Base-阴影ShadowMap相关

//比如3级Cascade 一个Tile 2048  ，大ShadowMap是6144x2048。实际每个Tile只画2040x2040，边界留4个像素。
//1 / 6144
const float InvBufferResolutionX = 1.0f / (float)ShadowBufferResolution.X;
//0.5 * 2040 * 1 / 6144
const float ShadowResolutionFractionX = 0.5f * (float)ResolutionX * InvBufferResolutionX;
//1 / 2048
const float InvBufferResolutionY = 1.0f / (float)ShadowBufferResolution.Y;
//0.5 * 2040 * 1 / 2048
const float ShadowResolutionFractionY = 0.5f * (float)ResolutionY * InvBufferResolutionY;

//上面可以算出来0级在ShadowMap的所在UV中心

const FMatrix WorldToShadowMatrix =
// Translate to the origin of the shadow's translated world space
FTranslationMatrix(PreShadowTranslation) *
// Transform into the shadow's post projection space
// This has to be the same transform used to render the shadow depths
SubjectAndReceiverMatrix *
// Scale and translate x and y to be texture coordinates into the ShadowInfo's rectangle in the shadow depth buffer
// Normalize z by MaxSubjectDepth, as was done when writing shadow depths
FMatrix(
    FPlane(ShadowResolutionFractionX,0,                            0,                                    0),
    FPlane(0,                         -ShadowResolutionFractionY,0,                                    0),
    FPlane(0,                        0,                            InvMaxSubjectDepth,    0),
    FPlane(
        (X + BorderSize) * InvBufferResolutionX + ShadowResolutionFractionX,  //第一级CasadeX偏移就是(2048 + 4) * 1 / 6144 + 0级在ShadowMap的所在UV.X
        (Y + BorderSize) * InvBufferResolutionY + ShadowResolutionFractionY,  //第一级CasadeY偏移就是(0 + 4) * 1 / 2048 + 0级在ShadowMap的所在UV.Y
        0,
        1
    )
);

FMatrix FProjectedShadowInfo::GetScreenToShadowMatrix(const FSceneView& View, uint32 TileOffsetX, uint32 TileOffsetY, uint32 TileResolutionX, uint32 TileResolutionY) const
{
  const FIntPoint ShadowBufferResolution = GetShadowBufferResolution();
  const float InvBufferResolutionX = 1.0f / (float)ShadowBufferResolution.X;
  const float ShadowResolutionFractionX = 0.5f * (float)TileResolutionX * InvBufferResolutionX;
  const float InvBufferResolutionY = 1.0f / (float)ShadowBufferResolution.Y;
  const float ShadowResolutionFractionY = 0.5f * (float)TileResolutionY * InvBufferResolutionY;
  // Calculate the matrix to transform a screenspace position into shadow map space

  FMatrix ScreenToShadow;
  FMatrix ViewDependentTransform =
    // Z of the position being transformed is actually view space Z,
      // Transform it into post projection space by applying the projection matrix,
      // Which is the required space before applying View.InvTranslatedViewProjectionMatrix
    FMatrix(
      FPlane(1,0,0,0),
      FPlane(0,1,0,0),
      FPlane(0,0,View.ViewMatrices.GetProjectionMatrix().M[2][2],1),
      FPlane(0,0,View.ViewMatrices.GetProjectionMatrix().M[3][2],0)) *
    // Transform the post projection space position into translated world space
    // Translated world space is normal world space translated to the view's origin,
    // Which prevents floating point imprecision far from the world origin.
    View.ViewMatrices.GetInvTranslatedViewProjectionMatrix() *
    FTranslationMatrix(-View.ViewMatrices.GetPreViewTranslation());

  FMatrix ShadowMapDependentTransform =
    // Translate to the origin of the shadow's translated world space
    FTranslationMatrix(PreShadowTranslation) *
    // Transform into the shadow's post projection space
    // This has to be the same transform used to render the shadow depths
    FMatrix(TranslatedWorldToClipInnerMatrix) *
    // Scale and translate x and y to be texture coordinates into the ShadowInfo's rectangle in the shadow depth buffer
    // Normalize z by MaxSubjectDepth, as was done when writing shadow depths
    FMatrix(
      FPlane(ShadowResolutionFractionX,0,							0,									0),
      FPlane(0,						 -ShadowResolutionFractionY,0,									0),
      FPlane(0,						0,							InvMaxSubjectDepth,	0),
      FPlane(
        (TileOffsetX + BorderSize) * InvBufferResolutionX + ShadowResolutionFractionX,
        (TileOffsetY + BorderSize) * InvBufferResolutionY + ShadowResolutionFractionY,
        0,
        1
        )
      );

  if (View.bIsMobileMultiViewEnabled && View.Family->Views.Num() > 0)
  {
    // In Multiview, we split ViewDependentTransform out into ViewUniformShaderParameters.MobileMultiviewShadowTransform
    // So we can multiply it later in shader.
    ScreenToShadow = ShadowMapDependentTransform;
  }
  else
  {
    ScreenToShadow = ViewDependentTransform * ShadowMapDependentTransform;
  }
  return ScreenToShadow;
}

MyShadowData data;
data.shadowMask.distance = false;
data.shadowMask.shadows = 1.0;
data.shadowMask.always = false;
data.cascadeBlend = 1.0;
//最大距离之外无阴影,做渐变
//这个距离衰减基于视角空间，作为3种阴影的全局衰减
data.strength = FadeShadowStrength(surfaceWS.depth, _ShadowDistanceFade.x, _ShadowDistanceFade.y);
int i = 0;
//计算出应该采样哪一级cascade，最后i就是级联层级
for(i = 0; i< _CascadeCount; i++){
  float4 sphere = _CascadeCullingSpheres[i];
  float distanceSqr = DistanceSquared(surfaceWS.position, sphere.xyz);
  //平方对比
  if(distanceSqr < sphere.w){
    float fade = FadeShadowStrength(
      distanceSqr, _CascadeData[i].x, _ShadowDistanceFade.z
    );

    if(i == _CascadeCount - 1)
    {
      data.strength *= fade;   //最大距离的
    }
    else
    {
      data.cascadeBlend = fade; //级联的Fade
    }
    break;
  }
}