView(normalWS));

// 좀 더 디테일한 버전
half3 normalVS = normalize(mul(UNITY_MATRIX_IT_MV, IN.normal));
half3 positionVS = UnityObjectToViewPos(IN.positionOS);
half3 r = reflect(positionVS, normalVS);
half m = 2.0 * sqrt(r.x * r.x + r.y * r.y + (r.z + 1) * (r.z + 1));
OUT.uv_Matcap = r.xy / m + 0.5;

// 회전
half s;
half c;
sincos(_RotateRadian, s, c);
half2x2 rot = {
	c, -s,
	s, c
};
normalVS = mul(rot, normalVS);

번외

• normalWS를 이용함으로써 View에 대해 변하는게 아닌 고정효과
  • 카메라 방향에 상관없는 Matcap 만들기

OUT.uv_Matcap = normalWS.xy * 0.5 + 0.5;

Ref

• MaCrea(MatCap 생성 도구) http://www.taron.de/
  • 사용법: MaCrea introduction
• https://en.wikipedia.org/wiki/Sphere_mapping
• http://wiki.polycount.com/wiki/Spherical_environment_map
• https://docs.microsoft.com/en-us/windows/win32/direct3d9/spherical-environment-mapping
• [Unite Seoul 2019] 최재영 류재성 - 일곱개의 대죄 : "애니메이션의 감성을 그대로＂와 “개발 최적화"
• https://github.com/nidorx/matcaps : MatCap Textures
• https://www.clicktorelease.com/blog/creating-spherical-environment-mapping-shader/
• 에픽 라이브 | 언리얼 엔진 카툰 렌더링 | PART 1

Outline

• 2-pass
  • Scale 확장
  • Normal 확장
• Rim
• SSO(Screen Space Outline)

2-pass

Tags
{
    "RenderPipeline" = "UniversalRenderPipeline"
}

Pass
{
    Name "Outline"
    Tags
    {
        "LightMode" = "SRPDefaultUnlit"
        "Queue" = "Geometry"
        "RenderType" = "Opaque"
    }
    Cull Front
    
    HLSLPROGRAM
    // outline code
    ENDHLSL
}

Pass
{
    Name "Front"
    Tags
    {
        "LightMode" = "UniversalForward"
        "Queue" = "Geometry"
        "RenderType" = "Opaque"
    }
    Cull Back

    HLSLPROGRAM
    // render code
    ENDHLSL
}

Scale 확장

• 1pass 원래 Model 확대하여 외곽선 색으로 칠한다
• 2pass 원래 Model을 덧그린다

half4 Scale(half4 positionOS, half3 s)
{
    // s : scale
    // m : scale matrix

    half4x4 m;
    m[0][0] = 1.0 + s.x; m[0][1] = 0.0;       m[0][2] = 0.0;       m[0][3] = 0.0;
    m[1][0] = 0.0;       m[1][1] = 1.0 + s.y; m[1][2] = 0.0;       m[1][3] = 0.0;
    m[2][0] = 0.0;       m[2][1] = 0.0;       m[2][2] = 1.0 + s.z; m[2][3] = 0.0;
    m[3][0] = 0.0;       m[3][1] = 0.0;       m[3][2] = 0.0;       m[3][3] = 1.0;
    return mul(m, positionOS);
}

OUT.positionCS = TransformObjectToHClip(Scale(IN.positionOS, _OutlineScale).xyz);

Normal 확장

• https://www.videopoetics.com/tutorials/pixel-perfect-outline-shaders-unity/
• 마둠파 - 유니티 외곽선 셰이더 완벽정리편

OUT.positionCS = TransformObjectToHClip(IN.positionOS);

half4 normalCS = TransformObjectToHClip(IN.normal);

// 아웃라인은 2차원이므로. `normalCS.xy`에 대해서만 계산 및 `normalize`.
// 카메라 거리에 따라 아웃라인의 크기가 변경되는것을 막기위해 `positionCS.w`를 곱해준다.
// _ScreenParams.xy (x/y는 카메라 타겟텍스쳐 넓이/높이)로 나누어서 [-1, +1] 범위로 만듬.
// 길이 2인 범위([-1, +1])와 비율을 맞추기 위해 OutlineWidth에 `*2`를 해준다.

half2 offset = (normalize(normalCS.xy) * normalCS.w) / _ScreenParams.xy * (2 * _OutlineWidth) * OUT.positionCS.w;

// 버텍스 칼라를 곱해주면서 디테일 조정.
// offset *= IN.color.r;
OUT.positionCS.xy += offset;

• 여러 갈래로 흩어진 normal을 부드럽게 하기
  • .fbx -> Model, Normal & Tangent Normals -> Normals:Calculate, Smoothing Angel:180
  • TCP2 : Smoothed Normals Utility

Rim

• 림라이트 효과를 이용

half3 NdotL = normalize(N, L);
half rim = abs(NdotL);
if (rim > _Outline)
{
    rim = 1;
}
else
{
    rim = -1;
}
final.rgb *= rim;

half3 NdotV = normalize(N, V);
half rim = 1 - NdotV;
final.rgb *= pow(rim, 3);

Post Processing 이용 - SSO(Screen Space Outline)

• https://en.wikipedia.org/wiki/Roberts_cross
• https://en.wikipedia.org/wiki/Sobel_operator

외곽선 검출(Edge detection) 필터

Ref

• https://roystan.net/articles/outline-shader.html
• https://alexanderameye.github.io/outlineshader.html
• https://alexanderameye.github.io/notes/rendering-outlines/
• https://musoucrow.github.io/2020/07/05/urp_outline/
• https://github.com/netpyoung/bs.introduction-to-shader-programming/blob/master/note/ch12.md
• https://www.ronja-tutorials.com/2020/07/23/sprite-outlines.html
• https://bgolus.medium.com/the-quest-for-very-wide-outlines-ba82ed442cd9
• https://www.codinblack.com/outline-effect-using-shader-graph-in-unity3d/
• 외곽선 렌더링 구현에 관한 허접한 정리

Parallax Mapping

• parallax mapping, offset mapping, photonic mapping, virtual displace mapping 이라고 부른다.
• 높이 정보를 활용하여 텍스처 좌표를 보정.

• Tangent Space 의 임의의 벡터 x는 (t, b, n)의 값을 갖게된다.
• t는 UV.u와 관련있을 것이고, b는 UV.v와 관련이 있게 된다.

Eye
= -V_TS.tbn / V_TS.n
= -(V_TS.t / V_TS.n,  V_TS.b / V_TS.n, V_TS.n / V_TS.n)
= -(V_TS.t / V_TS.n,  V_TS.b / V_TS.n,               1)

half3x3 TBN = half3x3(normalInputs.tangentWS, normalInputs.bitangentWS, normalInputs.normalWS);

// 시점에 대한 tangent space의 V
OUT.V_TS = mul(TBN, GetCameraPositionWS() - vertexInputs.positionWS);
// 광원 계산을 위한 tangent space의 L
OUT.L_TS = mul(TBN, mainLight.direction);

// OUT.V_TS.t = dot(V, T);
// OUT.V_TS.b = dot(V, B);
// OUT.V_TS.n = dot(V, N);

// OUT.L_TS.t = dot(L, T);
// OUT.L_TS.b = dot(L, B);
// OUT.L_TS.n = dot(L, N);


half2 ParallaxMapping(half2 uv, half3 V_TS)
{
    // 높이 맵에서 높이를 구하고,
    half height = SAMPLE_TEXTURE2D(_NormalDepthPackedTex, sampler_NormalDepthPackedTex, uv).b;

    // 시선에 대한 offset을 구한다.
    // 시선은 반대방향임으로 부호는 마이너스(-) 붙여준다.
    // TS.xyz == TS.tbn
    half2 E = -(V_TS.xy / V_TS.z);

    // 근사값이기에 적절한 strength를 곱해주자.
    return uv + E * (height * _HeightStrength);
}

half3 L_TS = normalize(IN.L_TS);
half3 V_TS = normalize(IN.V_TS);
half2 parallaxUV = ParallaxMapping(IN.uv, V_TS);
if ((parallaxUV.x < 0.0 || 1.0 < parallaxUV.x) || (parallaxUV.y < 0.0 || 1.0 < parallaxUV.y))
{
    // x, y가 범위([0, 1])를 벗어나면 discard.
    discard;
}

// 계산 절약을 위해 `tangent space`기준으로 반사 계산을 한다.
half3 N_TS = UnpackNormal(SAMPLE_TEXTURE2D(_NormalDepthPackedTex, sampler_NormalDepthPackedTex, parallaxUV));
half NdotL = saturate(dot(N_TS, L_TS));

// `world space` 기준이라면 계산이 더 들어가게 된다.
half3 N_TS = UnpackNormal(SAMPLE_TEXTURE2D(_NormalDepthPackedTex, sampler_NormalDepthPackedTex, parallaxUV));
half3 N = mul(normalTex, TBN);
half3 L = mul(L_TS, TBN);
half NdotL = saturate(dot(N, L));

• 유니티에도 헬퍼 함수들이 있지만, 별로 추천안함...
  • https://github.com/Unity-Technologies/Graphics/blob/master/com.unity.render-pipelines.core/ShaderLibrary/ParallaxMapping.hlsl
  • half4짜리 tangentWS를 원하지 않나... viewDirTS를 다시 normalze하질않나..

종류

• Parallax Mapping
• Parallax Mapping with offset limit
• Steep Parallax Mapping
• ReliefParallax
• Parallax Occlusion Mapping (POM)
• ....

Ref

• https://learnopengl.com/Advanced-Lighting/Parallax-Mapping
  • 번역본: https://gyutts.tistory.com/175
• bzyzhang -（三）表面凹凸技术
• http://blog.naver.com/sybershin/129399930

SSS

SSS(Sub-Surface Scattering) (피하산란)

• BSSRDF(Bidirectional surface-scattering reflectance distribution function)
  • 입사한 지점과 반사되는 지점이 다름.
  • 실시간으로 계산하기에는 부하가 큼
• 여러 기법들
  • Texture-Space Diffusion
  • Scren-Space SSS
  • Pre-Integrated Skin Shading
  • other Fake/Fast/Approximated SSS

TSD / Texture-Space Diffusion

SSSSS / Scren-Space Sub-Surface Scattering

• https://www.iryoku.com/sssss/
  • https://www.iryoku.com/screen-space-subsurface-scattering/
• https://unityshader.hatenablog.com/entry/2016/12/25/175333
• https://docs.unrealengine.com/udk/Three/ScreenSpaceSubsurfaceScattering.html
• 2018 - [Siggraph2018] Efficient screen-space subsurface scattering using Burley’s normalized diffusion in real-time - Evgenii Golubev (Unity Technologies)

PISS / Pre-Integrated Skin Shading

• LookUpTexture 이용

• fwidth : abs(ddx(x)) + abs(ddy(x))

• DirectX는 ddx_fine/ddy_fine함수도 있음.

half diffuse = LdotN;
half curvature = saturate(length(fwidth(N)) / length(fwidth(positionWS)) * curvatureScale);

half2 pissUV;
pissUV.x = diffuse;
pissUV.y = curvature;

half3 pissTex = SAMPLE_TEXTURE2D(_PissTex, sampler_PissTex, pissUV).rgb;

LocalThickness

// local thickness
half  localTicknessTex = SAMPLE_TEXTURE2D(_LocalThicknessTex, sampler_LocalThicknessTex, uv).r;
half3 H                = normalize(L + N * _Distortion);
half  VdotH            = pow(saturate(dot(V, -H)), _Power) * _Scale;
half  backLight        = _Attenuation * (VdotH + _Ambient) * localTicknessTex;

other Fake/Fast/Approximated SSS

half  halfLambert = NdotL * 0.5 + 0.5;
half3 fakeSSS     = (1 - halfLambert) * _SSSColor;
half3 color       = halfLambert + fakeSSS;

half  rim     = 1 - NdotL;
// 역광일때만 하려면 VdotL처리
// rim *= VdotL;
half3 fakeSSS = pow(rim, _SSSPower) * _SSSMultiplier * _SSSColor;
half3 color   = lambert * fakeSSS;

half  rim     = 1 - NdotL;
half3 fakeSSS = SAMPLE_TEXTURE2D(_SSS_RampTex, sampler_SSS_RampTex, half2(rim, 0)).rgb;

half2 brdfUV;
brdfUV.x = dot(N, L);
brdfUV.y = dot(N, H);

half3 brdfTex = SAMPLE_TEXTURE2D(_BrdfTex, sampler_BrdfTex, brdfUV * 0.5 + 0.5).rgb;

half LdotN = dot(L, N);
half LdotV = dot(L, V);

half2 rampUV;
rampUV.x = LdotN * 0.3 + 0.5;
rampUV.y = LdotV * 0.8;
half3 rampTex = SAMPLE_TEXTURE2D(_RampTex, sampler_RampTex, rampUV).rgb;

아니면 Albedo맵 / Normal맵 자체에 Blur(rgb에 가중치를 주어서)를 적용한다.

Ref

• Cheap realisticskinshading kor
• SIGGRAPH2010 - Uncharted 2: Character Lighting and Shading
• GDC2011 - Iterating Realistic Human Rendering: Boxers in FIGHT NIGHT CHAMPION
• https://zhuanlan.zhihu.com/p/97892884
• https://blog.naver.com/checkjei/60167971452
• https://therealmjp.github.io/posts/sss-intro/
• https://mgun.tistory.com/1600
• NVIDIA
  • [GPU Gems 1] Chapter 16. Real-Time Approximations to Subsurface Scattering
  • [GPU Gems 3] Chapter 14. Advanced Techniques for Realistic Real-Time Skin Rendering
  • https://github.com/NVIDIAGameWorks/FaceWorks
• 기법별
  • using Half lambert
    • SSS 쉐이더 만들었어요 뿌우
  • Pre-Integrated Skin Shading
    • GPU Pro 2 - Pre-Integrated Skin Shading
  • LocalThickness
    • GPU Pro 2 - Real-Time Approximation of Light Transport in Translucent Homogenous Media
    • GDC 2011 – Approximating Translucency for a Fast, Cheap and Convincing Subsurface Scattering Look
    • Fast Subsurface Scattering in Unity(1, 2)
    • 실시간 SSS 셰이더 구현하는 세가지 방법(1, 2, 3)
    • SSS Shader for Unity
  • Normal blur
    • [ 기고 ] Normal Blur Sub-Surface Scattering
    • NOC2012 - SubSurfaceScattering + UDK Custom Shader 권오찬
  • SIGGRAPH2011 - Penner pre-integrated skin rendering
  • BRDF
    • Brdf기반 사전정의 스킨 셰이더

_ForceFieldShield

FORCE FIELD in Unity - SHADER GRAPH Unity Shader Graph - Shield Effect Tutorial

• https://github.com/Brackeys/Force-Field
• https://github.com/vaxkun/ReinhardtLike-Shield-ShaderForge
• https://github.com/WorldOfZero/UnityVisualizations
• https://github.com/WorldOfZero/2D-Flat-Shape-Shader/blob/master/Assets/FlatSinShader.shader
• https://github.com/yanagiragi/Unity_Shader_Learn
• https://github.com/Toocanzs/Vertical-Billboard/blob/master/Toocanzs/Vertical%20Billboard/VerticalBillboard.cginc
• https://github.com/Xiexe

Cast Shadow> Off
var ScreenDepth
ScreenDepth -= ScreenPosition.a
var edge = smoothstep(0, 1, 1 - ScreenDepth) + fresnel
texture * edge

거기다 마스킹으로 강조 https://docs.unity3d.com/Packages/com.unity.shadergraph@11.0/manual/Sphere-Mask-Node.html

void Unity_SphereMask_float4(float4 Coords, float4 Center, float Radius, float Hardness, out float4 Out)
{
    Out = 1 - saturate((distance(Coords, Center) - Radius) / (1 - Hardness));
}

LightShaft_Mesh

• https://developer.arm.com/documentation/102259/0100/Light-shafts

Cull Off
ZWrite Off
Blend One One

// Project camera position onto cross section
float3 DIR_UP             = float3(0, 1, 0);
float  dotWithYAxis       = dot(cameraPositionOS, DIR_UP);
float3 projOnCrossSection = normalize(cameraPositionOS - (DIR_UP * dotWithYAxis));

// Dot product to fade the geometry at the edge of the cross section
float dotProd           = abs(dot(projOnCrossSection, input.normal));
output.overallIntensity = pow(dotProd, _FadingEdgePower) * _CurrLightShaftIntensity;

Ref

• https://assetstore.unity.com/packages/vfx/shaders/volumetric-light-beam-99888
• https://assetstore.unity.com/packages/tools/particles-effects/as-lightshafts-2-13196

Rim Light

카메라 각도 * 모델 90 | 그림 카메라 각도 * 모델 180 |안그림

• 피격 연출에도 들어간다.
• 응용가능
  • 반전하여 카메라 방향에 따른 라이트 효과로 응용가능.

half rim = 1 - NdotV;
half steppedRim = smoothstep(1.0 - _RimWidth, 1.0, rim);

Ref

• Unite '17 Seoul - 모바일 하드웨어에서 퀄리티있는 유니티 셰이더 구현
• [0806 박민근] 림 라이팅(rim lighting)
• http://cagetu.egloos.com/5883856
• GDC2011 - Cinematic Character Lighting in STAR WARS: THE OLD REPUBLIC

눈쌓기

월드 노말 기준으로

눈 각도

float _SnowSize;
float _SnowHeight;
float3 _SnowDirOS = float4(0, 1, 0, 1);

float3 snowDirWS = TransformObjectToWorldDir(normalize(_SnowDirOS));
float3 N = TransformObjectToWorldNormal(IN.normalOS);
if (dot(N, snowDirWS) >= _SnowSize)
{
    IN.positionOS.xyz += (v.positionOS.xyz * _SnowHeight);
}

• http://blog.naver.com/plasticbag0/221436480475
• KGC2013 - Company of Heroes 2 (COH2) Rendering Technology: The cold facts of recreating the hardest winter conditions of World War 2

Toon

• ceil / Ramp Texture / smoothstep

// ===== 계단 음영 표시 - ver. ceil() ====
// [0, 1]범위를 _ToonStep(int)을 곱해서 [0, _ToonStep]범위로 변경.
// ceil함수를 이용하여 올림(디테일 제거 효과).
// 다시 _ToonStep(int)으로 나눔으로써 [0, 1]범위로 변경.
half toonDiffuse = halfLambert;
toonDiffuse = ceil(toonDiffuse * _ToonStep) / _ToonStep;

// ===== 계단 음영 표시 - ver. Ramp Texture ====
// 아니면 Ramp Texture 이용
half3 toonColor = SAMPLE_TEXTURE2D(_RampTex, sampler_RampTex, half2(halfLambert, 0)).rgb;

// ===== 림라이트 ======================
// smoothstep으로 경계를 부드럽게 혼합.
half rim = 1 - NdotV;
half rimIntensity = smoothstep(0.715, 0.717, rim);

// 아니면, 빛방향으로 림
half rimIntensity = rim * pow(NdotL, 0.1);

// ===== 스펙큘러 ======================
// smoothstep으로 경계를 부드럽게 혼합.
half toonSpecular = smoothstep(0.005, 0.01, blinnphongSpecular);

아웃라인

• https://github.com/DumoeDss/AquaSmoothNormals

• 버텍스 확장

  • 단순 확장
  • 버텍스 칼라이용 세부 조절

• 포스트이펙트

기타 예제

• SSS 텍스쳐

half3 sssColor = mainTex * sssTex;
half3 afterSssColor = lerp(sssColor, mainTex, diffuse);

• maskTex

• vertex's color

원신

• ref: [번역]원신 셰이더 렌더링 복원 해석
• [번역/정리]만화 얼굴 그림자 매핑 생성 렌더링 원리
  • https://github.com/mattdesl/image-sdf
  • https://github.com/xudxud/Unity-SDF-Generator
  • https://github.com/Weesals/UnitySDF
  • https://github.com/zeroruka/GI-Assets
  • https://github.com/JMargevics/Unity-Texture-Packer
  • https://github.com/andydbc/unity-texture-packer
  • https://github.com/rstecca/ColorBands

Diffuse RampColor Glossiness Specular LightMap RampRange MetalMap FaceShadow

Ramp

• layer
  • 낮x5
  • 밤x5

Ref

• https://alexanderameye.github.io/simple-toon.html
• https://roystan.net/articles/toon-shader.html
  • https://github.com/IronWarrior/UnityToonShader
• The Art Direction of Street Fighter V: The Role of Art in Fighting Games
  • https://www.gdcvault.com/play/1024506/Art-Direction-of-Street-Fighter
• [마비노기] 마비노기 카툰렌더링 제작과정 영상 [Mabinogi] Cartoon rendering Production Process Video
• SIGGRAPH2006 - Shading In Valve's Source Engine
  • NPAR07_IllustrativeRenderingInTeamFortress2
  • Team Fortress 2 Shader for RenderMan RSL
• [GDC2024] - 3D Toon Rendering in 'Hi-Fi RUSH'

Toon Fire Write-up https://www.youtube.com/watch?v=x2AlfAON5F8

https://github.com/xraxra/IMP

Deferred

• Unity URP+Micro Gbuffer自定义延迟管线

Ref

• SIGGRAPH2018 - Maximizing Rendering Efficiency
• SIGGRAPH2020 - Deferred Shading in Unity URP
  • youtube, pdf
• Deferred rendering case study by ozlael
• [Kgc2012] deferred forward 이창희
• 2012 - [KGC 2012] 디퍼드 랜더링 케이스 스터디 : 레이더즈 & 건즈2

L, V 방향

빛(L)과 카메라(V)의 방향 부호는 취향에 따라 Vertex를 향하느냐, Vertex에서 나가느냐에 따라 결정된다.

보통 Vertex를 기준으로 나가는 방향을 많이 따른다.

• https://en.wikipedia.org/wiki/Phong_reflection_model

From Vertex (Unity)

// com.unity.render-pipelines.universal/ShaderLibrary/Lighting.hlsl
light.direction = half3(_MainLightPosition.xyz);

// com.unity.render-pipelines.universal/ShaderLibrary/ShaderVariablesFunctions.hlsl
float3 GetCurrentViewPosition()
{
    return GetCameraPositionWS();
}

float3 GetCameraPositionWS()
{
    return _WorldSpaceCameraPos;
}

float3 GetWorldSpaceViewDir(float3 positionWS)
{
    if (IsPerspectiveProjection())
    {
        // Perspective
        return GetCurrentViewPosition() - positionWS;
    }
    else
    {
        // Orthographic
        return -GetViewForwardDir();
    }
}

L = normalize(lightPositionWS - positionWS);
V = normalize(cameraPositionWS - positionWS);
H = normalize(V + L);
R = reflect(-L, N);

NdotL = dot(N, L);
RdotV = dot(R, V);

To Vertex

Pope 책에서는 Vertex를 향하도록 코드가 작성이 되어있다.

L = normalize(positionWS - lightPositionWS);
V = normalize(positionWS - cameraPositionWS);
H = normalize((-V) + (-L));
R = reflect(L, N);

NdotL = dot(N, -L);
RdotV = dot(R, -V);

코딩 스타일

// 괄호: BSD스타일 - 새행에서 괄호를 열자.
Shader "example/03_texture_uv"
{
    Properties
    {
        // Texture변수는 뒤에 Tex를 붙이자.
        _MainTex("_MainTex",     2D) = "white" {}
        _HeightTex("_HeightTex", 2D) = "gray" {}
    }

    SubShader
    {
        Tags
        {
            "RenderPipeline" = "UniversalRenderPipeline"
            "Queue" = "Geometry"
            "RenderType" = "Opaque"
        }

        Pass
        {
            // Name: 내부적으로 대문자로 처리되니, 처음부터 대문자로 쓰자.
            Name "HELLO_WORLD"
            
            Tags
            {
                "LightMode" = "UniversalForward"
            }

            HLSLPROGRAM
            // pragma
            // include
            // 변수선언
            // CBUFFER 선언
            // 구조체선언
            // 함수선언(vert / frag)

            #pragma prefer_hlslcc gles // gles is not using HLSLcc by default
            #pragma exclude_renderers d3d11_9x // DirectX 11 9.x는 더 이상 지원되지 않으므로 제외.

            #pragma target 3.5
            #pragma vertex vert
            #pragma fragment frag

            #include "Packages/com.unity.render-pipelines.universal/ShaderLibrary/Core.hlsl"

            // Texture와 sampler는 동일한 라인에 선언해주고, 중간엔 Tab으로 맞추어주자.
            TEXTURE2D(_MainTex);        SAMPLER(sampler_MainTex);

            CBUFFER_START(UnityPerMaterial)
                float4 _MainTex_ST;
            CBUFFER_END

            // Semantics는 특별히 Tab으로 정렬해주자.
            struct APPtoVS
            {
                float4 positionOS    : POSITION;
                float2 uv            : TEXCOORD0;
            };

            struct VStoFS
            {
                float4 positionCS    : SV_POSITION;
                float2 uv            : TEXCOORD0;
            };

            // vert/frag함수에서 입력/출력에는 IN/OUT을 쓴다.
            VStoFS vert(in APPtoVS IN)
            {
                VStoFS OUT;
                ZERO_INITIALIZE(VStoFS, OUT);

                OUT.positionHCS = TransformObjectToHClip(IN.positionOS.xyz);
                OUT.uv = TRANSFORM_TEX(IN.uv, _MainTex);

                // Time : https://docs.unity3d.com/Manual/SL-UnityShaderVariables.html
                OUT.uv += frac(float2(0, 1) * _Time.x);

                return OUT;
            }

            half4 frag(in VStoFS IN) : SV_Target
            {
                // if / for등 괄호({, })를 빼먹지 말자.
                if (...)
                {
                    ...
                }
                return SAMPLE_TEXTURE2D(_MainTex, sampler_MainTex, IN.uv);
            }
            ENDHLSL
        }
    }
}

// mainTex - _MainTex 이름 맞추기.
half3 mainTex = SAMPLE_TEXTURE2D(_MainTex, sampler_MainTex, IN.uv).rgb;
half3 normalTex = UnpackNormal(SAMPLE_TEXTURE2D(_NormalTex, sampler_NormalTex, IN.uv));

// 월드스페이스 방향. 대문자로.
Light light = GetMainLight();
half3 T = normalize(IN.B);
half3 N = CombineTBN(normalTex, IN.T, IN.B, IN.N);
half3 L = normalize(light.direction);
half3 V = TransformWorldToViewDir(IN.positionWS);
half3 H = normalize(L + V);

// dot연산 변수는 NdotL과 같은 형식으로
half NdotL = max(0.0, dot(N, L));
half TdotL = dot(T, L);
half TdotV = dot(T, V);

// 나머지 함수 연산은 sinTL 이런식으로.
half sinTL = sqrt(1 - TdotL * TdotL);

생각해 볼 것

• L, V
• 일단 흔히 사용되는 방식을 따르고, 좀 더 확신이 들면 바꾸던가 하자

NdotL or LdotN

• 어차피 동일한 값이지만 어떤 네이밍을 고를것인가

L = normalize(light.direction) or L = normalize(-light.direction)

• 유니티의 light.direction은 오브젝에서 라이트로 향하는 방향(normalize되지 않은)

  • light.direction = _MainLightPosition.xyz;

• postfix붙여볼까?

  • L_from , L_to

V?

• 눈(eye)을 뜻하는 E를 쓰는 사람도 있지만... E보다는 V고.. 방향이 문제인데..
• V = GetWorldSpaceNormalizeViewDir(positionWS);
  • 오브젝트에서 뷰로 향하는 방향임

float3 GetViewForwardDir()
{
    float4x4 viewMat = GetWorldToViewMatrix();
    return -viewMat[2].xyz;
}

float3 GetWorldSpaceNormalizeViewDir(float3 positionWS)
{
    if (IsPerspectiveProjection())
    {
        // Perspective
        float3 V = GetCurrentViewPosition() - positionWS;
        return normalize(V);
    }
    else
    {
        // Orthographic
        return -GetViewForwardDir();
    }
}

float3 GetPrimaryCameraPosition()
{
#if (SHADEROPTIONS_CAMERA_RELATIVE_RENDERING != 0)
    return float3(0, 0, 0);
#else
    return _WorldSpaceCameraPos;
#endif
}

// Could be e.g. the position of a primary camera or a shadow-casting light.
float3 GetCurrentViewPosition()
{
#if defined(SHADERPASS) && (SHADERPASS != SHADERPASS_SHADOWS)
    return GetPrimaryCameraPosition();
#else
    // This is a generic solution.
    // However, for the primary camera, using '_WorldSpaceCameraPos' is better for cache locality,
    // and in case we enable camera-relative rendering, we can statically set the position is 0.
    return UNITY_MATRIX_I_V._14_24_34;
#endif
}

Trouble Shooting

• zfighiting
  • https://blog.naver.com/shol9570/222407549035
  • 모바일 소수점 계산이 원할하지 않음
  • Near와 Far의 격차를 줄임

아트팀

• 노말이 뒤집혔다
  • normal 표시기를 만들어 둬서 확인

Command Buffer

• https://docs.microsoft.com/en-us/windows/win32/direct3d11/overviews-direct3d-11-render-multi-thread-render
• https://docs.microsoft.com/en-us/windows/win32/direct3d11/overviews-direct3d-11-render-multi-thread-command-list

ChromaSubsampling

• 사람의 시각은 밝기의 변화에 비하여 색상의 변화에는 둔감.

• https://en.wikipedia.org/wiki/Chroma_subsampling

• 원본이미지를 Y'CbCr로 바뀌어 하나의 채널만을 이용하여 저장. (이미지 사이즈가 늘어나기에 POT이미지인 경우 NPOT로 바뀌게 됨)
• 메모리를 아껴주긴하니, 모션블러나같이 RenderTexture사용시 메모리 아낄때 이용하면 좋을듯.

ChromaPack

• https://github.com/keijiro/ChromaPack
  • 4:2:0 Y'CbCr를 이용한다.
  • 알파있는 것은 Y'의 8비트중 1비트를 이용하여 처리
  • 변형하여 YCgCo를 이용한 버전 : https://github.com/n-yoda/unity-ycca-subsampling
  • 이미지 압축시 품질 손실을 막기위해 고안
    • 예로 유니티 내장 PVRTC 변환툴로 변환하면 텍스쳐 압축시 품질저하가 일어남(일러스트같은경우 문제가됨)
    • 품질저하를 줄인 상용 이미지 편집기가 있기도 함(OPTPiX iMageStudio)
    • ASTC가 나온이상, 이게 필요할까? 라는 의문점이 있음

Ref

• https://blog.uwa4d.com/archives/TechSharing_186.html
• https://techblog.kayac.com/texture-compression-in-yuv
  • https://github.com/hiryma/UnitySamples/tree/master/Yuv
• 2020 - Dev Weeks: 성능 프로파일링과 최적화

Optimize Combine Texture

• 여러장의 텍스쳐를 쓰는 대신, 안쓰는 채널이 없도록 packing한다면, 추가적인 텍스쳐를 불러오는 로딩시간과 메모리를 줄일 수 있다.

• 예)

  • Albedo + Specular
  • NormalMap + HeightMap
  • ...

• EasyChannelPacking

ex) NormalMap와 ParallaxMapping

• NormalMap에선 R,G채널 2개의 채널을
• ParallaxMapping에선 깊이에 대한 하나의 채널을 추가해서 사용한다

NormaMap Packing시 주의

BC5 (x, y, 0, 1)을 보면, RGBA채널중에서 RG채널만 사용하고 있다. 이를 이용하여, BA채널에 마스킹이나 다른 데이터값을 체워 넣을 수 있다.

B채널은 공짜로 사용할 수 있으나, A채널까지 사용하려면 유니티의 UnpackNormal함수는 다음과 같이 채널을 바꾸는 기능이 있어, 따로 함수를 작성해 주어야 한다.

// Unpack normal as DXT5nm (1, y, 0, x) or BC5 (x, y, 0, 1)
real3 UnpackNormalmapRGorAG(real4 packedNormal, real scale = 1.0)
{
    // Convert to (?, y, 0, x)
    // R과 A를 혼합하는 과정이 있는데, 이미 텍스쳐의 포맷을 알고 있으면 이 과정은 불필요하다.
    packedNormal.a *= packedNormal.r;
    return UnpackNormalAG(packedNormal, scale);
}

ex

Optimize tip

from Optimizing unity games (Google IO 2014)

• Shader.SetGlobalVector
• OnWillRenderObject(오브젝트가 보일때만), propertyID(string보다 빠름)

void OnWillRenderObject()
{
    material.SetMatrix(propertyID, matrix);
}

Tangent Space 라이트 계산

• 월드 스페이스에서 라이트 계산값과 탄젠트 스페이스에서 라이트 계산값과 동일.
• vertex함수에서 tangent space V, L을 구하고 fragment함수에 넘겨줌.
  • 월드 스페이스로 변환 후 계산하는 작업을 단축 할 수 있음

데미지폰트

• 셰이더로 한꺼번에 출력
• https://blog.naver.com/jinwish/221577786406

NPOT 지원안하는 텍스쳐 포맷

• NPOT지원안하는 ETC/PVRTC같은경우 POT로 자르고 셰이더로 붙여주는걸 작성해서 최적화
  • https://blog.naver.com/jinwish/221576705990

GGX 공식 간략화

• Optimizing PBR for Mobile
  • pdf, note

밉맵 디테일 올리기

샤픈

• https://blog.popekim.com/ko/2013/06/24/mipmap-quality.html
• https://zhuanlan.zhihu.com/p/413834301

FAST SRGB

• pow에 계산되는 비용을 줄이기 위한 방법
• ShaderFeatures.UseFastSRGBLinearConversion
• sRGB Approximations for HLSL

// com.unity.postprocessing/PostProcessing/Shaders/Colors.hlsl
half3 SRGBToLinear(half3 c)
{
#if USE_VERY_FAST_SRGB
    return c * c;
#elif USE_FAST_SRGB
    return c * (c * (c * 0.305306011 + 0.682171111) + 0.012522878);
#else
    half3 linearRGBLo = c / 12.92;
    half3 linearRGBHi = PositivePow((c + 0.055) / 1.055, half3(2.4, 2.4, 2.4));
    half3 linearRGB = (c <= 0.04045) ? linearRGBLo : linearRGBHi;
    return linearRGB;
#endif
}

half3 LinearToSRGB(half3 c)
{
#if USE_VERY_FAST_SRGB
    return sqrt(c);
#elif USE_FAST_SRGB
    return max(1.055 * PositivePow(c, 0.416666667) - 0.055, 0.0);
#else
    half3 sRGBLo = c * 12.92;
    half3 sRGBHi = (PositivePow(c, half3(1.0 / 2.4, 1.0 / 2.4, 1.0 / 2.4)) * 1.055) - 0.055;
    half3 sRGB = (c <= 0.0031308) ? sRGBLo : sRGBHi;
    return sRGB;
#endif
}

// com.unity.postprocessing/PostProcessing/Shaders/StdLib.hlsl
float3 PositivePow(float3 base, float3 power)
{
    return pow(max(abs(base), float3(FLT_EPSILON, FLT_EPSILON, FLT_EPSILON)), power);
}

Anaylize

• https://developer.android.com/agi/frame-trace/shader-performance
• ALU(arithmetic logic unit)
• Unity Shader Performance: How to Quickly Measure the GPU Cycles Your Shaders Take
  • malioc
    • https://github.com/Delt06/malioc-unity
    • https://github.com/Alunice/TaTa/tree/master/com.shac.shaderanalysis
    • https://github.com/Unity-Technologies/sol-games-unity-samples/tree/main/Graphics/MaliCompilerReport
    • https://github.com/wotakuro/UnityMaliocPlugin
    • https://blog.unity.com/games/tackling-profiling-for-mobile-games-with-unity-and-arm
• https://shader-playground.timjones.io/
  • https://forum.unity.com/threads/2023-3-heatmap-color-mode.1507490/#post-9425870

Etc

• https://www.slideshare.net/LeeJungpyo/unite17-shanghailee-jungpyoneteasepangufull-chinese

PostProcessUberShader

• 여러패스가 아닌 하나의 패스로 관리
  • 여러번 전채화면이 Blit되는 것을 방지
  • 공통되게 사용되는 정보(예: 밝기)등 이용

#pragma multi_compile_local_fragment _ _KEY_A

material.EnableKeyword("_KEY_A");
material.DisableKeyword("_KEY_A");

Ref

• https://github.com/Unity-Technologies/Graphics/blob/master/com.unity.render-pipelines.universal/Shaders/PostProcessing/UberPost.shader

Specular Pow Approximation

제곱하는 횟수 n을 줄여보자면(m만큼)?

Specular 구할때 보통 pow쓰는데 이때 제곱되어 지는 횟수를 줄여봐보자

이 그래프에서 예를들어 n=4와 n=16을 비교할때, n=4의 그래프의 기울기와 시작 위치를 얼추 조절하면 오차야 있겠지만, n=16그래프와 비슷해 질 것이다.

그럼 기울기와 오프셋을 구하는 것은 위 문서에 나와있다.

이걸 활용해보자.

예를들어, pow(x, 18)을 구하고자 한다면,

x^n 
max(0, Ax + B)^m // 단,  x < pow(256, -1/n )은 0으로 취급

테이블에서 다음 라인을 확인 할 수 있을 것이다.

공식에 대입하면 다음과 같은 식을 얻을 수 있다.

inline half n18Approximation(half x)
{
    // n | 18
    // m | 2
    //     pow(x, n)
    //     pow(x, 18)
    //     pow(max(0, Ax        + B     ), m)
    return pow(max(0, 6.645 * x + -5.645), 2);
}

half3 specularColor = pow(max(0, dot(R, V)), 18);

half3 specularColor = n18Approximation(max(0, dot(R, V)));

Ref

• A Non-Integer Power Function on the Pixel Shader - by Juan Guardado, Phillippe Beaudoin
• Fast Specular 계산

VertexMultipleLight

• God of war 3에서 여러 광원처리 기법
  • pixel shader가 느렸던 기기여서 vertex에서 처리

// ref: https://github.com/Unity-Technologies/Graphics/blob/master/com.unity.render-pipelines.universal/ShaderLibrary/Lighting.hlsl

half3 VertexLighting(float3 positionWS, half3 normalWS)
{
    half3 vertexLightColor = half3(0.0, 0.0, 0.0);

#ifdef _ADDITIONAL_LIGHTS_VERTEX
    uint lightsCount = GetAdditionalLightsCount();
    LIGHT_LOOP_BEGIN(lightsCount)
        Light light = GetAdditionalLight(lightIndex, positionWS);
        half3 lightColor = light.color * light.distanceAttenuation;
        vertexLightColor += LightingLambert(lightColor, light.direction, normalWS);
    LIGHT_LOOP_END
#endif

    return vertexLightColor;
}

struct VStoFS
{
    half4 positionCS          : SV_POSITION;
    half2 uv                  : TEXCOORD0;
    half3 N                   : TEXCOORD1;
    half3 H_Sun               : TEXCOORD2;
    half3 H_Points            : TEXCOORD3;
    half3 Diffuse_Sun         : TEXCOORD4;
    half3 Diffuse_Points      : TEXCOORD5;
};

VStoFS vert(in APPtoVS IN)
{
    half3 L_points = half3(0, 0, 0);

    uint additionalLightsCount = min(GetAdditionalLightsCount(), 3);
    for (uint i = 0; i < additionalLightsCount; ++i)
    {
        Light additionalLight = GetAdditionalLight(i, positionWS);
        half3 L_attenuated = additionalLight.direction * additionalLight.distanceAttenuation;

        OUT.Diffuse_Points += saturate(dot(N, L_attenuated)) * additionalLight.color;
        L_points += L_attenuated;
    }
    OUT.H_Points = normalize(L_points) + V;

    OUT.Diffuse_Sun = saturate(dot(N, L * mainLight.distanceAttenuation)) * mainLight.color;
    OUT.H_Sun = normalize(L + V);
}

half4 frag(in VStoFS IN) : SV_Target
{
{
    half3 diffuse = diffuseTex * (IN.Diffuse_Sun + IN.Diffuse_Points);

    half2 highlights;
    highlights.x = pow(saturate(dot(N, H_Sun)), _SpecularPower);
    highlights.y = pow(saturate(dot(N, H_Points)), _SpecularPower);
    half3 specular = specularMaskTex * ((IN.Diffuse_Sun * highlights.x) + (IN.Diffuse_Points * highlights.y));

    half3 result = diffuse + specular;
}

Ref

• SIGGRAPH2011 - Dynamic lighting in God of War 3
  • Comments on Advances in Real Time Rendering 2011

AnimationTexture

Ref

• https://github.com/sugi-cho/Animation-Texture-Baker

DrawCall

• 유니티 초보자도 이해하는 URP와 성능, 그리고 모바일까지🔥🔥

• DrawCall : CPU가 GPU에게 그려라(Draw)라고 명령하는 것

  • CPU성능에 영향

• Dev Weeks: 성능을 고려한 파이프라인, Universal Render Pipeline

• [유니티 TIPS] 유니티 최적화를 위한 필수 기본기! Batching 방법 소개

• Batching
  • 하나로 묶음
  • 장점: 드로우 콜 감소
  • 단점: 메모리를 더 필요함.

SRP Batcher

• https://docs.unity3d.com/Manual/SRPBatcher.html
• Material의 정보와 Object의 정보를 GPU 메모리에 상주

// 메터리얼별로 상수버퍼(Constant buffer)를 만들어 주어
// 동일한 셰이더를 여러 머터리얼이 사용하는 경우 앞서 만든 상수버퍼를 이용하셔 성능 향상 도모.
// (상수버퍼안 개별 값을 전달하는게 아니라 상수버퍼 자체를 전달)
// 메모리 정렬(16바이트)가 되어 있는게 좋은데 유니티에서는 어떻게 내부적으로 되었는지 모르겠다.
CBUFFER_START(UnityPerMaterial)
float _Blabla;
CBUFFER_END

CBUFFER_START(UnityPerDraw)
    float4x4 unity_ObjectToWorld;
    float4x4 unity_WorldToObject;
    real4 unity_WorldTransformParams;
CBUFFER_END

public CustomRenderPipeline ()
{
    GraphicsSettings.useScriptableRenderPipelineBatching = true;
}

GPU Instancing

• https://docs.unity3d.com/Manual/GPUInstancing.html
• 동일한 메시끼리 한 번의 드로우콜로 처리
• 배칭과 달리 동일한 메시의 복사본들을 만든다는 점에서 구분. 배칭보다 런타임 오버헤드가 적다.
• CPU에서 처리해서 보내준 정보를 GPU에서 별도의 버퍼에 담고 인스턴싱 처리를 함.
• Renderer
  • Mesh Renderer에만 사용
  • Skinned Mesh Renderer에는 사용 불가(케릭터 불가)

#pragma multi_compile_instancing

// ref: https://github.com/Unity-Technologies/Graphics/blob/master/com.unity.render-pipelines.core/ShaderLibrary/UnityInstancing.hlsl
#include "Packages/com.unity.render-pipelines.core/ShaderLibrary/UnityInstancing.hlsl"

UNITY_INSTANCING_BUFFER_START(UnityPerMaterial)
    UNITY_DEFINE_INSTANCED_PROP(float4, _BaseColor)
UNITY_INSTANCING_BUFFER_END(UnityPerMaterial)

struct APPtoVS
{
    UNITY_VERTEX_INPUT_INSTANCE_ID
};

struct VStoFS
{
    UNITY_VERTEX_INPUT_INSTANCE_ID 
};

VStoFS vert(in APPtoVS IN)
{
    UNITY_SETUP_INSTANCE_ID(IN);

    VStoFS OUT;
    ZERO_INITIALIZE(VStoFS, OUT);
    UNITY_TRANSFER_INSTANCE_ID(IN, OUT);
    return OUT;
}

half4 frag(in VStoFS IN)
{
    UNITY_SETUP_INSTANCE_ID(IN);
    half4 color = UNITY_ACCESS_INSTANCED_PROP(UnityPerMaterial, _Color);
    return 1;
}

static readonly int _BaseColor = Shader.PropertyToID("_BaseColor");

Mesh _mesh;
Material _material;
MaterialPropertyBlock _block;

Matrix4x4[] _matrices;
Vector4[] _baseColors ;

_block = new MaterialPropertyBlock();
_block.SetVectorArray(baseColorId, baseColors);

// ref: https://docs.unity3d.com/ScriptReference/Graphics.DrawMeshInstanced.html
Graphics.DrawMeshInstanced(_mesh, 0, _material, _matrices, _matrices.Length, block);

Dynamic Batching

• 동일한 머티리얼을 공유하는 여러 개의 작은 메시를 하나의 더 큰 메시로 결합하여 그림
  • 공유하지 않은 개별 머티리얼에는 동작 안함
  • 작은 메시에만 적합
• 우선순위
  • SRP Batcher가 Dynamic Batching보다 우선됨.

Tool

• RenderDoc

• NVIDIA Texture Tools Exporter

• 텍스쳐 생성 : https://github.com/mtwoodard/TextureGenerator

• FlowMap Painter

• EasyChannelPacking

• https://docs.unity3d.com/Packages/com.unity.progrids@3.0/manual/index.html

• https://docs.unity3d.com/Packages/com.unity.polybrush@1.0/manual/index.html

• https://docs.unity3d.com/Packages/com.unity.formats.fbx@4.0/manual/index.html

• https://catlikecoding.com/sdf-toolkit/docs/texture-generator/

• https://github.com/andydbc/unity-texture-packer

• https://github.com/unity3d-jp/MeshSync

• https://github.com/unity3d-jp/NormalPainter

• https://github.com/M-Fatah/texture_maker

• http://mebiusbox.github.io/contents/EffectTextureMaker/

• BRDF explorer

  • https://oduska.com/datahoarder/Walt%20Disney%20Animation%20Studios/BRDF%20Explorer/Website/BRDF%20Explorer.html
  • https://github.com/wdas/brdf/downloads

• MERL BRDF Database - https://www.merl.com/brdf/

Texture maker

• https://github.com/njbrown/texturelab
  • Free, Cross-Platform, GPU-Accelerated Procedural Texture Generator
• https://github.com/alelievr/Mixture
  • on unity
• https://www.materialmaker.org/
• 유료
  • substance designer

모델

• https://www.mixamo.com/
• https://www.models-resource.com/
• https://en.wikipedia.org/wiki/List_of_common_3D_test_models

Etc

• FumeFX

Presentation

• 년도별 작품 위주 셰이더관련 발표자료 모음.
• 일단 모바일 URP에서 쓰일법한 기술 위주로 담자.

국내

• 2011 - [TA] 2011. 03. 테라에 사용된 렌더링 테크닉 - 임신형 (valhashi)
• 2012 - [kgc2012] multi plaform Full3D MMO 만들기 "삼국지를 품다"의 테크니컬 아트
• 2012 - [NDC2012] 전형규, 가성비 좋은 렌더링 테크닉 10선
  • 업스케일: 세로 해상도를 더 확보하는게 인지적 측면 유리
  • Radiosity normal map
• 2013 - [NDC2013] 김동석:UDK로 물리기반 셰이더 만들기
• 2013 - [NDC2013] 구세대 엔진 신데렐라 만들기 최종본
• 2014 - [KGC2014] 울프나이츠 엔진 프로그래밍 기록
• 2014 - [유나이트코리아] 영웅의 군단의 테크니컬 아트 - 황재철
• 2014 - [NDC2014] 뉴비탈출 넘버원: 그래픽스 프로그래밍 길들이기
• 2015 - [NDC2015] 광개토태왕 테크니컬 아트
• 2016 - [NDC2016] 프로젝트 A1의 AAA급 캐릭터 렌더링 기술
• 2016 - [IGC 2016] 넷게임즈 김영희 - Unreal4를 사용해 모바일 프로젝트 제작하기
• 2017 - [IGC2017] 펄어비스 개발자가 직접 알려주는 '검은사막의 압도적인 비주얼 노하우'
• 2019 - [NDC2019] 모바일에서 사용 가능한 유니티 커스텀 섭스턴스 PBR 셰이더 만들기
• 2019 - [NDC2019]〈드래곤하운드〉 비주얼이펙트 연출
  • 연기/나무에 불 붙이기
• 2019 - [Unite Seoul 2019] 최재영 류재성 - 일곱개의 대죄 : “애니메이션의 감성을 그대로＂와 “개발 최적화
  • https://youtu.be/8iKWEYvozws?t=800
  • MatCap / Imposter / LookAt
  • ChromaticAberration / Bloom / Noise / RadialBlur / Vignette / InvertColor / ScratchedFilm / ColorGradingLUT
• 2020 - Dev Weeks: A3 Still Alive - Technical Art Review

디퍼드

• 2012 - [KGC 2012] 디퍼드 랜더링 케이스 스터디 : 레이더즈 & 건즈2

일본

• 2018 - 【Unite Tokyo 2018】『崩壊3rd』開発者が語るアニメ風レンダリングの極意
  • https://www.youtube.com/watch?v=egHSE0dpWRw
• 2021 - [SIGGRAPH Asia 2021] 劇場クオリティのアニメをリアルタイムエンジンで実現するには - Unityステーション

중국

• 2022 - 유나이트 서울 2020 - 원신 콘솔 플랫폼 개발 경험 및 렌더링 파이프라인 기술 소개 Track3-1
  • 2022 - Unite Seoul 2020 세션 다시보기 - 원신 콘솔 플랫폼 개발 및 렌더링 파이프라인 기술 소개

해외

• 2013 - [GDC2013] Technical Artist Bootcamp: The VFX of Diablo

• 2016 - [Unite2016] - A Guide for Achieving World-Class Visuals on Mobile Devices

• 2016 - [GDC2016] Delivering Console Car Visuals on Mobile With 'CSR Racing 2'

• 2016 - [GDC2016] The Environment Pipeline of 'CSR Racing 2'

  • https://archive.org/details/GDC2016Harber/

• RIME

  • ADDON2017 - GAT #65: Stylized VFX in RIME
  • UNREALFEST2018 - GAT #67: Stylized VFX in RIME 2.0 - Water Edition

Reference

Books

• 2004 - Directx 9 셰이더 프로그래밍 - 타카시 이마기레
  • DirectX기준
  • 저자 블로그 : https://t-pot.com/program/index.html
  • 부록소스
• 2012 - 셰이더 프로그래밍 입문 - Pope Kim
  • DirectX기준, rendermonkey
  • log
• 2016 - Unity 5.x Shaders and Effects Cookbook - Alan Zucconi , Kenneth Lammers
  • Built-in(Legacy)
• 2017 - 유니티 쉐이더 스타트업 - 정종필
  • Built-in(Legacy) / Surface
  • log
• 2017 - Physically Based Shader Development for Unity 2017 - Zignaigo, Claudia
  • Built-in(Legacy)
  • log
• Become a Unity Shaders Guru
  • https://github.com/PacktPublishing/Become-a-Unity-Shaders-Guru
• Building Quality Shaders for Unity
  • 12장
  • 14장
    • GraphicsBuffer
  • https://github.com/Apress/building-quality-shaders-unity
• Unity Shaders Bible
  • built-in(Legacy), 기초
  • compute, ray tracing and sphere tracing

Video

• ShaderDev - Chayan Vinayak Goswami
  • Built-in(Legacy)
  • log
• Interactive 3D Graphics by Autodesk

Tutorial

• https://github.com/netpyoung/unity.shader.sandbox
  • setchi - ビルトイン関数の使い方いろいろ！シェーダアートの表現力を高める小技集
    • 번역: 빌트인 함수 사용법 이것저것! 셰이더 아트의 표현력을 높이는 잔기술
  • setchi - 楽しい！Unityシェーダーお絵描き入門！
    • 번역: 신나는! Unity 셰이더 그림 그리기 입문!

돈아까움

• the GAME GRAPHICS: 유니티 비주얼 테크닉
• 유니티짱 툰쉐이더 2.0 슈퍼테크닉

EtcLink

그래픽 링크 모음집

• https://kesen.realtimerendering.com/
• http://trowley.org/
• https://cgworld.jp/
• https://discourse.techart.online/
• https://archive.org/details/gdcextras
• https://indievisuallab.github.io/
• 1mafx
• Shader Tutorial Series

리서치

• GDC
• tri-Ace
• ARM

그래프

• https://www.geogebra.org/graphing
• https://www.desmos.com/calculator
• https://graphtoy.com/
• http://tobyschachman.com/Shadershop/editor/

모델

• https://3dnchu.com/archives/cg-test-models/

튜토리얼

• https://www.ronja-tutorials.com/
• https://guru/unity-cpu-performance/mesh-baker-interview-ian-deane/
• https://www.gamedevpensieve.com/graphics/3d/3d_lighting
• 게임 아티스트를 위한 CG 최적화이론