Init
This commit is contained in:
BIN
03-UnrealEngine/Rendering/Shader/Effect/Ocean/水下逻辑判断.jpg
Normal file
BIN
03-UnrealEngine/Rendering/Shader/Effect/Ocean/水下逻辑判断.jpg
Normal file
Binary file not shown.
|
After Width: | Height: | Size: 56 KiB |
BIN
03-UnrealEngine/Rendering/Shader/Effect/Ocean/水下逻辑判断2.jpg
Normal file
BIN
03-UnrealEngine/Rendering/Shader/Effect/Ocean/水下逻辑判断2.jpg
Normal file
Binary file not shown.
|
After Width: | Height: | Size: 2.6 KiB |
76
03-UnrealEngine/Rendering/Shader/Effect/Ocean/海洋SSS效果论文笔记.md
Normal file
76
03-UnrealEngine/Rendering/Shader/Effect/Ocean/海洋SSS效果论文笔记.md
Normal file
@@ -0,0 +1,76 @@
|
||||
## 论文与相关资料
|
||||
### 寒霜的快速SSS
|
||||
在GDCVault中搜索对应的演讲，之后就可以下载了。一些PPT比较大，可以直接去控件里找下载地址。
|
||||
|
||||
https://www.slideshare.net/colinbb/colin-barrebrisebois-gdc-2011-approximating-translucency-for-a-fast-cheap-and-convincing-subsurfacescattering-look-7170855
|
||||
|
||||
#### 视频
|
||||
https://www.gdcvault.com/play/1014536/Approximating-Translucency-for-a-Fast
|
||||
|
||||
因为视频是blob模式的，所以可以去下面的网站下载：
|
||||
http://downloadblob.com/
|
||||
|
||||
#### ppt
|
||||
https://twvideo01.ubm-us.net/o1/vault/gdc2011/slides/Colin_BarreBrisebois_Programming_ApproximatingTranslucency.pptx
|
||||
|
||||
### SIGGRAPH2019ppt
|
||||
http://advances.realtimerendering.com/s2019/index.htm
|
||||
|
||||
## Approximating Translucency for a Fast, Cheap and Convincing Subsurface Scattering Look
|
||||
### 数据管理
|
||||
数据可以分成材质相关与灯光类型相关。在寒霜中材质相关会使用GBuffer传递（UE4可以使用CustomData吧），光类型相关会在LightPass中传递。
|
||||
|
||||
### 计算厚度值
|
||||
通过AO的方式来计算厚度:
|
||||
- 反转表面法线
|
||||
- 渲染AO
|
||||
- 反转颜色最后渲染到贴图中
|
||||
|
||||
### 技术细节
|
||||
|
||||
|
||||
LT应该是Light Translucency的意思。v代表Vector，f代表float，i代表int。
|
||||
```
|
||||
half3 vLTLight = vLight + vNormal * fLTDistortion;
|
||||
half fLTDot = pow(saturate(dot(vEye, -vLTLight)), iLTPower) * fLTScale;
|
||||
half3 fLT = fLightAttenuation * (fLTDot + fLTAmbient) * fLTThickness;
|
||||
outColor.rgb += cDiffuseAlbedo * cLightDiffuse * fLT;
|
||||
```
|
||||
|
||||
#### fLTAmbient
|
||||
Ambient项，代表了始终存的各个方向的透射值。材质相关变量。
|
||||
|
||||
|
||||
|
||||
#### iLTPower
|
||||
强度衰减项，直接透射强度。与视口相关。可以通过预计算进行优化。材质相关变量。
|
||||
|
||||
|
||||
|
||||
#### fLTDistortion
|
||||
透射方向形变项，用于模拟光线传输的不规则的效果，类似于毛玻璃的效果。主要的功能是控制法线对于透射光方向的影响。与视口相关。材质相关变量。
|
||||
|
||||
|
||||
|
||||
#### fLTThickness
|
||||
厚底项，预计算的Local坐标的表面厚度值。材质相关变量。
|
||||
|
||||
|
||||
|
||||
#### fLTScale
|
||||
缩放项，用于缩放直接透射效果。视口相关。灯光相关变量。
|
||||
|
||||
|
||||
|
||||
### 最终效果
|
||||
|
||||
|
||||
### GBuffer设计
|
||||
最低的要求是GBufer中使用8位位宽灰度贴图的方式来存储translucency。使用24位位宽以颜色贴图的方式，可以实现材质对于不同光谱的光线不同的散射效果。
|
||||
|
||||
### 技术缺点
|
||||
因为是一种近似技术，所以只适合在凸物体上演示。这种技术对变形物体不起作用，因为需要烘焙厚度贴图。此外，我们可以使用实时AO算法配合倒置法线来计算厚度。
|
||||
|
||||
|
||||
|
||||
PS.该技术的详细描述可以在《GPU PRO2》中找到。
|
||||
202
03-UnrealEngine/Rendering/Shader/Effect/Ocean/海洋论文笔记.md
Normal file
202
03-UnrealEngine/Rendering/Shader/Effect/Ocean/海洋论文笔记.md
Normal file
@@ -0,0 +1,202 @@
|
||||
## UE4 渲染功能探究
|
||||
New: Planar Reflections
|
||||
New: High Quality Reflections
|
||||
|
||||
## UE4.26 SingleLayerWater笔记
|
||||
官方论坛讨论
|
||||
|
||||
https://forums.unrealengine.com/development-discussion/rendering/1746626-actually-realistic-water-shader#post1789028
|
||||
### SingleLayerCommon.ush
|
||||
计算光照强度、透明度。
|
||||
struct WaterVolumeLightingOutput
|
||||
{
|
||||
float3 Luminance;
|
||||
float3 WaterToSceneTransmittance;
|
||||
float3 WaterToSceneToLightTransmittance;
|
||||
};
|
||||
|
||||
Output.Luminance = WaterVisibility * (ScatteredLuminance + Transmittance * BehindWaterSceneLuminance);
|
||||
Output.WaterToSceneTransmittance = Transmittance;
|
||||
Output.WaterToSceneToLightTransmittance;
|
||||
|
||||
目前没有开启RayMarching,所以核心代码为：
|
||||
```
|
||||
const float3 OpticalDepth = ExtinctionCoeff * BehindWaterDeltaDepth;
|
||||
float3 Transmittance = exp(-OpticalDepth);
|
||||
float3 ScatteredLuminance = ScatteringCoeff * (AmbScattLuminance + SunScattLuminance * DirectionalLightShadow);
|
||||
ScatteredLuminance = (ScatteredLuminance - ScatteredLuminance * Transmittance) / ExtinctionCoeffSafe;
|
||||
|
||||
// Apply Fresnel effect to out-scattering towards the view
|
||||
ScatteredLuminance *= CameraIsUnderWater ? 1.0 : (1.0 - EnvBrdf); // Under water is less visible due to Fresnel effect
|
||||
Transmittance *= CameraIsUnderWater ? (1.0 - EnvBrdf) : 1.0; // Above " " " " "
|
||||
|
||||
// Add single in-scattering apply colored transmittance to scene color
|
||||
Output.Luminance = WaterVisibility * (ScatteredLuminance + Transmittance * (BehindWaterSceneLuminance* ColorScaleBehindWater));
|
||||
Output.WaterToSceneTransmittance = Transmittance;
|
||||
Output.WaterToSceneToLightTransmittance = Transmittance * MeanTransmittanceToLightSources;
|
||||
```
|
||||
|
||||
|
||||
```c++
|
||||
const float BehindWaterDeltaDepth = CameraIsUnderWater ? WaterDepth : max(0.0f, SceneDepth - WaterDepth);
|
||||
|
||||
const float3 ScatteringCoeff = max(0.0f, GetSingleLayerWaterMaterialOutput0(MaterialParameters));
|
||||
const float3 AbsorptionCoeff = max(0.0f, GetSingleLayerWaterMaterialOutput1(MaterialParameters));
|
||||
const float PhaseG = clamp(GetSingleLayerWaterMaterialOutput2(MaterialParameters), -1.0f, 1.0f);
|
||||
//Sample the optional Material Input ColorScaleBehindWater and fade it out at shorelines to avoid hard edge intersections
|
||||
float3 ColorScaleBehindWater = lerp(1.0f, max(0.0f, GetSingleLayerWaterMaterialOutput3(MaterialParameters)), saturate(BehindWaterDeltaDepth * 0.02f));
|
||||
|
||||
const float3 ExtinctionCoeff = ScatteringCoeff + AbsorptionCoeff;
|
||||
// Max to avoid division by 0 with the analytical integral below.
|
||||
// 1e-5 is high enough to avoid denorms on mobile
|
||||
const float3 ExtinctionCoeffSafe = max(ScatteringCoeff + AbsorptionCoeff, 1e-5);
|
||||
|
||||
float DirLightPhaseValue = 0.0f; // Default when Total Internal Reflection happens.
|
||||
{
|
||||
#if SIMPLE_SINGLE_LAYER_WATER
|
||||
DirLightPhaseValue = IsotropicPhase();
|
||||
#else
|
||||
float IorFrom = 1.0f; // assumes we come from air
|
||||
float IorTo = DielectricF0ToIor(DielectricSpecularToF0(Specular)); // Wrong if metal is set to >1. But we still keep refraction on the water surface nonetheless.
|
||||
const float relativeIOR = IorFrom / IorTo;
|
||||
float3 UnderWaterRayDir = 0.0f;
|
||||
if (WaterRefract(MaterialParameters.CameraVector, MaterialParameters.WorldNormal, relativeIOR, UnderWaterRayDir))
|
||||
{
|
||||
DirLightPhaseValue = SchlickPhase(PhaseG, dot(-ResolvedView.DirectionalLightDirection.xyz, UnderWaterRayDir));
|
||||
}
|
||||
#endif
|
||||
}
|
||||
|
||||
// We also apply transmittance from light to under water surface. However, the scene has been lit by many sources already.
|
||||
// So the transmittance toabove surface is simply approximated using the travel distance from the scene pixel to the water top, assuming a flat water surface.
|
||||
// We cannot combine this transmittance with the transmittance from view because this would change the behavior of the analytical integration of light scattering integration.
|
||||
const float3 BehindWaterSceneWorldPos = SvPositionToWorld(float4(MaterialParameters.SvPosition.xy, SceneDeviceZ, 1.0));
|
||||
const float DistanceFromScenePixelToWaterTop = max(0.0, MaterialParameters.AbsoluteWorldPosition.z - BehindWaterSceneWorldPos.z);
|
||||
const float3 MeanTransmittanceToLightSources = exp(-DistanceFromScenePixelToWaterTop * ExtinctionCoeff);
|
||||
|
||||
#if SIMPLE_SINGLE_LAYER_WATER
|
||||
const float3 BehindWaterSceneLuminance = 0.0f; // Cannot read back the scene color in this case
|
||||
#else
|
||||
// We use the pixel SvPosition instead of the scene one pre refraction/distortion to avoid those extra ALUs.
|
||||
float3 BehindWaterSceneLuminance = SceneColorWithoutSingleLayerWaterTexture.SampleLevel(SceneColorWithoutSingleLayerWaterSampler, ViewportUV, 0).rgb;
|
||||
BehindWaterSceneLuminance = MeanTransmittanceToLightSources * (USE_PREEXPOSURE ? ResolvedView.OneOverPreExposure : 1.0f) * BehindWaterSceneLuminance;
|
||||
#endif
|
||||
|
||||
float3 SunScattLuminance = DirLightPhaseValue * SunIlluminance;
|
||||
float3 AmbScattLuminance = IsotropicPhase() * AmbiantIlluminance;
|
||||
|
||||
#define VOLUMETRICSHADOW 0
|
||||
#if !VOLUMETRICSHADOW || SIMPLE_SINGLE_LAYER_WATER
|
||||
|
||||
const float3 OpticalDepth = ExtinctionCoeff * BehindWaterDeltaDepth;
|
||||
float3 Transmittance = exp(-OpticalDepth);
|
||||
float3 ScatteredLuminance = ScatteringCoeff * (AmbScattLuminance + SunScattLuminance * DirectionalLightShadow);
|
||||
ScatteredLuminance = (ScatteredLuminance - ScatteredLuminance * Transmittance) / ExtinctionCoeffSafe;
|
||||
|
||||
#else
|
||||
// TODO Make the volumetric shadow part work again
|
||||
float3 Transmittance = 1.0f;
|
||||
float3 ScatteredLuminance = 0.0f;
|
||||
const float RayMarchMaxDistance = min(BehindWaterDeltaDepth, 200.0f); // 20 meters
|
||||
const float RayMarchStepSize = RayMarchMaxDistance / 10.0f; // Less samples wil lresult in a bit brighter look due to TransmittanceToLightThroughWater being 1 on a longer first sample. Would need it part of analiytical integration
|
||||
const float ShadowDither = RayMarchStepSize * GBufferDither;
|
||||
for (float s = 0.0f; s < RayMarchMaxDistance; s += RayMarchStepSize)
|
||||
{
|
||||
// Only jitter shadow map sampling to not lose energy on first sample
|
||||
float Shadow = ComputeDirectionalLightDynamicShadowing(MaterialParameters.AbsoluteWorldPosition - (s + ShadowDither)*MaterialParameters.CameraVector, GBuffer.Depth);
|
||||
|
||||
float3 WP = MaterialParameters.AbsoluteWorldPosition - s * MaterialParameters.CameraVector;
|
||||
float WaterHeightAboveSample = max(0.0, MaterialParameters.AbsoluteWorldPosition.z - WP.z);
|
||||
float3 TransmittanceToLightThroughWater = 1.0; // no self shadow, same energy as above analytical solution
|
||||
//float3 TransmittanceToLightThroughWater = exp(-ExtinctionCoeff * WaterHeightAboveSample); // self shadow as transmittance to water level, close to reference, depends a bit on sample count due to first sample being critical for dense medium
|
||||
|
||||
float3 SampleTransmittance = exp(-ExtinctionCoeff * RayMarchStepSize); // Constant
|
||||
float3 SS = (ScatteringCoeff * TransmittanceToLightThroughWater * (SunScattLuminance * Shadow + AmbScattLuminance));
|
||||
ScatteredLuminance += Transmittance * (SS - SS * SampleTransmittance) / ExtinctionCoeffSafe;
|
||||
Transmittance *= SampleTransmittance;
|
||||
}
|
||||
|
||||
// The rest of the medium
|
||||
const float3 OpticalDepth2 = ExtinctionCoeff * max(0.0, BehindWaterDeltaDepth - RayMarchMaxDistance);
|
||||
if (any(OpticalDepth2 > 0.0f))
|
||||
{
|
||||
float3 Transmittance2 = exp(-OpticalDepth2);
|
||||
float3 ScatteredLuminance2 = ScatteringCoeff * (SunScattLuminance + AmbScattLuminance);
|
||||
ScatteredLuminance += Transmittance * (ScatteredLuminance2 - ScatteredLuminance2 * Transmittance2) / ExtinctionCoeffSafe;
|
||||
Transmittance *= Transmittance2;
|
||||
}
|
||||
#endif
|
||||
|
||||
// Apply Fresnel effect to out-scattering towards the view
|
||||
ScatteredLuminance *= CameraIsUnderWater ? 1.0 : (1.0 - EnvBrdf); // Under water is less visible due to Fresnel effect
|
||||
Transmittance *= CameraIsUnderWater ? (1.0 - EnvBrdf) : 1.0; // Above " " " " "
|
||||
|
||||
// Add single in-scattering apply colored transmittance to scene color
|
||||
Output.Luminance = WaterVisibility * (ScatteredLuminance + Transmittance * (BehindWaterSceneLuminance* ColorScaleBehindWater));
|
||||
Output.WaterToSceneTransmittance = Transmittance;
|
||||
Output.WaterToSceneToLightTransmittance = Transmittance * MeanTransmittanceToLightSources;
|
||||
}
|
||||
|
||||
```
|
||||
海洋是不透明的，使用SceneColor缓存合成出的透明效果。
|
||||
|
||||
## GDC2012 神秘海域3演讲
|
||||
### 渲染方案
|
||||
|
||||
### FlowShader
|
||||
没看懂，为什么需要用2张贴图叠加，是因为要过度么？
|
||||
|
||||
4.5.1.1 Flow Map变体：《神秘海域3》Flow Map + Displacement
|
||||
另外，Flow Map可以和其他渲染技术结合使用，比如《神秘海域3》中的Flow Map + Displacement：
|
||||
|
||||
|
||||
4.5.1.2 Flow Map变体：《堡垒之夜》Flow Map + Distance Fields + Normal Maps
|
||||
以及《堡垒之夜》中的Flow Map + Distance Fields + Normal Maps [GDC 2019, Technical Artist Bootcamp Distance Fields and Shader Simulation Tricks]
|
||||
|
||||
4.5.1.3 Flow Map变体：《神秘海域4》Flow Map + Wave Particles
|
||||
或者《神秘海域4》中的Flow Map + Wave Particles[SIGGRAPH 2016, Rendering Rapids in Uncharted 4]，都是进阶模拟水体表面流动与起伏效果的不错选择。
|
||||
|
||||
### Wave System
|
||||
如果我们能找到一个好的模型，程序化的几何和动画是不错的。
|
||||
仿真计算成本太高（即使在SPU中），设计者也很难控制。
|
||||
Perlin噪音效果在视觉上不是很好，往往看起来很人工化
|
||||
FFT技术很好，但是参数很难被艺术家控制和调整。也是很难搞好的
|
||||
|
||||
**Gerstner waves**
|
||||
简单易于控制效果，但高频细节不够多，只能叠加几组大浪，否则太消耗资源。
|
||||
|
||||
**FFT Waves**
|
||||
真实，细节丰富。但是美术难以控制效果。
|
||||
|
||||
神秘海域3采用4组Gerstner waves+4组波动粒子的方式来实现Wave Vector Displacement。
|
||||
|
||||
#### 大浪
|
||||
大浪采用贝塞尔曲线建模完成
|
||||
**之后再叠加大浪**。
|
||||
|
||||
|
||||
|
||||
这是整个波系的局部公式。
|
||||
bspline是一个均匀的、非理性的bspline。我们本可以使用贝塞尔，但它需要更多的代码。
|
||||
grid(u,v)函数返回一个给定坐标u,v的标量值。在这种情况下，我们有一个波标的乘数
|
||||
|
||||
## Sea of Thieves approach [Ang, 2018]
|
||||
|
||||
## Crest Siggraph2019
|
||||
### Light Scattering
|
||||
使用了类似盗贼之海的光线散射算法，光线散射项是基于海面置换项的水平长度。这里补充一下：使它在我们的框架中更好地工作--我们通过将置换项除以波长来做一种特殊的归一化，并将这个归一化版本用于光散射项。
|
||||
|
||||
基于海平面高度的散射在海洋参数发生改变时容易出问题。
|
||||
|
||||
如果将置换项除以波长，就可以针对大波与小波进行缩放。
|
||||
|
||||
### Shadering
|
||||
Cascade scale used to scale shading inputs
|
||||
Normals
|
||||
Foam
|
||||
Underwater bubbles
|
||||
Works for range of viewpoints
|
||||
Breaks up patterns
|
||||
Combats mipmapping
|
||||
Increase visible range of detail
|
||||
Doubles texture samples
|
||||
252
03-UnrealEngine/Rendering/Shader/Effect/PostProcess 后处理材质笔记.md
Normal file
252
03-UnrealEngine/Rendering/Shader/Effect/PostProcess 后处理材质笔记.md
Normal file
@@ -0,0 +1,252 @@
|
||||
|
||||
|
||||
|
||||
|
||||
```c++
|
||||
// Manually clamp scene texture UV as if using a clamp sampler.
|
||||
MaterialFloat2 ClampSceneTextureUV(MaterialFloat2 BufferUV, const uint SceneTextureId)
|
||||
{
|
||||
float4 MinMax = GetSceneTextureUVMinMax(SceneTextureId);
|
||||
|
||||
return clamp(BufferUV, MinMax.xy, MinMax.zw);
|
||||
}
|
||||
```
|
||||
|
||||
## 相关函数的HLSL代码
|
||||
```c++
|
||||
float GetPixelDepth(FMaterialVertexParameters Parameters)
|
||||
{
|
||||
FLATTEN
|
||||
if (View.ViewToClip[3][3] < 1.0f)
|
||||
{
|
||||
// Perspective
|
||||
return GetScreenPosition(Parameters).w;
|
||||
}
|
||||
else
|
||||
{
|
||||
// Ortho
|
||||
return ConvertFromDeviceZ(GetScreenPosition(Parameters).z);
|
||||
}
|
||||
}
|
||||
|
||||
float GetPixelDepth(FMaterialPixelParameters Parameters)
|
||||
{
|
||||
FLATTEN
|
||||
if (View.ViewToClip[3][3] < 1.0f)
|
||||
{
|
||||
// Perspective
|
||||
return GetScreenPosition(Parameters).w;
|
||||
}
|
||||
else
|
||||
{
|
||||
// Ortho
|
||||
return ConvertFromDeviceZ(GetScreenPosition(Parameters).z);
|
||||
}
|
||||
}
|
||||
|
||||
float2 GetSceneTextureUV(FMaterialVertexParameters Parameters)
|
||||
{
|
||||
return ScreenAlignedPosition(GetScreenPosition(Parameters));
|
||||
}
|
||||
|
||||
float2 GetSceneTextureUV(FMaterialPixelParameters Parameters)
|
||||
{
|
||||
return SvPositionToBufferUV(Parameters.SvPosition);
|
||||
}
|
||||
|
||||
float2 GetViewportUV(FMaterialVertexParameters Parameters)
|
||||
{
|
||||
#if POST_PROCESS_MATERIAL
|
||||
return Parameters.WorldPosition.xy;
|
||||
#else
|
||||
return BufferUVToViewportUV(GetSceneTextureUV(Parameters));
|
||||
#endif
|
||||
}
|
||||
|
||||
float2 GetPixelPosition(FMaterialVertexParameters Parameters)
|
||||
{
|
||||
return GetViewportUV(Parameters) * View.ViewSizeAndInvSize.xy;
|
||||
}
|
||||
|
||||
|
||||
#if POST_PROCESS_MATERIAL
|
||||
|
||||
float2 GetPixelPosition(FMaterialPixelParameters Parameters)
|
||||
{
|
||||
return Parameters.SvPosition.xy - float2(PostProcessOutput_ViewportMin);
|
||||
}
|
||||
|
||||
float2 GetViewportUV(FMaterialPixelParameters Parameters)
|
||||
{
|
||||
return GetPixelPosition(Parameters) * PostProcessOutput_ViewportSizeInverse;
|
||||
}
|
||||
|
||||
#else
|
||||
|
||||
float2 GetPixelPosition(FMaterialPixelParameters Parameters)
|
||||
{
|
||||
return Parameters.SvPosition.xy - float2(View.ViewRectMin.xy);
|
||||
}
|
||||
|
||||
float2 GetViewportUV(FMaterialPixelParameters Parameters)
|
||||
{
|
||||
return SvPositionToViewportUV(Parameters.SvPosition);
|
||||
}
|
||||
|
||||
```
|
||||
|
||||
## 本质上是调用了SceneTextureLookup
|
||||
float4 SceneTextureLookup(float2 UV, int SceneTextureIndex, bool bFiltered)
|
||||
{
|
||||
FScreenSpaceData ScreenSpaceData = GetScreenSpaceData(UV, false);
|
||||
switch(SceneTextureIndex)
|
||||
{
|
||||
// order needs to match to ESceneTextureId
|
||||
|
||||
case PPI_SceneColor:
|
||||
return float4(CalcSceneColor(UV), 0);
|
||||
case PPI_SceneDepth:
|
||||
return ScreenSpaceData.GBuffer.Depth;
|
||||
case PPI_DiffuseColor:
|
||||
return float4(ScreenSpaceData.GBuffer.DiffuseColor, 0);
|
||||
case PPI_SpecularColor:
|
||||
return float4(ScreenSpaceData.GBuffer.SpecularColor, 0);
|
||||
case PPI_SubsurfaceColor:
|
||||
return IsSubsurfaceModel(ScreenSpaceData.GBuffer.ShadingModelID) ? float4( ExtractSubsurfaceColor(ScreenSpaceData.GBuffer), ScreenSpaceData.GBuffer.CustomData.a ) : ScreenSpaceData.GBuffer.CustomData;
|
||||
case PPI_BaseColor:
|
||||
return float4(ScreenSpaceData.GBuffer.BaseColor, 0);
|
||||
case PPI_Specular:
|
||||
return ScreenSpaceData.GBuffer.Specular;
|
||||
case PPI_Metallic:
|
||||
return ScreenSpaceData.GBuffer.Metallic;
|
||||
case PPI_WorldNormal:
|
||||
return float4(ScreenSpaceData.GBuffer.WorldNormal, 0);
|
||||
case PPI_SeparateTranslucency:
|
||||
return float4(1, 1, 1, 1); // todo
|
||||
case PPI_Opacity:
|
||||
return ScreenSpaceData.GBuffer.CustomData.a;
|
||||
case PPI_Roughness:
|
||||
return ScreenSpaceData.GBuffer.Roughness;
|
||||
case PPI_MaterialAO:
|
||||
return ScreenSpaceData.GBuffer.GBufferAO;
|
||||
case PPI_CustomDepth:
|
||||
return ScreenSpaceData.GBuffer.CustomDepth;
|
||||
|
||||
case PPI_PostProcessInput0:
|
||||
return Texture2DSample(PostProcessInput_0_Texture, bFiltered ? PostProcessInput_BilinearSampler : PostProcessInput_0_SharedSampler, UV);
|
||||
case PPI_PostProcessInput1:
|
||||
return Texture2DSample(PostProcessInput_1_Texture, bFiltered ? PostProcessInput_BilinearSampler : PostProcessInput_1_SharedSampler, UV);
|
||||
case PPI_PostProcessInput2:
|
||||
return Texture2DSample(PostProcessInput_2_Texture, bFiltered ? PostProcessInput_BilinearSampler : PostProcessInput_2_SharedSampler, UV);
|
||||
case PPI_PostProcessInput3:
|
||||
return Texture2DSample(PostProcessInput_3_Texture, bFiltered ? PostProcessInput_BilinearSampler : PostProcessInput_3_SharedSampler, UV);
|
||||
case PPI_PostProcessInput4:
|
||||
return Texture2DSample(PostProcessInput_4_Texture, bFiltered ? PostProcessInput_BilinearSampler : PostProcessInput_4_SharedSampler, UV);
|
||||
|
||||
case PPI_DecalMask:
|
||||
return 0; // material compiler will return an error
|
||||
case PPI_ShadingModelColor:
|
||||
return float4(GetShadingModelColor(ScreenSpaceData.GBuffer.ShadingModelID), 1);
|
||||
case PPI_ShadingModelID:
|
||||
return float4(ScreenSpaceData.GBuffer.ShadingModelID, 0, 0, 0);
|
||||
case PPI_AmbientOcclusion:
|
||||
return ScreenSpaceData.AmbientOcclusion;
|
||||
case PPI_CustomStencil:
|
||||
return ScreenSpaceData.GBuffer.CustomStencil;
|
||||
case PPI_StoredBaseColor:
|
||||
return float4(ScreenSpaceData.GBuffer.StoredBaseColor, 0);
|
||||
case PPI_StoredSpecular:
|
||||
return float4(ScreenSpaceData.GBuffer.StoredSpecular.rrr, 0);
|
||||
|
||||
case PPI_WorldTangent:
|
||||
return float4(ScreenSpaceData.GBuffer.WorldTangent, 0);
|
||||
case PPI_Anisotropy:
|
||||
return ScreenSpaceData.GBuffer.Anisotropy;
|
||||
default:
|
||||
return float4(0, 0, 0, 0);
|
||||
}
|
||||
}
|
||||
##
|
||||
FScreenSpaceData GetScreenSpaceData(float2 UV, bool bGetNormalizedNormal = true)
|
||||
{
|
||||
FScreenSpaceData Out;
|
||||
|
||||
Out.GBuffer = GetGBufferData(UV, bGetNormalizedNormal);
|
||||
float4 ScreenSpaceAO = Texture2DSampleLevel(SceneTexturesStruct.ScreenSpaceAOTexture, SceneTexturesStruct_ScreenSpaceAOTextureSampler, UV, 0);
|
||||
|
||||
Out.AmbientOcclusion = ScreenSpaceAO.r;
|
||||
|
||||
return Out;
|
||||
}
|
||||
|
||||
## GetGBufferData使用的GBufferXTexture的传入位置
|
||||
比如AOUpsamplePS（c++中为FDistanceFieldAOUpsamplePS）就带有FSceneTextureUniformParameters, SceneTextures。都是由对应的渲染函数的TRDGUniformBufferRef<FSceneTextureUniformParameters> SceneTexturesUniformBuffer形参传入。
|
||||
|
||||
```c#
|
||||
FGBufferData GetGBufferData(float2 UV, bool bGetNormalizedNormal = true)
|
||||
{
|
||||
float4 GBufferA = Texture2DSampleLevel(SceneTexturesStruct.GBufferATexture, SceneTexturesStruct_GBufferATextureSampler, UV, 0);
|
||||
float4 GBufferB = Texture2DSampleLevel(SceneTexturesStruct.GBufferBTexture, SceneTexturesStruct_GBufferBTextureSampler, UV, 0);
|
||||
float4 GBufferC = Texture2DSampleLevel(SceneTexturesStruct.GBufferCTexture, SceneTexturesStruct_GBufferCTextureSampler, UV, 0);
|
||||
float4 GBufferD = Texture2DSampleLevel(SceneTexturesStruct.GBufferDTexture, SceneTexturesStruct_GBufferDTextureSampler, UV, 0);
|
||||
float CustomNativeDepth = Texture2DSampleLevel(SceneTexturesStruct.CustomDepthTexture, SceneTexturesStruct_CustomDepthTextureSampler, UV, 0).r;
|
||||
|
||||
int2 IntUV = (int2)trunc(UV * View.BufferSizeAndInvSize.xy);
|
||||
uint CustomStencil = SceneTexturesStruct.CustomStencilTexture.Load(int3(IntUV, 0)) STENCIL_COMPONENT_SWIZZLE;
|
||||
|
||||
#if ALLOW_STATIC_LIGHTING
|
||||
float4 GBufferE = Texture2DSampleLevel(SceneTexturesStruct.GBufferETexture, SceneTexturesStruct_GBufferETextureSampler, UV, 0);
|
||||
#else
|
||||
float4 GBufferE = 1;
|
||||
#endif
|
||||
|
||||
float4 GBufferF = Texture2DSampleLevel(SceneTexturesStruct.GBufferFTexture, SceneTexturesStruct_GBufferFTextureSampler, UV, 0);
|
||||
|
||||
#if WRITES_VELOCITY_TO_GBUFFER
|
||||
float4 GBufferVelocity = Texture2DSampleLevel(SceneTexturesStruct.GBufferVelocityTexture, SceneTexturesStruct_GBufferVelocityTextureSampler, UV, 0);
|
||||
#else
|
||||
float4 GBufferVelocity = 0;
|
||||
#endif
|
||||
|
||||
float SceneDepth = CalcSceneDepth(UV);
|
||||
|
||||
//FGBufferData DecodeGBufferData()通过解码GBuffer之后返回FGBufferData 结构体
|
||||
return DecodeGBufferData(GBufferA, GBufferB, GBufferC, GBufferD, GBufferE, GBufferF, GBufferVelocity, CustomNativeDepth, CustomStencil, SceneDepth, bGetNormalizedNormal, CheckerFromSceneColorUV(UV));
|
||||
}
|
||||
```
|
||||
|
||||
## GetGBufferDataFromSceneTextures中使用的GBufferXTexture的传入位置
|
||||
SingleLayerWaterCompositePS中使用了GetGBufferDataFromSceneTextures()来获取GBuffer，这个数据是在SingleLayerWaterRendering.cpp中传入的。传入的变量FSingleLayerWaterCommonShaderParameters->FSceneTextureParameters SceneTextures中带有GBufferABCEDF与Depth、Stencil、Velocity贴图。
|
||||
```c++
|
||||
Texture2D SceneDepthTexture;
|
||||
Texture2D<uint2> SceneStencilTexture;
|
||||
Texture2D GBufferATexture;
|
||||
Texture2D GBufferBTexture;
|
||||
Texture2D GBufferCTexture;
|
||||
Texture2D GBufferDTexture;
|
||||
Texture2D GBufferETexture;
|
||||
Texture2D GBufferVelocityTexture;
|
||||
Texture2D GBufferFTexture;
|
||||
Texture2D<uint> SceneLightingChannels;
|
||||
|
||||
FGBufferData GetGBufferDataFromSceneTextures(float2 UV, bool bGetNormalizedNormal = true)
|
||||
{
|
||||
float4 GBufferA = GBufferATexture.SampleLevel(GBufferATextureSampler, UV, 0);
|
||||
float4 GBufferB = GBufferBTexture.SampleLevel(GBufferBTextureSampler, UV, 0);
|
||||
float4 GBufferC = GBufferCTexture.SampleLevel(GBufferCTextureSampler, UV, 0);
|
||||
float4 GBufferD = GBufferDTexture.SampleLevel(GBufferDTextureSampler, UV, 0);
|
||||
float4 GBufferE = GBufferETexture.SampleLevel(GBufferETextureSampler, UV, 0);
|
||||
float4 GBufferF = GBufferFTexture.SampleLevel(GBufferFTextureSampler, UV, 0);
|
||||
float4 GBufferVelocity = GBufferVelocityTexture.SampleLevel(GBufferVelocityTextureSampler, UV, 0);
|
||||
|
||||
uint CustomStencil = 0;
|
||||
float CustomNativeDepth = 0;
|
||||
|
||||
float DeviceZ = SampleDeviceZFromSceneTextures(UV);
|
||||
|
||||
float SceneDepth = ConvertFromDeviceZ(DeviceZ);
|
||||
|
||||
//FGBufferData DecodeGBufferData()通过解码GBuffer之后返回FGBufferData 结构体
|
||||
return DecodeGBufferData(GBufferA, GBufferB, GBufferC, GBufferD, GBufferE, GBufferF, GBufferVelocity, CustomNativeDepth, CustomStencil, SceneDepth, bGetNormalizedNormal, CheckerFromSceneColorUV(UV));
|
||||
}
|
||||
```
|
||||
53
03-UnrealEngine/Rendering/Shader/Effect/UE4剖切效果实现.md
Normal file
53
03-UnrealEngine/Rendering/Shader/Effect/UE4剖切效果实现.md
Normal file
@@ -0,0 +1,53 @@
|
||||
## 前言
|
||||
思路是使用平面方程来判断模型裁切，之后在另一面使用UnLit的自发光材质显示剖面。但Ue4的BlendingMaterialAttributes不能指定UnLit作为ShaderModel。所以可以使用我之前开发的多Pass插件搞定。
|
||||
|
||||
## 外部普通表面材质
|
||||
使用平面方程来做一个Mask即可。
|
||||
|
||||
|
||||
## 内部剖面材质
|
||||
坡面材质需要使用UnLit材质模型，这样就不会有阴影与法线的干扰了。但Unlit不能翻转法线，所以需要再勾选**双面渲染**选项。
|
||||
|
||||
|
||||
## 使用方法说明
|
||||
https://zhuanlan.zhihu.com/p/69139579
|
||||
|
||||
1. 使用上面链接介绍的代码，新建一个蓝图类并且挂载UStrokeStaticMeshComponent。
|
||||
2. 赋予UStrokeStaticMeshComponent所需剖切的模型，以及上文制作好的外部表面材质。
|
||||
3. 赋予SecondPassMaterial内部剖面材质，并勾选UStrokeStaticMeshComponent的NeedSecondPass变量。
|
||||
4. 将蓝图类拖到关卡中，并且设置2个材质的PlaneNormal与PlanePoint。PlanePoint为世界坐标，PlaneNormal需要归一化。
|
||||
|
||||
## 最终效果
|
||||
锯齿是因为垃圾笔记本散热差，把效果开低的关系。
|
||||
|
||||
|
||||
|
||||
## 错误的尝试
|
||||
之前还尝试WorldPositionOffset的思路来做，不过后来发现没有实用性。因为剖面的大小与形状是不确定的，用投射的方式来平移内部可见顶点，会造成多余的顶点平移到平面外的问题。所以只适合拿来做规整模型的剖切效果。
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
错误结果：
|
||||
|
||||
|
||||
实现节点如下：
|
||||
|
||||
|
||||
CustomNode代码：
|
||||
```
|
||||
// Input PlaneNormal
|
||||
// Input PlanePoint
|
||||
// Input WorldPosition
|
||||
float NormalPower2= pow(PlaneNormal.x,2) + pow(PlaneNormal.y,2) + pow(PlaneNormal.z,2);
|
||||
float D=dot(PlaneNormal.xyz,PlanePoint.xyz)*-1;
|
||||
|
||||
float posX= ((pow(PlaneNormal.y,2) + pow(PlaneNormal.z,2))*WorldPosition.x-PlaneNormal.x*(PlaneNormal.y*WorldPosition.y+PlaneNormal.z*WorldPosition.z+D))/NormalPower2;
|
||||
|
||||
float posY=((pow(PlaneNormal.x,2) + pow(PlaneNormal.z,2))*WorldPosition.y-PlaneNormal.y*(PlaneNormal.x*WorldPosition.x+PlaneNormal.z*WorldPosition.z+D))/NormalPower2;
|
||||
|
||||
float posZ=((pow(PlaneNormal.x,2) + pow(PlaneNormal.y,2))*WorldPosition.z-PlaneNormal.z*(PlaneNormal.x*WorldPosition.x+PlaneNormal.y*WorldPosition.y+D))/NormalPower2;
|
||||
|
||||
return float3(posX,posY,posZ)-WorldPosition.xyz;
|
||||
```
|
||||
@@ -0,0 +1,57 @@
|
||||
早几年学习后处理Shader的时候，遇到编辑器视口中的渲染结果与视口大小不匹配的问题：
|
||||
|
||||
|
||||
PS.该问题不会在使用SceneTexture节点制作的后处理效果材质中发生。但它不易于维护与修改。所以我花了些时间来解决这个问题。
|
||||
|
||||
## 问题成因与解决方法
|
||||
可以看得出这个问题是因为ViewSize与BufferSize不一造成的。经过一系列测试使用ViewportUVToBufferUV()对UV坐标进行转换就可以解决问题，该函数位于Common.h。
|
||||
```c#
|
||||
float2 ViewportUVToBufferUV(float2 ViewportUV)
|
||||
{
|
||||
float2 PixelPos = ViewportUV * View.ViewSizeAndInvSize.xy;
|
||||
return (PixelPos + View.ViewRectMin.xy) * View.BufferSizeAndInvSize.zw;
|
||||
}
|
||||
```
|
||||
ScreenUV使用GetViewportUV(Parameters)来获取，Step为ResolvedView.ViewSizeAndInvSize.zw，用法大致如下：
|
||||
```c#
|
||||
float GetDepthTestWeight(float2 ScreenUV,float Stepx, float Stepy, float3x3 KernelX, float3x3 KernelY,float DepthTestThreshold)
|
||||
{
|
||||
float3x3 image;
|
||||
float CurrentPixelDepth=GetScreenSpaceData(ViewportUVToBufferUV(ScreenUV)).GBuffer.Depth;
|
||||
image = float3x3(
|
||||
length(GetScreenSpaceData(ViewportUVToBufferUV(ScreenUV + float2(-Stepx,Stepy))).GBuffer.Depth-CurrentPixelDepth),
|
||||
length(GetScreenSpaceData(ViewportUVToBufferUV(ScreenUV + float2(0,Stepy))).GBuffer.Depth-CurrentPixelDepth),
|
||||
length(GetScreenSpaceData(ViewportUVToBufferUV(ScreenUV + float2(Stepx,Stepy))).GBuffer.Depth-CurrentPixelDepth),
|
||||
length(GetScreenSpaceData(ViewportUVToBufferUV(ScreenUV + float2(-Stepx,0))).GBuffer.Depth-CurrentPixelDepth),
|
||||
0,
|
||||
length(GetScreenSpaceData(ViewportUVToBufferUV(ScreenUV + float2(Stepx,0))).GBuffer.Depth-CurrentPixelDepth),
|
||||
length(GetScreenSpaceData(ViewportUVToBufferUV(ScreenUV + float2(-Stepx,-Stepy))).GBuffer.Depth-CurrentPixelDepth),
|
||||
length(GetScreenSpaceData(ViewportUVToBufferUV(ScreenUV + float2(0,-Stepy))).GBuffer.Depth-CurrentPixelDepth),
|
||||
length(GetScreenSpaceData(ViewportUVToBufferUV(ScreenUV + float2(Stepx,-Stepy))).GBuffer.Depth-CurrentPixelDepth)
|
||||
);
|
||||
|
||||
UNROLL
|
||||
for (int i = 0; i < 3; i++) {
|
||||
for (int j = 0; j < 3; j++) {
|
||||
// image[i][j] = image[i][j] > CurrentPixelDepth * DepthTestThreshold ? 1 : 0;
|
||||
image[i][j] = saturate(image[i][j] / (CurrentPixelDepth * DepthTestThreshold));
|
||||
}
|
||||
}
|
||||
|
||||
float2 StrokeWeight;
|
||||
StrokeWeight.x = convolve(KernelX, image);
|
||||
StrokeWeight.y = convolve(KernelY, image);
|
||||
|
||||
return length(StrokeWeight);
|
||||
}
|
||||
```
|
||||
|
||||
当然如果是直接在渲染管线里添加渲染Pass的方式来实现后处理效果，那也不会出现这个问题，因为每帧都会按照View大小创建新的buffer，所以不会出现这个问题。
|
||||
|
||||
## 材质编辑器的相关大小节点。
|
||||
- ScreenPosition：代码为GetViewportUV(FMaterialVertexParameters Parameters)。
|
||||
- ViewSize：可视视口的大小。代码为View.ViewSizeAndInvSize.xy，zw为倒数值。
|
||||
- MF_ScreenResolution：具体有VisibleResolution(就是ViewSize节点输出结果) 与BufferResolution(ViewProperty的RenderTargetSize)
|
||||
- SceneTexelSize（场景纹素大小）：这是一个float2的值，对应着UV，（u，v），uv均为正数。
|
||||
- MF_ScreenAlignedPixelToPixelUVs：RenderTargetSize / TextureSize。 其中TextureSize可以是ViewSize。可以用来实现一些修改View大小后，后处理尺寸不变得效果。
|
||||
- 后处理节点的Size：ClampSceneTextureUV(ViewportUVToSceneTextureUV，SceneTextureId)
|
||||
@@ -0,0 +1,8 @@
|
||||
# 在UE4中实现LowPoly效果
|
||||
|
||||
## 代码
|
||||
|
||||
|
||||
## 效果
|
||||
|
||||
|
||||
8
03-UnrealEngine/Rendering/Shader/Effect/技能CD转圈效果实现.md
Normal file
8
03-UnrealEngine/Rendering/Shader/Effect/技能CD转圈效果实现.md
Normal file
@@ -0,0 +1,8 @@
|
||||
---
|
||||
title: 技能CD转圈效果实现
|
||||
date: 2022-11-04 09:38:11
|
||||
excerpt:
|
||||
tags:
|
||||
rating: ⭐
|
||||
---
|
||||
|
||||
209
03-UnrealEngine/Rendering/Shader/Effect/海洋、云Demo制作方案.md
Normal file
209
03-UnrealEngine/Rendering/Shader/Effect/海洋、云Demo制作方案.md
Normal file
@@ -0,0 +1,209 @@
|
||||
---
|
||||
title: 海洋、云Demo制作方案
|
||||
date: 2022-11-04 09:38:11
|
||||
excerpt:
|
||||
tags:
|
||||
rating: ⭐
|
||||
---
|
||||
# 海洋方案
|
||||
浅墨的总结：https://zhuanlan.zhihu.com/p/95917609
|
||||
|
||||
波形：Gerstner 波、快速傅立叶变换、空间-频谱混合方法、离线FFT贴图烘焙（Offline FFT Texture）
|
||||
着色：Depth Based LUT Approach、Sub-Surface Scattering Approximation Approach
|
||||
白沫：浪尖=>[Tessendorf 2001] 基于雅克比矩阵的方法、岸边=》[GDC 2018]《孤岛惊魂5》基于有向距离场的方法、交互白浪：ASher的Niagara方案https://zhuanlan.zhihu.com/p/100700549
|
||||
|
||||
## 波形方案
|
||||
### 线性波形叠加方法
|
||||
正弦波（Sinusoids Wave）[Max 1981]
|
||||
Gerstner 波 （Gerstner Wave） [Fournier 1986]
|
||||
|
||||
### 统计学模型法
|
||||
快速傅立叶变换（Fast Fourier Transform）[Mastin 1987]
|
||||
空间-频谱混合方法（Spatial -Spectral Approaches）[Thon 2000]
|
||||
|
||||
### 波动粒子方法（不考虑）
|
||||
波动粒子方法（（Wave Particles） [Yuksel 2007]
|
||||
水波小包方法（Water Wave Packets）[Jeschke 2017]
|
||||
水面小波方法（Water Surface Wavelets）[Jeschke 2018]
|
||||
|
||||
### 预渲染
|
||||
流型图（Flow Map）[Vlachos 2010]
|
||||
离线FFT贴图烘焙（Offline FFT Texture） [Torres 2012]
|
||||
离线流体帧动画烘焙（bake to flipbook）[Bowles 2017]
|
||||
|
||||
## 水体渲染中的着色方案
|
||||
《神秘海域3》在2012年SIGGRAPH上的技术分享中有一张分析水体渲染技术非常经典的图，如下。
|
||||
|
||||
|
||||
|
||||
- 漫反射（Diffuse）
|
||||
- 镜面反射（Specular）
|
||||
- 法线贴图（Normal Map）
|
||||
- 折射（Reflection）
|
||||
- 通透感（Translucency）
|
||||
- 基于深度的查找表方法（Depth Based LUT Approach）
|
||||
- 次表面散射（Subsurface Scattering）
|
||||
- 白沫（Foam/WhiteCap）
|
||||
- 流动表现（Flow）
|
||||
- 水下雾效（Underwater Haze）
|
||||
|
||||
### 通透感（Translucency）
|
||||
#### 水体散射效果
|
||||
##### 基于深度的查找表方法（Depth Based LUT Approach）
|
||||
Depth Based-LUT方法的思路是，计算视线方向的水体像素深度，然后基于此深度值采样吸收/散射LUT（Absorb/Scatter LUT）纹理，以控制不同深度水体的上色，得到通透的水体质感表现。
|
||||
|
||||
|
||||
|
||||
|
||||
##### 次表面散射近似方法（Sub-Surface Scattering Approximation Approach）
|
||||
- [SIGGRAPH 2019] Crest Ocean System改进的《盗贼之海》SSS方案。
|
||||
- [GDC 2011] 寒霜引擎的Fast SSS方案。
|
||||
|
||||
经过Crest Ocean System改进的《盗贼之海》的SSS方案可以总结如下：
|
||||
|
||||
假设光更有可能在波浪的一侧被水散射与透射。
|
||||
基于FFT模拟产生的顶点偏移，为波的侧面生成波峰mask
|
||||
根据视角，光源方向和波峰mask的组合，将深水颜色和次表面散射水体颜色之间进行混合，得到次表面散射颜色。
|
||||
将位移值（Displacement）除以波长，并用此缩放后的新的位移值计算得出次表面散射项强度。
|
||||
对应的核心实现代码如下：
|
||||
```hlsl
|
||||
float v = abs(i_view.y);
|
||||
half towardsSun = pow(max(0., dot(i_lightDir, -i_view)),_SubSurfaceSunFallOff);
|
||||
half3 subsurface = (_SubSurfaceBase + _SubSurfaceSun * towardsSun) *_SubSurfaceColour.rgb * _LightColor0 * shadow;
|
||||
subsurface *= (1.0 - v * v) * sssIndensity;
|
||||
```
|
||||
col += subsurface;
|
||||
其中，sssIndensity，即散射强度，由采样位移值计算得出。
|
||||
|
||||
|
||||
图 《Crest Ocean System》中基于次表面散射近似的水体表现
|
||||
|
||||
|
||||
|
||||
[GDC 2011] 寒霜引擎的Fast SSS方案
|
||||
[GDC 2011]上，Frostbite引擎提出的Fast Approximating Subsurface Scattering方案，也可以用于水体渲染的模拟。
|
||||
|
||||
|
||||
|
||||
### 白沫（Foam/WhiteCap）
|
||||
白沫（Foam），在有些文献中也被称为Whitecap，White Water，是一种复杂的现象。即使白沫下方的材质具有其他颜色，白沫也通常看起来是白色的。出现这种现象的原因是因为白沫是由包含空气的流体薄膜组成的。随着每单位体积薄膜的数量增加，所有入射光都被反射而没有任何光穿透到其下方。这种光学现象使泡沫看起来比材质本身更亮，以至于看起来几乎是白色的。
|
||||
|
||||
|
||||
对于白沫的渲染而言，白沫可被视为水面上的纹理，其可直接由对象交互作用，浪花的飞溅，或气泡与水表面碰撞而产生。
|
||||
|
||||
白沫的渲染方案，按大的渲染思路而言，可以分为两类：
|
||||
- 基于动态纹理（dynamic texture）
|
||||
- 基于粒子系统（particle system）
|
||||
|
||||
按照类型，可以将白沫分为三种：
|
||||
- 浪尖白沫
|
||||
- 岸边白沫
|
||||
- 交互白沫
|
||||
|
||||
而按照渲染方法，可将白沫渲染的主流方案总结如下：
|
||||
- 基于粒子系统的方法[Tessendorf 2001]
|
||||
- 基于Saturate高度的模拟方法 [GPU Gems 2]
|
||||
- 基于雅可比矩阵的方法 [Tessendorf 2001]
|
||||
- 屏幕空间方法 [Akinci 2013]
|
||||
- 基于场景深度的方法 [Kozin 2018][Sea of Thieves]
|
||||
- 基于有向距离场的方法 [GDC 2018][Far Cry 5]
|
||||
|
||||
#### 浪尖白沫
|
||||
##### 5.2.1 浪尖白沫：[Tessendorf 2001] 基于雅克比矩阵的方法
|
||||
Tessendorf在其著名的水体渲染paper《Simulating Ocean Water》[Tessendorf 2001]中提出了可以基于雅克比矩阵（Jacobian）为负的部分作为求解白沫分布区域的方案。据此，即可导出一张或多张标记了波峰白沫区域的Folding Map贴图。
|
||||
|
||||
|
||||
《战争雷霆（War Thunder）》团队在CGDC 2015上对此的改进方案为，取雅克比矩阵小于M的区域作为求解白沫的区域，其中M~0.3...05。
|
||||
|
||||
|
||||
图 《战争雷霆（War Thunder）》选取雅克比矩阵小于M的区域作为求解白沫的区域 [CGDC 2015]
|
||||
|
||||
另外，《盗贼之海（Sea of Thieves）》团队在SIGGRPAPH 2018上提出，可以对雅可比矩阵进行偏移，以获得更多白沫。且可以采用渐进模糊（Progressive Blur）来解决噪点（noisy）问题以及提供风格化的白沫表现。
|
||||
|
||||
!(https://vdn1.vzuu.com/SD/cfb76434-ec75-11ea-acfd-5ab503a75443.mp4?disable_local_cache=1&bu=pico&expiration=1601438226&auth_key=1601438226-0-0-1b75c02f8faa2d5a94b74ad8ebb6e595&f=mp4&v=hw)
|
||||
|
||||
图 《盗贼之海》基于雅可比矩阵偏移 + 渐进模糊（Progressive Blur）的风格化白沫表现
|
||||
|
||||
##### 浪尖白沫：[GPU Gems 2] 基于Saturate高度的方法
|
||||
《GPU Gems 2》中提出的白沫渲染方案，思路是将一个预先创建的白沫纹理在高于某一高度H0的顶点上进行混合。白沫纹理的透明度根据以下公式进行计算：
|
||||
|
||||
`$ Foam.a=saturate(\frac {H-H_0} {H_{max}-H_0}) $`
|
||||
|
||||
- 其中， `$H_{max}$` 是泡沫最大时的高度， `$H_0$` 是基准高度，`$H$` 是当前高度。
|
||||
- 白沫纹理可以做成序列帧来表示白沫的产生和消失的演变过程。此动画序列帧既可以由美术师进行制作，也可以采用程序化生成。
|
||||
- 将上述结果和噪声图进行合理的结合，可以得到更真实的泡沫表现。
|
||||
|
||||
#### 岸边白沫：[2012]《刺客信条3》基于Multi Ramp Map的方法
|
||||
《刺客信条3》中的岸边白沫的渲染方案可以总结为：
|
||||
|
||||
- 以规则的间距对地形结构进行离线采样，标记出白沫出现的区域。
|
||||
- 采用高斯模糊和Perlin噪声来丰富泡沫的表现形式，以模拟海岸上泡沫的褪色现象。
|
||||
- 由于白沫是白色的，因此在R，G和B通道中的每个通道中都放置三张灰度图，然后颜色ramp图将定义三者之间的混合比率，来实现稠密、中等、稀疏三个级别的白沫。要修改白沫表现，美术师只需对ramp图进行颜色校正即可。如下图所示：
|
||||
|
||||
|
||||
|
||||
|
||||
#### 浪尖白沫&岸边白沫：[GDC 2018]《孤岛惊魂5》基于有向距离场的方法
|
||||
GDC 2018上《孤岛惊魂5》团队分享的白沫渲染技术也不失为一种优秀的方案，主要思路是基于单张Noise贴图控制白沫颜色，结合两个offset采样Flow Map控制白沫的流动，并基于有向距离场（Signed Distance Field，SDF）控制岩石和海岸线处白沫的出现，然后根据位移对白沫进行混合。
|
||||
|
||||
|
||||
|
||||
### 水面高光
|
||||
|
||||
|
||||
### 水焦散
|
||||
水焦散的渲染 （Rendering Water Caustics）
|
||||
|
||||
GPU Gems 1
|
||||
ShaderBits.com
|
||||
### 其他Demo
|
||||
6.5 NVIDIA UE4 WaveWorks
|
||||
GDC 2017上，NVIDIA和Unreal Engine合作推出了WaveWorks，以集成到Unreal Engine 4.15引擎中的形式放出。
|
||||
|
||||
源代码传送门：https://github.com/NvPhysX/UnrealEngine/tree/WaveWorks
|
||||
|
||||
demo视频：https://www.youtube.com/watch?v=DhrNvZLPBGE&list=PLN8o8XBheMDxCCfKijfZ3IlTP3cUCH6Jq&index=11&t=0s
|
||||
|
||||
## 云方案
|
||||
|
||||
eg2020的一篇很思路很新的论文
|
||||
有空写篇文章分享一下思路
|
||||
下面是改编了一下的shader
|
||||
raymarch的次数降低了很多倍（同样也增加了不少sdf的采样次数）
|
||||
优化思路想到再发。
|
||||
|
||||
https://www.shadertoy.com/view/WdKczW
|
||||
### ASher的方案
|
||||
https://zhuanlan.zhihu.com/p/107016039
|
||||
https://www.bilibili.com/video/BV1FE411n7rU
|
||||
|
||||
https://www.bilibili.com/video/BV1L54y1271W
|
||||
|
||||
### 大表哥2
|
||||
分帧渲染
|
||||
https://zhuanlan.zhihu.com/p/127435500
|
||||
|
||||
安泊霖的分析：
|
||||
https://zhuanlan.zhihu.com/p/91359071
|
||||
|
||||
### 地平线：黎明时分
|
||||
在有道云里有pdf
|
||||
https://zhuanlan.zhihu.com/p/97257247
|
||||
https://zhuanlan.zhihu.com/p/57463381
|
||||
|
||||
### UE4云层阴影实现
|
||||
https://zhuanlan.zhihu.com/p/130785313
|
||||
|
||||
## 其他相关
|
||||
### GPU Gems 1
|
||||
https://developer.nvidia.com/gpugems/gpugems/foreword
|
||||
|
||||
http://http.download.nvidia.com/developer/GPU_Gems/CD_Image/Index.html
|
||||
|
||||
浅墨的总结：
|
||||
https://zhuanlan.zhihu.com/p/35974789
|
||||
https://zhuanlan.zhihu.com/p/36499291
|
||||
|
||||
真实感水体渲染（Realistic Water Rendering）
|
||||
无尽草地的渲染（Rendering Countless Blades of Waving Grass）
|
||||
水焦散渲染（Rendering Water Caustics）
|
||||
57
03-UnrealEngine/Rendering/Shader/Effect/皮皮虾材质方案.md
Normal file
57
03-UnrealEngine/Rendering/Shader/Effect/皮皮虾材质方案.md
Normal file
@@ -0,0 +1,57 @@
|
||||
---
|
||||
title: 皮皮虾材质方案
|
||||
date: 2022-12-06 16:06:06
|
||||
excerpt:
|
||||
tags:
|
||||
rating: ⭐
|
||||
---
|
||||
## 效果分析
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
由观察可得皮皮虾由外表面的透明甲壳材质与内部的肉材质组成。结构如下图：
|
||||
|
||||
解决方案：
|
||||
- 外层甲壳材质：
|
||||
- 透明度：以最薄处作为基础的透明度，叠加上 BumpOffset/POM 的HeightTexture的采样结果（深度）乘以参数计算出的透明度。
|
||||
- Roughness：很光滑光滑 0~0.2
|
||||
- Metallic：非金属（我们不需要反射）
|
||||
- 内部肉材质：
|
||||
- 贴图使用 BumpOffset/POM 进行UV偏移。
|
||||
- Roughness：粗糙 1。
|
||||
- Metallic：非金属
|
||||
- 中间层
|
||||
- 中间发光亮片：通过建模实现，内部放上一些小亮片，并且添加独立的自发光材质控制。
|
||||
- 中间雾气效果：通过建模实现，里面放上几个条带，添加类似雾气粒子的材质。
|
||||
|
||||
## 多层材质的距离感
|
||||
可以使用2种方案，主要都是通过偏移UV的方式来实现视差效果：
|
||||
1. BumpOffset
|
||||
2. POM
|
||||
|
||||
POM的会有较大的性能损耗，但与BumpOffset相比在视角较小时依然有着不错的效果。部分效果不佳的地方mask掉，防止穿帮。所以BumpOffset可以作为其下位代替。
|
||||
|
||||
|
||||
如果这2个方法不行就只能考虑使用2个模型，也就是一个外壳一个内部肉（不推荐）
|
||||
|
||||
## 具体方案
|
||||
### 两层模型
|
||||
- 外壳使用DefaultLit、Translucent
|
||||
- 内部使用Subsurface
|
||||
|
||||
缺点：内部模型需要蒙皮。
|
||||
|
||||
### MultiDraw
|
||||
MultiDraw在5.1支持不透明材质渲染透明效果了。所以可以是只用一个模型+HeightTexture配合MultiDraw二次绘制第二个材质效果。
|
||||
|
||||
缺点：有穿帮的危险。
|
||||
|
||||
### BumpOffset/POM
|
||||
使用DefaultLit或者车漆材质尝试 外壳/内部肉（视差），参考龙虾人的眼睛
|
||||
|
||||
1. 实现DefaultLit版本
|
||||
2. 查看车漆材质并且实现
|
||||
|
||||
缺点：5.1版本前无法做到材质物理上的正确。
|
||||
118
03-UnrealEngine/Rendering/Shader/SNN_Kuwahara Filter.md
Normal file
118
03-UnrealEngine/Rendering/Shader/SNN_Kuwahara Filter.md
Normal file
@@ -0,0 +1,118 @@
|
||||
---
|
||||
title: SNN/Kuwahara Filter
|
||||
date: 2022-10-19 13:44:01
|
||||
excerpt:
|
||||
tags: PostProcess
|
||||
rating: ⭐⭐
|
||||
---
|
||||
## 参考文章
|
||||
|
||||
## Kuwahara（算法发明者是日本人）
|
||||
Wiki：https://en.wikipedia.org/wiki/Kuwahara_filter
|
||||
各种Kuwahara算法：https://zhuanlan.zhihu.com/p/354523349
|
||||
|
||||
- Kuwahara：
|
||||
- 广义Kuwahara（Generalized Kuwahara filter）：
|
||||
- 各向异性Kuwahara（ Anisotropic Kuwahara filter）：https://www.shadertoy.com/view/ltyyDd
|
||||
- Adaptative Kuwahara filter：https://link.springer.com/article/10.1007/s11760-015-0791-3
|
||||
- Ganeralized Kuwahara：https://www.shadertoy.com/view/ltGcWD
|
||||
|
||||
## SNN（Symmetric nearest neighbour filter）
|
||||
Wiki：https://subsurfwiki.org/wiki/Symmetric_nearest_neighbour_filter
|
||||
|
||||
https://www.shadertoy.com/view/MlyfWd
|
||||
|
||||
|
||||
|
||||
|
||||
比较相反方向的像素对，并选择那些值最接近中心输入像素的像素（见图）。然后它返回所选像素的平均值。通过这种方式，它可以保留更多的边缘细节。
|
||||
|
||||
|
||||
|
||||
```c++
|
||||
vec3 CalcSNN(in vec2 fragCoord) {
|
||||
vec2 src_size = iResolution.xy;
|
||||
vec2 inv_src_size = 1.0f / src_size;
|
||||
vec2 uv = fragCoord * inv_src_size;
|
||||
|
||||
vec3 c0 = texture(iChannel0, uv).rgb;
|
||||
|
||||
vec4 sum = vec4(0.0f, 0.0f, 0.0f, 0.0f);
|
||||
|
||||
//{(1,0)~(5,0)}
|
||||
for (int i = 1; i <= half_width; ++i) {
|
||||
vec3 c1 = texture(iChannel0, uv + vec2(+i, 0) * inv_src_size).rgb;
|
||||
vec3 c2 = texture(iChannel0, uv + vec2(-i, 0) * inv_src_size).rgb;
|
||||
|
||||
float d1 = CalcDistance(c1, c0);
|
||||
float d2 = CalcDistance(c2, c0);
|
||||
if (d1 < d2) {
|
||||
sum.rgb += c1;
|
||||
} else {
|
||||
sum.rgb += c2;
|
||||
}
|
||||
sum.a += 1.0f;
|
||||
}
|
||||
|
||||
//{(-5,1)~(5,5)}
|
||||
for (int j = 1; j <= half_width; ++j) {
|
||||
for (int i = -half_width; i <= half_width; ++i) {
|
||||
vec3 c1 = texture(iChannel0, uv + vec2(+i, +j) * inv_src_size).rgb;
|
||||
vec3 c2 = texture(iChannel0, uv + vec2(-i, -j) * inv_src_size).rgb;
|
||||
|
||||
float d1 = CalcDistance(c1, c0);
|
||||
float d2 = CalcDistance(c2, c0);
|
||||
if (d1 < d2) {
|
||||
sum.rgb += c1;
|
||||
} else {
|
||||
sum.rgb += c2;
|
||||
}
|
||||
sum.a += 1.0f;
|
||||
}
|
||||
}
|
||||
|
||||
return sum.rgb / sum.a;
|
||||
}
|
||||
|
||||
```
|
||||
|
||||
### 双边滤波
|
||||
https://www.shadertoy.com/view/4dfGDH
|
||||
https://zhuanlan.zhihu.com/p/180497579
|
||||
高斯是基于距离的权重，双边是这个基础上增加了颜色权重。
|
||||
|
||||
### 破晓传说后处理
|
||||
|
||||
BasePass中。角色会渲染出奇怪的黄色。
|
||||
|
||||
- R通道为1
|
||||
- G通道为猜测为灯光光照结果，因为能看得出场景的光照。
|
||||
- B通道为猜测为天光+AO（很不确定）
|
||||
- A通道为角色Mask
|
||||
G、B概率是存了2种数据。但最后依然是在Lighting渲染出整个场景与角色的光照。
|
||||
|
||||
9650：之前是BasePass
|
||||
14046~14139：脸的SSS效果
|
||||
14146：眼睛
|
||||
14258~14296：角色头发
|
||||
14318~14420：
|
||||
14438~14589：
|
||||
14620~14848：渲染角色外的场景
|
||||
14868：开始渲染角色
|
||||
|
||||
光照渲染用了很多Ramp
|
||||
|
||||
15117~15159：锐化。智能锐化 1半径、130%强度效果接近，但可以看得出，角色暗部的锐化效果会更加明显。
|
||||
15182：SNN处理场景（不对角色进行处理）
|
||||
15202：描边。法线+深度描边。
|
||||
15219：描边合并。
|
||||
15240：提取暗部细节。
|
||||
|
||||
中间是粒子
|
||||
|
||||
17738：LUT 调整颜色
|
||||
17758：有点像计算 sobel 过滤u与v之后，`vec3 col = vec3(dot(u, u), dot(v, v), dot(u, v));`的结果。
|
||||
17776：感觉是重要步骤，但看不出是啥
|
||||
17790：边缘模糊
|
||||
|
||||
绘制UI
|
||||
@@ -0,0 +1,71 @@
|
||||
---
|
||||
title: ScreenPercentage
|
||||
date: 2022-11-09 10:32:14
|
||||
excerpt:
|
||||
tags:
|
||||
rating: ⭐
|
||||
---
|
||||
|
||||
## ScreenPercentage
|
||||
基本都与后处理盒子的ScreenPercentage相关。
|
||||
- ScreenPercentage：Scene.h
|
||||
- ScreenPercentageInterface：SceneView.h
|
||||
- ScreenPercentage：ShowFlag.h
|
||||
- ScreenPercentage：TemporalAA.cpp，FTAAStandaloneCS::FParameters
|
||||
- ScreenPercentage：PlanarReflectionComponent.h
|
||||
最后用于TAA上采样计算。
|
||||
|
||||
## DPI
|
||||
使用`UserInterfaceSettings->GetDPIScaleBasedOnSize(ViewportSize);`作为采样步长的因子即可。
|
||||
```c++
|
||||
float SGameLayerManager::GetGameViewportDPIScale() const
|
||||
{
|
||||
const FSceneViewport* Viewport = SceneViewport.Get();
|
||||
|
||||
if (Viewport == nullptr)
|
||||
{ return 1;
|
||||
}
|
||||
const auto UserInterfaceSettings = GetDefault<UUserInterfaceSettings>(UUserInterfaceSettings::StaticClass());
|
||||
|
||||
if (UserInterfaceSettings == nullptr)
|
||||
{ return 1;
|
||||
}
|
||||
FIntPoint ViewportSize = Viewport->GetSize();
|
||||
float GameUIScale;
|
||||
|
||||
if (bUseScaledDPI)
|
||||
{ float DPIValue = UserInterfaceSettings->GetDPIScaleBasedOnSize(ScaledDPIViewportReference);
|
||||
float ViewportScale = FMath::Min((float)ViewportSize.X / (float)ScaledDPIViewportReference.X, (float)ViewportSize.Y / (float)ScaledDPIViewportReference.Y);
|
||||
|
||||
GameUIScale = DPIValue * ViewportScale;
|
||||
} else
|
||||
{
|
||||
GameUIScale = UserInterfaceSettings->GetDPIScaleBasedOnSize(ViewportSize);
|
||||
}
|
||||
// Remove the platform DPI scale from the incoming size. Since the platform DPI is already
|
||||
// attempt to normalize the UI for a high DPI, and the DPI scale curve is based on raw resolution // for what a assumed platform scale of 1, extract that scale the calculated scale, since that will // already be applied by slate. const float FinalUIScale = GameUIScale / Viewport->GetCachedGeometry().Scale;
|
||||
|
||||
return FinalUIScale;
|
||||
}
|
||||
```
|
||||
|
||||
```c++
|
||||
// this function requires the UserInterfaceSettings header to be included
|
||||
#include Runtime/Engine/Classes/Engine/UserInterfaceSettings.h
|
||||
|
||||
// this function can be marked as Blueprint Pure in its declaration, as it simply returns a float
|
||||
float MyBPFL::GetUMG_DPI_Scale() {
|
||||
// need a variable here to pass to the GetViewportSize function
|
||||
FVector2D viewportSize;
|
||||
// as this function returns through the parameter, we just need to call it by passing in our FVector2D variable
|
||||
GEngine->GameViewport->GetViewportSize(viewportSize);
|
||||
|
||||
// we need to floor the float values of the viewport size so we can pass those into the GetDPIScaleBasedOnSize function
|
||||
int32 X = FGenericPlatformMath::FloorToInt(viewportSize.X);
|
||||
int32 Y = FGenericPlatformMath::FloorToInt(viewportSize.Y);
|
||||
|
||||
// the GetDPIScaleBasedOnSize function takes an FIntPoint, so we construct one out of the floored floats of the viewport
|
||||
// the fuction returns a float, so we can return the value out of our function here
|
||||
return GetDefault<UUserInterfaceSettings>(UUserInterfaceSettings::StaticClass())->GetDPIScaleBasedOnSize(FIntPoint(X,Y));
|
||||
}
|
||||
```
|
||||
113
03-UnrealEngine/Rendering/Shader/UE GBuffer存储数据.md
Normal file
113
03-UnrealEngine/Rendering/Shader/UE GBuffer存储数据.md
Normal file
@@ -0,0 +1,113 @@
|
||||
---
|
||||
title: UE GBuffer存储数据
|
||||
date: 2022-10-11 13:07:27
|
||||
excerpt:
|
||||
tags: Rendering
|
||||
rating: ⭐
|
||||
---
|
||||
|
||||
https://zhuanlan.zhihu.com/p/400677108
|
||||
| MRT | R | G | B | A |
|
||||
| -------------------------------- | -------------------------- | -------------------------- | -------------------------- | ------------------------------------- |
|
||||
| MRT0 | SceneColor.r | SceneColor.g | SceneColor.b | Alpha(Opacity?) |
|
||||
| MRT1 GBufferA | WorldNormal.r | WorldNormal.g | WorldNormal.b | PerObjectGBufferData |
|
||||
| MRT2 GBUfferB | Metallic | Specular | Roughness | ShadingModelId && SelectiveOutputMask |
|
||||
| MRT3 GBUfferC | BaseColor.r | BaseColor.g | BaseColor.b | IndirectIrradiance |
|
||||
| MRT4 GBUfferD(根据需求插入) | Velocity.r | Velocity.g | Velocity.b | Velocity.a |
|
||||
| MRT5 GBUfferE(Velocity不渲染为D) | CustomData.r | CustomData.g | CustomData.b | CustomData.a |
|
||||
| MRT6 GBUfferF(Velocity不渲染为E) | PrecomputedShadowFactors.r | PrecomputedShadowFactors.g | PrecomputedShadowFactors.b | PrecomputedShadowFactors.a |
|
||||
| MRT7 | | | | |
|
||||
|
||||
## Depth Stencil
|
||||
[0] sandbox bit (bit to be use by any rendering passes, but must be properly reset to 0 after using)
|
||||
STENCIL LOD抖动过渡，后处理，贴花，Local光源都有使用，用来绘制临时Mask
|
||||
**[1] unallocated**
|
||||
**[2] unallocated**
|
||||
[3] Temporal AA mask for translucent object.
|
||||
标记后TAA将不再读取History数据，用来处理半透物体的TAA问题
|
||||
|
||||
**1-3在移动延迟会被借用储存ShadingMode，但小于ShadingMode总数量，不清楚怎么处理的。但因为毕竟有重复使用，不适合直接占用这两个值。**
|
||||
[4] Lighting channels
|
||||
[5] Lighting channels
|
||||
[6] Lighting channels
|
||||
[7] Primitive receive decal bit
|
||||
|
||||
## GBuffer
|
||||
**SceneColorDeferred（场景颜色）**
|
||||
RGBAHalf（移动默认R11G11B10）
|
||||
Emissive直接写入这里，非自发光写入BaseColor。A通道正常储存Alpha值，通常情况没用。
|
||||
>**原神将其和BaseColor合并了，由Alpha来标记自发光强度。这使得自发光颜色必须和BaseColor相同，限制了其表达。在需要不同颜色的时候，原神则用叠加半透材质的方法实现。**
|
||||
**用这个方法需要给原本在BaseColor.a的IndirectIrradiance * AO 再找个位置。**
|
||||
|
||||
**GBufferA（法线）**
|
||||
R10G10B10A2
|
||||
移动延迟用八面体Encode法线到RG，B为PrecomputedShadowFactors.x。
|
||||
|
||||
PC直接储存法线到RGB（禁用的分支里可以Encode法线到RG，如果这样做了B可以空出一个10bit，但移动没法使用。不如和移动一样把PrecomputedShadowFactors.x移到这来。不过PrecomputedShadowFactors.x其实也不需要10bit，可以拆2bit做特殊用途）
|
||||
|
||||
A为PerObjectGBufferData（bit1:CastContactShadow，bit2:HasDynamicIndirectShadowCasterRepresentation）
|
||||
|
||||
**GBufferB**
|
||||
RGBA8888
|
||||
金属，高光，粗糙，ShadingModelID+SelectiveOutputMask（各占4bit，Shading Mode最大值16）
|
||||
|
||||
SelectiveOutputMask记录了绘制时以下宏的开启结果：
|
||||
- MATERIAL_USES_ANISOTROPY 禁止计算各向异性
|
||||
- !GBUFFER_HAS_PRECSHADOWFACTOR 禁止读取GBufferE数据作为预计算阴影
|
||||
- GBUFFER_HAS_PRECSHADOWFACTOR && WRITES_PRECSHADOWFACTOR_ZERO 当不读取GBufferE时，若此值为1时，预计算阴影设为0，否则为1。
|
||||
- WRITES_VELOCITY_TO_GBUFFER 禁止从Gbuffer中读取速度值。
|
||||
|
||||
**因为没写速度的时候Buffer里是0，并不是零速度，所以必须靠它来Mask，但如果将速度图Clear为零速度，按说可以不需要这个Mask，就可以用在别处**
|
||||
**并不觉得高光值得占用8bit的容量，可以压缩其他数据到高光上。**
|
||||
> 常见的做法还有将F0储存至RGB，让SSR无需采样BaseColor，但这样一来粗糙度又需要另找地方储存（可储存在法线B通道上）
|
||||
|
||||
**GBufferC（BaseColor）**
|
||||
RGBA8888
|
||||
RGB储存Lit物体的BaseColor，A在有静态光照时候储存随机抖动过的IndirectIrradiance * Material AO，否则直接储存Material AO。
|
||||
|
||||
**GBufferD（自定义数据）**
|
||||
|
||||
**GBufferE（静态阴影）**
|
||||
|
||||
**GBufferVelocity（速度）**
|
||||
|
||||
|
||||
## 实用函数
|
||||
LightingPass中的调用：
|
||||
```c++
|
||||
uint PixelShadingModelID = GetScreenSpaceData(ScreenUV).GBuffer.ShadingModelID;
|
||||
```
|
||||
最终：
|
||||
```c++
|
||||
FGBufferData GetGBufferData(float2 UV, bool bGetNormalizedNormal = true)
|
||||
{
|
||||
float4 GBufferA = Texture2DSampleLevel(SceneTexturesStruct.GBufferATexture, SceneTexturesStruct_GBufferATextureSampler, UV, 0);
|
||||
float4 GBufferB = Texture2DSampleLevel(SceneTexturesStruct.GBufferBTexture, SceneTexturesStruct_GBufferBTextureSampler, UV, 0);
|
||||
float4 GBufferC = Texture2DSampleLevel(SceneTexturesStruct.GBufferCTexture, SceneTexturesStruct_GBufferCTextureSampler, UV, 0);
|
||||
float4 GBufferD = Texture2DSampleLevel(SceneTexturesStruct.GBufferDTexture, SceneTexturesStruct_GBufferDTextureSampler, UV, 0);
|
||||
float4 ToonDataA = Texture2DSampleLevel(SceneTexturesStruct.ToonBufferATexture, SceneTexturesStruct_ToonBufferATextureSampler, UV, 0);
|
||||
float CustomNativeDepth = Texture2DSampleLevel(SceneTexturesStruct.CustomDepthTexture, SceneTexturesStruct_CustomDepthTextureSampler, UV, 0).r;
|
||||
|
||||
int2 IntUV = (int2)trunc(UV * View.BufferSizeAndInvSize.xy);
|
||||
uint CustomStencil = SceneTexturesStruct.CustomStencilTexture.Load(int3(IntUV, 0)) STENCIL_COMPONENT_SWIZZLE;
|
||||
|
||||
#if ALLOW_STATIC_LIGHTING
|
||||
float4 GBufferE = Texture2DSampleLevel(SceneTexturesStruct.GBufferETexture, SceneTexturesStruct_GBufferETextureSampler, UV, 0);
|
||||
#else
|
||||
float4 GBufferE = 1;
|
||||
#endif
|
||||
|
||||
float4 GBufferF = Texture2DSampleLevel(SceneTexturesStruct.GBufferFTexture, SceneTexturesStruct_GBufferFTextureSampler, UV, 0);
|
||||
|
||||
#if WRITES_VELOCITY_TO_GBUFFER
|
||||
float4 GBufferVelocity = Texture2DSampleLevel(SceneTexturesStruct.GBufferVelocityTexture, SceneTexturesStruct_GBufferVelocityTextureSampler, UV, 0);
|
||||
#else
|
||||
float4 GBufferVelocity = 0;
|
||||
#endif
|
||||
|
||||
float SceneDepth = CalcSceneDepth(UV);
|
||||
return DecodeGBufferData(GBufferA, GBufferB, GBufferC, GBufferD, GBufferE, GBufferF, GBufferVelocity, ToonDataA, CustomNativeDepth, CustomStencil, SceneDepth, bGetNormalizedNormal, CheckerFromSceneColorUV(UV));
|
||||
}
|
||||
```
|
||||
|
||||
>GBuffer.ShadingModelID = DecodeShadingModelId(InGBufferB.a);
|
||||
83
03-UnrealEngine/Rendering/Shader/UE4.25传统渲染方式的代码迁移指南.md
Normal file
83
03-UnrealEngine/Rendering/Shader/UE4.25传统渲染方式的代码迁移指南.md
Normal file
@@ -0,0 +1,83 @@
|
||||
---
|
||||
title: UE4.25传统渲染方式的代码迁移指南
|
||||
date: 2020-11-10 13:55:15
|
||||
tags: Shader
|
||||
rating: ⭐️
|
||||
---
|
||||
## 前言
|
||||
4.25的GlobalShader的调用代码发生了一下变化，我在写完RDG例子后顺便写了一下GlobalShader的使用案例。具体的可以参考我的插件：
|
||||
|
||||
[https://github.com/blueroseslol/BRPlugins](https://github.com/blueroseslol/BRPlugins)
|
||||
|
||||
代码位于SimpleGlobalShader.cpp中。
|
||||
|
||||
## 变量使用LAYOUT_FIELD宏进行包裹
|
||||
对于新定义用于设置变量的GlobalShader基类使用，需要使用**DECLARE_INLINE_TYPE_LAYOUT**(FYourShaderName,NonVirtual)宏进行声明。另外还有**DECLARE_TYPE_LAYOUT**(FYourShaderName,NonVirtual)宏，不太清楚差别，但作用应该是一样的。
|
||||
|
||||
除此之外还有
|
||||
```
|
||||
class FSimpleGlobalShader : public FGlobalShader
|
||||
{
|
||||
DECLARE_INLINE_TYPE_LAYOUT(FSimpleGlobalShader, NonVirtual);
|
||||
public:
|
||||
static bool ShouldCompilePermutation(const FGlobalShaderPermutationParameters& Parameters)
|
||||
{
|
||||
return IsFeatureLevelSupported(Parameters.Platform, ERHIFeatureLevel::SM5);
|
||||
}
|
||||
|
||||
static void ModifyCompilationEnvironment(const FGlobalShaderPermutationParameters& Parameters, FShaderCompilerEnvironment& OutEnvironment)
|
||||
{
|
||||
FGlobalShader::ModifyCompilationEnvironment(Parameters, OutEnvironment);
|
||||
OutEnvironment.SetDefine(TEXT("TEST_MICRO"), 1);
|
||||
}
|
||||
|
||||
FSimpleGlobalShader(){}
|
||||
|
||||
FSimpleGlobalShader(const ShaderMetaType::CompiledShaderInitializerType& Initializer):FGlobalShader(Initializer)
|
||||
{
|
||||
SimpleColorVal.Bind(Initializer.ParameterMap, TEXT("SimpleColor"));
|
||||
TextureVal.Bind(Initializer.ParameterMap, TEXT("TextureVal"));
|
||||
TextureSampler.Bind(Initializer.ParameterMap, TEXT("TextureSampler"));
|
||||
}
|
||||
|
||||
template<typename TRHIShader>
|
||||
void SetParameters(FRHICommandList& RHICmdList,TRHIShader* ShaderRHI,const FLinearColor &MyColor, FTexture2DRHIRef InInputTexture)
|
||||
{
|
||||
SetShaderValue(RHICmdList, ShaderRHI, SimpleColorVal, MyColor);
|
||||
SetTextureParameter(RHICmdList, ShaderRHI, TextureVal, TextureSampler,TStaticSamplerState<SF_Trilinear,AM_Clamp,AM_Clamp,AM_Clamp>::GetRHI(), InInputTexture);
|
||||
}
|
||||
|
||||
private:
|
||||
LAYOUT_FIELD(FShaderResourceParameter, TextureVal);
|
||||
LAYOUT_FIELD(FShaderResourceParameter, TextureSampler);
|
||||
LAYOUT_FIELD(FShaderParameter, SimpleColorVal);
|
||||
};
|
||||
```
|
||||
|
||||
第二个形参视类内是否有虚函数，选择Abstract或者NonVirtual。源代码里基本上都是使用NonVirtual的。
|
||||
|
||||
另一个不常用的宏是**DECLARE_EXPORTED_TYPE_LAYOUT**(FYourShaderName, YOUR_API, Virtual)，它用于声明需要被其他类继承的基类。具体的使用方法还请参照UE4源码。
|
||||
|
||||
Shader变量需要使用LAYOUT_FIELD宏进行包裹，第一个形参为类型，普通类型声明为FShaderParameter，Texture2d与Sampler声明为FShaderResourceParameter。例如：
|
||||
```
|
||||
LAYOUT_FIELD(FShaderParameter, SimpleColorVal);
|
||||
LAYOUT_FIELD(FShaderResourceParameter, TextureVal);
|
||||
LAYOUT_FIELD(FShaderResourceParameter, TextureSampler);
|
||||
```
|
||||
## Shader声明改变
|
||||
声明宏由DECLARE_SHADER_TYPE(FSimplePixelShader, Global)转变为DECLARE_GLOBAL_SHADER(FSimplePixelShader)。
|
||||
|
||||
## 去除序列化相关函数
|
||||
去掉virtual bool Serialize(FArchive& Ar) override函数。
|
||||
|
||||
LAYOUT_FIELD宏会实现这部分功能。
|
||||
|
||||
## FImageWriteTask
|
||||
FImageWriteTask现在接受TArray64的数据。而不是 TArray。如果你想传递TArray数据，可以这么写：
|
||||
```
|
||||
TArray<FColor> OutBMP;
|
||||
// Fill OutBMP somehow
|
||||
TUniquePtr<TImagePixelData<FColor>> PixelData = MakeUnique<TImagePixelData<FColor>>(DestSize);
|
||||
PixelData->Pixels = OutBMP;
|
||||
ImageTask->PixelData = MoveTemp(PixelData);
|
||||
```
|
||||
175
03-UnrealEngine/Rendering/Shader/UE的透明渲染功能笔记.md
Normal file
175
03-UnrealEngine/Rendering/Shader/UE的透明渲染功能笔记.md
Normal file
@@ -0,0 +1,175 @@
|
||||
---
|
||||
title: UE的透明渲染功能笔记
|
||||
date: 2022-10-30 21:36:08
|
||||
excerpt:
|
||||
tags: Rendering
|
||||
rating: ⭐
|
||||
---
|
||||
|
||||
其主函数为FDeferredShadingSceneRenderer::RenderTranslucency()，位于FXSystem与头发合成之后，渲染Distortion、速度之前。
|
||||
```c++
|
||||
// Draw translucency.
|
||||
if (bCanOverlayRayTracingOutput && TranslucencyViewsToRender != ETranslucencyView::None)
|
||||
{ RDG_CSV_STAT_EXCLUSIVE_SCOPE(GraphBuilder, RenderTranslucency);
|
||||
SCOPE_CYCLE_COUNTER(STAT_TranslucencyDrawTime);
|
||||
|
||||
// Raytracing doesn't need the distortion effect.
|
||||
const bool bShouldRenderDistortion = TranslucencyViewsToRender != ETranslucencyView::RayTracing;
|
||||
|
||||
#if RHI_RAYTRACING
|
||||
if (EnumHasAnyFlags(TranslucencyViewsToRender, ETranslucencyView::RayTracing))
|
||||
{ RenderRayTracingTranslucency(GraphBuilder, SceneColorTexture);
|
||||
EnumRemoveFlags(TranslucencyViewsToRender, ETranslucencyView::RayTracing);
|
||||
}#endif
|
||||
|
||||
// Render all remaining translucency views.
|
||||
AddSetCurrentStatPass(GraphBuilder, GET_STATID(STAT_CLM_Translucency));
|
||||
RenderTranslucency(GraphBuilder, SceneColorTexture, SceneDepthTexture, HairDatas, &SeparateTranslucencyTextures, TranslucencyViewsToRender);
|
||||
AddServiceLocalQueuePass(GraphBuilder);
|
||||
TranslucencyViewsToRender = ETranslucencyView::None;
|
||||
...
|
||||
}
|
||||
```
|
||||
|
||||
## Pass Event String
|
||||
static const TCHAR* TranslucencyPassToString(ETranslucencyPass::Type TranslucencyPass)
|
||||
{
|
||||
switch (TranslucencyPass)
|
||||
{ case ETranslucencyPass::TPT_StandardTranslucency:
|
||||
return TEXT("Standard");
|
||||
case ETranslucencyPass::TPT_TranslucencyAfterDOF:
|
||||
return TEXT("AfterDOF");
|
||||
case ETranslucencyPass::TPT_TranslucencyAfterDOFModulate:
|
||||
return TEXT("AfterDOFModulate");
|
||||
case ETranslucencyPass::TPT_AllTranslucency:
|
||||
return TEXT("All");
|
||||
} checkNoEntry();
|
||||
return TEXT("");
|
||||
}
|
||||
|
||||
|
||||
##
|
||||
半透明渲染分为可分离与单Pass渲染。
|
||||
- 可分离渲染调用RenderViewTranslucencyInner()渲染内部，之后调用AddEndSeparateTranslucencyTimerPass()完成渲染。
|
||||
- 单Pass渲染完所有物体。调用RenderViewTranslucencyInner()渲染内部，之后调用AddEndSeparateTranslucencyTimerPass()完成渲染。
|
||||
|
||||
Shader变量使用FTranslucentBasePassParameters。
|
||||
MeshDrawPass名为。
|
||||
```c++
|
||||
EMeshPass::TranslucencyStandard
|
||||
EMeshPass::TranslucencyAfterDOF
|
||||
EMeshPass::TranslucencyAfterDOFModulate
|
||||
EMeshPass::TranslucencyAfterMotionBlur
|
||||
EMeshPass::TranslucencyAll
|
||||
```
|
||||
|
||||
最后在AddPass()调用RenderViewTranslucencyInner()进行实际MeshDraw渲染：
|
||||
```c++
|
||||
if (bRenderInParallel)
|
||||
{
|
||||
GraphBuilder.AddPass(
|
||||
RDG_EVENT_NAME("Translucency(%s Parallel) %dx%d",
|
||||
TranslucencyPassToString(TranslucencyPass),
|
||||
int32(View.ViewRect.Width() * ViewportScale),
|
||||
int32(View.ViewRect.Height() * ViewportScale)),
|
||||
PassParameters,
|
||||
ERDGPassFlags::Raster | ERDGPassFlags::SkipRenderPass,
|
||||
[&SceneRenderer, &View, PassParameters, ViewportScale, Viewport, TranslucencyPass](FRHICommandListImmediate& RHICmdList)
|
||||
{
|
||||
FRDGParallelCommandListSet ParallelCommandListSet(RHICmdList, GET_STATID(STAT_CLP_Translucency), SceneRenderer, View, FParallelCommandListBindings(PassParameters), ViewportScale);
|
||||
RenderViewTranslucencyInner(RHICmdList, SceneRenderer, View, Viewport, ViewportScale, TranslucencyPass, &ParallelCommandListSet, PassParameters->InstanceCullingDrawParams);
|
||||
});
|
||||
}
|
||||
else
|
||||
{
|
||||
GraphBuilder.AddPass(
|
||||
RDG_EVENT_NAME("Translucency(%s) %dx%d",
|
||||
TranslucencyPassToString(TranslucencyPass),
|
||||
int32(View.ViewRect.Width() * ViewportScale),
|
||||
int32(View.ViewRect.Height() * ViewportScale)),
|
||||
PassParameters,
|
||||
ERDGPassFlags::Raster,
|
||||
[&SceneRenderer, &View, ViewportScale, Viewport, TranslucencyPass, PassParameters](FRHICommandListImmediate& RHICmdList)
|
||||
{
|
||||
RenderViewTranslucencyInner(RHICmdList, SceneRenderer, View, Viewport, ViewportScale, TranslucencyPass, nullptr, PassParameters->InstanceCullingDrawParams);
|
||||
});
|
||||
}
|
||||
```
|
||||
最后使用FBasePassMeshProcessor。
|
||||
```c++
|
||||
DrawDynamicMeshPass(View, RHICmdList,
|
||||
[&View, &DrawRenderState, TranslucencyPass](FDynamicPassMeshDrawListContext* DynamicMeshPassContext)
|
||||
{
|
||||
FBasePassMeshProcessor PassMeshProcessor(
|
||||
View.Family->Scene->GetRenderScene(),
|
||||
View.GetFeatureLevel(),
|
||||
&View,
|
||||
DrawRenderState,
|
||||
DynamicMeshPassContext,
|
||||
FBasePassMeshProcessor::EFlags::CanUseDepthStencil,
|
||||
TranslucencyPass);
|
||||
|
||||
const uint64 DefaultBatchElementMask = ~0ull;
|
||||
|
||||
for (int32 MeshIndex = 0; MeshIndex < View.ViewMeshElements.Num(); MeshIndex++)
|
||||
{ const FMeshBatch& MeshBatch = View.ViewMeshElements[MeshIndex];
|
||||
PassMeshProcessor.AddMeshBatch(MeshBatch, DefaultBatchElementMask, nullptr);
|
||||
}});
|
||||
```
|
||||
|
||||
## 透明物体判断
|
||||
IsTranslucentBlendMode()位于MaterialShared.h：
|
||||
```c++
|
||||
inline bool IsTranslucentBlendMode(enum EBlendMode BlendMode)
|
||||
{
|
||||
return BlendMode != BLEND_Opaque && BlendMode != BLEND_Masked;
|
||||
}
|
||||
```
|
||||
位于`bool FBasePassMeshProcessor::ShouldDraw(const FMaterial& Material)`
|
||||
```c++
|
||||
const bool bIsTranslucent = IsTranslucentBlendMode(BlendMode);
|
||||
if (bTranslucentBasePass)
|
||||
{
|
||||
if (bIsTranslucent && !Material.IsDeferredDecal())
|
||||
{
|
||||
switch (TranslucencyPassType)
|
||||
{
|
||||
case ETranslucencyPass::TPT_StandardTranslucency:
|
||||
bShouldDraw = !Material.IsTranslucencyAfterDOFEnabled();
|
||||
break;
|
||||
|
||||
case ETranslucencyPass::TPT_TranslucencyAfterDOF:
|
||||
bShouldDraw = Material.IsTranslucencyAfterDOFEnabled();
|
||||
break;
|
||||
|
||||
// only dual blended or modulate surfaces need background modulation
|
||||
case ETranslucencyPass::TPT_TranslucencyAfterDOFModulate:
|
||||
bShouldDraw = Material.IsTranslucencyAfterDOFEnabled() && (Material.IsDualBlendingEnabled(GetFeatureLevelShaderPlatform(FeatureLevel)) || BlendMode == BLEND_Modulate);
|
||||
break;
|
||||
|
||||
case ETranslucencyPass::TPT_AllTranslucency:
|
||||
bShouldDraw = true;
|
||||
break;
|
||||
}
|
||||
}
|
||||
}
|
||||
```
|
||||
|
||||
## 为什么MultiDraw无法同时渲染透明与不透明材质
|
||||
首先透明会使用FBasePassMeshProcessor进行渲染（不同BasePass会有不同的设置），针对图元创建MeshProcessor时是以每个图元为单位进行的。
|
||||
|
||||
|
||||
```c++
|
||||
void FMaterialRenderProxy::UpdateUniformExpressionCacheIfNeeded(ERHIFeatureLevel::Type InFeatureLevel) const
|
||||
{
|
||||
if (!UniformExpressionCache[InFeatureLevel].bUpToDate)
|
||||
{ // Don't cache uniform expressions if an entirely different FMaterialRenderProxy is going to be used for rendering
|
||||
const FMaterial* Material = GetMaterialNoFallback(InFeatureLevel);
|
||||
if (Material)
|
||||
{ FMaterialRenderContext MaterialRenderContext(this, *Material, nullptr);
|
||||
MaterialRenderContext.bShowSelection = GIsEditor;
|
||||
EvaluateUniformExpressions(UniformExpressionCache[InFeatureLevel], MaterialRenderContext);
|
||||
}
|
||||
}
|
||||
}
|
||||
```
|
||||
18
03-UnrealEngine/Rendering/Shader/时间抗锯齿TemporalAA与上采样.md
Normal file
18
03-UnrealEngine/Rendering/Shader/时间抗锯齿TemporalAA与上采样.md
Normal file
@@ -0,0 +1,18 @@
|
||||
---
|
||||
title: 时间抗锯齿TemporalAA与上采样
|
||||
date: 2022-11-04 09:45:21
|
||||
excerpt:
|
||||
tags:
|
||||
rating: ⭐
|
||||
---
|
||||
## 网址
|
||||
- 文档：[https://docs.unrealengine.com/5.0/en-US/screen-percentage-with-temporal-upscale-in-unreal-engine/](https://docs.unrealengine.com/5.0/en-US/screen-percentage-with-temporal-upscale-in-unreal-engine/)
|
||||
- 论坛上介绍的一种方法：[https://forums.unrealengine.com/unreal-engine/feedback-for-epic/1821872-gen-5-temporal-anti-aliasing](https://forums.unrealengine.com/unreal-engine/feedback-for-epic/1821872-gen-5-temporal-anti-aliasing)
|
||||
|
||||
### 步骤
|
||||
UE5中大概率不适用了
|
||||
- r.temporalAA.Algorithm 1 //切换到gen5
|
||||
- r.temporalAA.Upsampling 1 //开启上采样效果
|
||||
- r.upscale.quality 5 //选择上采样算法5(最猛的一种)
|
||||
- r.tonemapper.sharpen 2 // 最后再加一点点后处理锐化抵消边缘模糊感
|
||||
- r.TemporalAA.R11G11B10History=1 //TAA缓存格式
|
||||
28
03-UnrealEngine/Rendering/Shader/给Pass传入SceneColor.md
Normal file
28
03-UnrealEngine/Rendering/Shader/给Pass传入SceneColor.md
Normal file
@@ -0,0 +1,28 @@
|
||||
---
|
||||
title: 给Pass传入SceneColor
|
||||
date: 2022-10-10 15:11:46
|
||||
excerpt:
|
||||
tags: Rendering
|
||||
rating: ⭐
|
||||
---
|
||||
|
||||
## 添加变量
|
||||
```c++
|
||||
SHADER_PARAMETER_RDG_TEXTURE(Texture2D, SceneColorTexture)
|
||||
SHADER_PARAMETER_SAMPLER(SamplerState, SceneColorTextureSampler)
|
||||
```
|
||||
传入变量
|
||||
```c++
|
||||
PassParameters->SceneColorTexture = SceneColorTexture.Target;
|
||||
PassParameters->SceneColorTextureSampler = TStaticSamplerState<>::GetRHI();
|
||||
```
|
||||
|
||||
## Shader
|
||||
```c++
|
||||
Texture2D SceneColorTexture;
|
||||
SamplerState SceneColorTextureSampler;
|
||||
```
|
||||
|
||||
```c++
|
||||
float4 SceneColor = SceneColorTexture.SampleLevel(SceneColorTextureSampler, SceneBufferUV, 0);
|
||||
```
|
||||
9
03-UnrealEngine/Rendering/Shader/薄膜干涉材质.md
Normal file
9
03-UnrealEngine/Rendering/Shader/薄膜干涉材质.md
Normal file
@@ -0,0 +1,9 @@
|
||||
---
|
||||
title: 薄膜干涉材质
|
||||
date: 2023-03-14 17:36:48
|
||||
excerpt:
|
||||
tags:
|
||||
rating: ⭐
|
||||
---
|
||||
|
||||
https://80.lv/articles/building-an-iridescence-shader-in-ue4/
|
||||
Reference in New Issue
Block a user