vault backup: 2023-12-08 18:00:40
This commit is contained in:
113
03-UnrealEngine/卡通渲染相关资料/渲染功能/ShaderModel/Material&BasePass.md
Normal file
113
03-UnrealEngine/卡通渲染相关资料/渲染功能/ShaderModel/Material&BasePass.md
Normal file
@@ -0,0 +1,113 @@
|
||||
---
|
||||
title: Material&BasePass
|
||||
date: 2023-12-08 17:34:58
|
||||
excerpt:
|
||||
tags:
|
||||
rating: ⭐
|
||||
---
|
||||
# GBuffer
|
||||
```c#
|
||||
OutGBufferA = WorldNormal/PerObjectGBufferData
|
||||
OutGBufferB = Metallic/Specular/Roughness/EncodeShadingModelIdAndSelectiveOutputMask(GBuffer.ShadingModelID, GBuffer.SelectiveOutputMask);
|
||||
OutGBufferC = BaseColor/GBufferAO
|
||||
OutGBufferD = GBuffer.CustomData;
|
||||
OutGBufferE = GBuffer.PrecomputedShadowFactors;
|
||||
```
|
||||
```c#
|
||||
GBufferB:Metallic/Specular/Roughness=>ToonHairMask OffsetShadowMask/SpcularMask/SpecularValue
|
||||
OutGBufferD = CustomData.xyzw=》ShaderColor.rgb/NoL
|
||||
OutGBufferE = GBuffer.PrecomputedShadowFactors.xyzw=》 /RimLightMask/DiffuseOffset/RimLightWidth
|
||||
OutGBufferF = velocity => OutlineWidth/OutlineID/OutlinePaint/OutlineZShift
|
||||
```
|
||||
|
||||
```
|
||||
| GBuffer | 表头 |
|
||||
| -------- | ------------------------------------------------------------------------------------- |
|
||||
| GBufferB | OffsetShadowMask SpcularMask SpecularValue EncodeShadingModelIdAndSelectiveOutputMask |
|
||||
| GBufferD | ShaderColor.rgb NoL |
|
||||
| GBufferE | |
|
||||
| GBufferF | ID |
|
||||
```
|
||||
|
||||
## BaseColor与ShadowColor
|
||||
- 原神里ShadowColor还会接收其他物体的阴影投射，没有自投影;蓝色协议可能也没有自投影。
|
||||
|
||||
BaseColor与ShadowColor的过渡需要Step、Feather、Offset等参数，可以直接制作一个HalfLambert的渐变贴图之后使用View传递。因为有多个贴图所以还需要ID贴图指定。但这样需要考虑一个问题：
|
||||
|
||||
- 一个物体上的同一个ID区域的BaseColor与ShadowColor是否都是一样的
|
||||
- 如果不一样就需要再传递一个ShadowColor.rgb到GBuffer里。
|
||||
- 不管如何手绘的补充暗部也是需要加到GBuffer中的
|
||||
|
||||
这决定传递到View里面的渐变贴图是彩色还是暗色
|
||||
|
||||
### 预计算贴图方案（构想）
|
||||
Toon渲染一般会使用HalfLambda。之后使用Feather、Step等参数对过渡边界进行调整
|
||||
使用 渐变贴图查表来实现 渐变、二阶化。以此代替羽化、step等参数。
|
||||
使用ID贴图指定，或者通过BaseColor值来查询？
|
||||
|
||||
## 高光
|
||||
- PBR高光（使用Roughness控制是否可行？是否需要传入GBuffer一个Mask贴图）
|
||||
- 自定义高光：高光贴图、高光颜色、参数化高光形状、多层高光
|
||||
|
||||
## 描边
|
||||
- 原神的描边好像是后处理
|
||||
- 蓝色协议
|
||||
![[08-Assets/Images/ImageBag/UrealEngineNPR/原神_描边.png]]
|
||||
![[08-Assets/Images/ImageBag/UrealEngineNPR/原神截图_描边.png]]
|
||||
|
||||
TODO：考虑使用顶点色来控制宽度，使用顶点色G
|
||||
|
||||
## 顶点色
|
||||
用于存储一些低精度数据，插值即可
|
||||
- R:
|
||||
- G:描边宽度
|
||||
- B:
|
||||
|
||||
蓝色协议的R:阴影区域标记 与 B:Ao，而罪恶装备使用贴图来传递信息。
|
||||
|
||||
## lightmap
|
||||
### 罪恶装备
|
||||
,G为阴影控（AO），R为高光强度参数，金属和光滑材质的部分设置的更大一些。B通道：用于照明控制。最大值为高光，反之，值越小高光越淡。
|
||||
|
||||
|
||||
https://zhuanlan.zhihu.com/p/360229590一文中介绍了崩坏3与原神的计算方式
|
||||
|
||||
崩坏3的LightMap计算方式：
|
||||
```c++
|
||||
half4 baseColor = SAMPLE_TEXTURE2D(_BaseMap, sampler_BaseMap, input.uv.xy);
|
||||
half4 LightMapColor = SAMPLE_TEXTURE2D(_LightMap, sampler_LightMap, input.uv.xy);
|
||||
half3 ShadowColor = baseColor.rgb * _ShadowMultColor.rgb;
|
||||
half3 DarkShadowColor = baseColor.rgb * _DarkShadowMultColor.rgb;
|
||||
|
||||
//如果SFactor = 0,ShallowShadowColor为一级阴影色,否则为BaseColor。
|
||||
float SWeight = (LightMapColor.g * input.color.r + input.lambert) * 0.5 + 1.125;
|
||||
float SFactor = floor(SWeight - _ShadowArea);
|
||||
half3 ShallowShadowColor = SFactor * baseColor.rgb + (1 - SFactor) * ShadowColor.rgb;
|
||||
```
|
||||
|
||||
二级阴影计算:
|
||||
```c++
|
||||
//如果SFactor = 0,DarkShadowColor为二级阴影色,否则为一级阴影色。
|
||||
SFactor = floor(SWeight - _DarkShadowArea);
|
||||
DarkShadowColor = SFactor * (_FixDarkShadow * ShadowColor + (1 - _FixDarkShadow) * ShallowShadowColor) + (1 - SFactor) * DarkShadowColor;
|
||||
|
||||
// 平滑阴影边缘
|
||||
half rampS = smoothstep(0, _ShadowSmooth, input.lambert - _ShadowArea);
|
||||
half rampDS = smoothstep(0, _DarkShadowSmooth, input.lambert - _DarkShadowArea);
|
||||
ShallowShadowColor.rgb = lerp(ShadowColor, baseColor.rgb, rampS);
|
||||
DarkShadowColor.rgb = lerp(DarkShadowColor.rgb, ShadowColor, rampDS);
|
||||
|
||||
//如果SFactor = 0,FinalColor为二级阴影，否则为一级阴影。
|
||||
SFactor = floor(LightMapColor.g * input.color.r + 0.9f);
|
||||
half4 FinalColor;
|
||||
FinalColor.rgb = SFactor * ShallowShadowColor + (1 - SFactor) * DarkShadowColor;
|
||||
```
|
||||
|
||||
|
||||
**罪恶装备**：
|
||||
对阴影判断阈值的偏移。（见前面着色部分，顶点AO+手绘修正）
|
||||
G : 轮廓线根据与相机的距离扩大多少的系数
|
||||
B : 等高线 Z 轴偏移值
|
||||
A : 轮廓厚度系数。0.5为标准，1为最大厚度，0为无等高线
|
||||
|
||||
### 米哈游
|
||||
123
03-UnrealEngine/卡通渲染相关资料/渲染功能/其他渲染功能/Toon眉毛渲染.md
Normal file
123
03-UnrealEngine/卡通渲染相关资料/渲染功能/其他渲染功能/Toon眉毛渲染.md
Normal file
@@ -0,0 +1,123 @@
|
||||
---
|
||||
title: Toon眉毛渲染
|
||||
date: 2023-12-08 17:48:07
|
||||
excerpt:
|
||||
tags:
|
||||
rating: ⭐
|
||||
---
|
||||
## 有关眉毛、表情需要使用 模板功能
|
||||
UTS使用模板
|
||||
![[08-Assets/Images/ImageBag/UrealEngineNPR/UTS表情.png]]
|
||||
![[08-Assets/Images/ImageBag/UrealEngineNPR/UTS表情_StencilOut.png]]
|
||||
![[08-Assets/Images/ImageBag/UrealEngineNPR/UTS表情_StencilMask.png]]
|
||||
|
||||
```
|
||||
Stencil
|
||||
{
|
||||
Ref[_StencilNo] //设置渲染的模板缓存值，0~255
|
||||
Comp[_StencilComp] //模板测试的通过条件，有除了equal，还有Greater、Less、Always、Never等，类似ZTest。
|
||||
Pass[_StencilOpPass] //表示通过模板测试和Z测试（注意是都通过）的像素，怎么处置它的模板值。
|
||||
Fail[_StencilOpFail] //表示通过了模板测试但没通过Z测试的像素，怎么处置它的模板值。
|
||||
}
|
||||
|
||||
_UTS_StencilMode mode = (_UTS_StencilMode)(material.GetInt(ShaderPropStencilMode));
|
||||
switch (mode)
|
||||
{
|
||||
case _UTS_StencilMode.Off:
|
||||
// material.SetInt(ShaderPropStencilNo,0);
|
||||
material.SetInt(ShaderPropStencilComp, (int)_StencilCompFunction.Disabled);
|
||||
material.SetInt(ShaderPropStencilOpPass, (int)_StencilOperation.Keep);
|
||||
material.SetInt(ShaderPropStencilOpFail, (int)_StencilOperation.Keep);
|
||||
break;
|
||||
case _UTS_StencilMode.StencilMask:
|
||||
// material.SetInt(ShaderPropStencilNo,0);
|
||||
material.SetInt(ShaderPropStencilComp, (int)_StencilCompFunction.Always);
|
||||
material.SetInt(ShaderPropStencilOpPass, (int)_StencilOperation.Replace);
|
||||
material.SetInt(ShaderPropStencilOpFail, (int)_StencilOperation.Replace);
|
||||
break;
|
||||
case _UTS_StencilMode.StencilOut:
|
||||
// material.SetInt(ShaderPropStencilNo,0);
|
||||
material.SetInt(ShaderPropStencilComp, (int)_StencilCompFunction.NotEqual);
|
||||
material.SetInt(ShaderPropStencilOpPass, (int)_StencilOperation.Keep);
|
||||
material.SetInt(ShaderPropStencilOpFail, (int)_StencilOperation.Keep);
|
||||
|
||||
break;
|
||||
}
|
||||
```
|
||||
|
||||
七大罪中使用使用了深度测试 Greater Equal.(默认是 less Equal)。但这个方式可能在UE4里不太行，因为UE4的深度测试是全局的？
|
||||
|
||||
这个应该需要创建一个MeshProcessor来实现：
|
||||
```c#
|
||||
FSingleLayerWaterPassMeshProcessor::FSingleLayerWaterPassMeshProcessor(const FScene* Scene, const FSceneView* InViewIfDynamicMeshCommand, const FMeshPassProcessorRenderState& InPassDrawRenderState, FMeshPassDrawListContext* InDrawListContext)
|
||||
: FMeshPassProcessor(Scene, Scene->GetFeatureLevel(), InViewIfDynamicMeshCommand, InDrawListContext)
|
||||
, PassDrawRenderState(InPassDrawRenderState)
|
||||
{
|
||||
if (SingleLayerWaterUsesSimpleShading(Scene->GetShaderPlatform()))
|
||||
{
|
||||
// Force non opaque, pre multiplied alpha, transparent blend mode because water is going to be blended against scene color (no distortion from texture scene color).
|
||||
FRHIBlendState* ForwardSimpleWaterBlendState = TStaticBlendState<CW_RGBA, BO_Add, BF_One, BF_InverseSourceAlpha>::GetRHI();
|
||||
PassDrawRenderState.SetBlendState(ForwardSimpleWaterBlendState);
|
||||
}
|
||||
}
|
||||
|
||||
//默认是CF_DepthNearOrEqual，做这个效果可能就要使用CF_DepthFartherOrEqual
|
||||
void SetupBasePassState(FExclusiveDepthStencil::Type BasePassDepthStencilAccess, const bool bShaderComplexity, FMeshPassProcessorRenderState& DrawRenderState)
|
||||
{
|
||||
DrawRenderState.SetDepthStencilAccess(BasePassDepthStencilAccess);
|
||||
if (bShaderComplexity)
|
||||
{
|
||||
// Additive blending when shader complexity viewmode is enabled.
|
||||
DrawRenderState.SetBlendState(TStaticBlendState<CW_RGBA, BO_Add, BF_One, BF_One, BO_Add, BF_Zero, BF_One>::GetRHI());
|
||||
// Disable depth writes as we have a full depth prepass.
|
||||
DrawRenderState.SetDepthStencilState(TStaticDepthStencilState<false, CF_DepthNearOrEqual>::GetRHI());
|
||||
}
|
||||
}
|
||||
```
|
||||
设置位置在于FXXXPassMeshProcessor::Process()中的SetDepthStencilStateForBasePass()中。需要bMaskedInEarlyPass为false，GetDepthStencilAccess为DepthRead。
|
||||
```c#
|
||||
const bool bMaskedInEarlyPass = (MaterialResource.IsMasked() || Mesh.bDitheredLODTransition) && MaskedInEarlyPass(GShaderPlatformForFeatureLevel[FeatureLevel]);
|
||||
|
||||
if (bEnableReceiveDecalOutput)
|
||||
{
|
||||
// Set stencil value for this draw call
|
||||
// This is effectively extending the GBuffer using the stencil bits
|
||||
const uint8 StencilValue = GET_STENCIL_BIT_MASK(RECEIVE_DECAL, PrimitiveSceneProxy ? !!PrimitiveSceneProxy->ReceivesDecals() : 0x00)
|
||||
| STENCIL_LIGHTING_CHANNELS_MASK(PrimitiveSceneProxy ? PrimitiveSceneProxy->GetLightingChannelStencilValue() : 0x00);
|
||||
|
||||
if (bMaskedInEarlyPass)
|
||||
{
|
||||
DrawRenderState.SetDepthStencilState(TStaticDepthStencilState<
|
||||
false, CF_Equal,
|
||||
true, CF_Always, SO_Keep, SO_Keep, SO_Replace,
|
||||
false, CF_Always, SO_Keep, SO_Keep, SO_Keep,
|
||||
0xFF, GET_STENCIL_BIT_MASK(RECEIVE_DECAL, 1) | STENCIL_LIGHTING_CHANNELS_MASK(0x7)
|
||||
>::GetRHI());
|
||||
DrawRenderState.SetStencilRef(StencilValue);
|
||||
}
|
||||
else if (DrawRenderState.GetDepthStencilAccess() & FExclusiveDepthStencil::DepthWrite)
|
||||
{
|
||||
DrawRenderState.SetDepthStencilState(TStaticDepthStencilState<
|
||||
true, CF_GreaterEqual,
|
||||
true, CF_Always, SO_Keep, SO_Keep, SO_Replace,
|
||||
false, CF_Always, SO_Keep, SO_Keep, SO_Keep,
|
||||
0xFF, GET_STENCIL_BIT_MASK(RECEIVE_DECAL, 1) | STENCIL_LIGHTING_CHANNELS_MASK(0x7)
|
||||
>::GetRHI());
|
||||
DrawRenderState.SetStencilRef(StencilValue);
|
||||
}
|
||||
else
|
||||
{
|
||||
DrawRenderState.SetDepthStencilState(TStaticDepthStencilState<
|
||||
false, CF_GreaterEqual,
|
||||
true, CF_Always, SO_Keep, SO_Keep, SO_Replace,
|
||||
false, CF_Always, SO_Keep, SO_Keep, SO_Keep,
|
||||
0xFF, GET_STENCIL_BIT_MASK(RECEIVE_DECAL, 1) | STENCIL_LIGHTING_CHANNELS_MASK(0x7)
|
||||
>::GetRHI());
|
||||
DrawRenderState.SetStencilRef(StencilValue);
|
||||
}
|
||||
}
|
||||
else if (bMaskedInEarlyPass)
|
||||
{
|
||||
DrawRenderState.SetDepthStencilState(TStaticDepthStencilState<false, CF_Equal>::GetRHI());
|
||||
}
|
||||
```
|
||||
202
03-UnrealEngine/卡通渲染相关资料/渲染功能/其他渲染功能/Toon眼睛渲染.md
Normal file
202
03-UnrealEngine/卡通渲染相关资料/渲染功能/其他渲染功能/Toon眼睛渲染.md
Normal file
@@ -0,0 +1,202 @@
|
||||
# 资料整理
|
||||
|
||||
## 凹模型
|
||||
直接贴图即可。
|
||||
折射
|
||||
模型高光
|
||||
|
||||
## 凸模型
|
||||
1. 使用视差贴图来控制瞳孔效果 + 折射模拟
|
||||
```
|
||||
float2 viewL = mul(viewW, (float3x2) worldInverse);
|
||||
float2 offset = height * viewL;
|
||||
offset.y = -offset.y;
|
||||
texcoord -= parallaxScale * offset;
|
||||
```
|
||||
|
||||
|
||||
2. Physically based refraction
|
||||
```c++
|
||||
// 角膜区域突起的模型
|
||||
// Alternatively, use a displacement map
|
||||
// height = max(-positionL.z – eyeIrisDepth, 0.0);
|
||||
|
||||
// 球形模型
|
||||
// Plot[Max[1.0 - 18.4 * r * r, 0.0], {r, 0, 0.3}]
|
||||
height = anteriorChamberDepth * saturate( 1.0 - 18.4 * radius * radius );
|
||||
|
||||
// refractedW
|
||||
float w = n * dot( normalW, viewW );
|
||||
float k = sqrt( 1.0 + ( w - n ) * ( w + n ) );
|
||||
float3 refractedW = ( w - k ) * normalW - n * viewW;
|
||||
|
||||
float cosAlpha = dot(frontNormalW, -refractedW);
|
||||
float dist = height / cosAlpha;
|
||||
float3 offsetW = dist * refractedW;
|
||||
float2 offsetL = mul(offsetW, (float3x2) worldInverse);
|
||||
texcoord += float2(mask, -mask) * offsetL;
|
||||
```
|
||||
|
||||
|
||||
代码中首先计算了height，即前房的高度，PPT中height有两种计算方式，分别对应两种眼睛的模型结构，对应的结构写在注释中了。
|
||||
然后计算了refracted，这个是rtr中快速拟合的计算方法，n是空气与介质折射率的比值，关于refracted的推论可以参考：
|
||||
YivanLee：虚幻4渲染编程(人物篇)【第三卷：Human Eye Rendering】
|
||||
[129 赞同 · 12 评论文章](https://zhuanlan.zhihu.com/p/151786832)
|
||||
|
||||
最后一段，先通过frontNormalW与refractedW的点积计算出α角的cos值（上图中的α应该是标识错误，α是-refractedW与frontNormalW的夹角）。然后已知height，通过比值可以计算出refractedW的模长dist。offsetW即为完整的refractedW向量。最后转换到本地空间，乘上眼睛的Mask，加到原本的UV上。
|
||||
之后就是使用偏转后的UV去采样贴图了。
|
||||
与视差相同，这里也是在本地与世界空间中进行的计算，同样会有轴向问题，主要是normalW、viewW和frontNormalW参与的计算，normal与view可以转换到切线空间计算，而frontNormalW代表的是模型向前的朝向，这个必须要指定，不过图方便的话，把frontNormalW改成切线空间法线也不是不可以。
|
||||
|
||||
### 多层复合模型
|
||||
樱花大战cedec2020分享：https://blog.ch-wind.com/cedec2020-new-sakura-wars-note/
|
||||
|
||||
>本作的眼睛分为三个部分，眼白的部分是一个内凹的形状，瞳孔的部分则分为了向内凹的部分和向外突出的部分。
|
||||
|
||||
|
||||
>瞳孔的高光叠加在其突出的半透明部分上。根据摄像角度的不同，各个部分的贴图会分开进行移动，使得在哪个角度高光都能处在一个刚好的位置。
|
||||
控制上，有针对高光上下左右的移动强度与控制移动范围的参数共同作用。
|
||||
|
||||
|
||||
>从左边开始，是作为基础颜色的Albedo，以及用于Mask瞳孔的Alpha贴图，用于在Albedo上进行叠加的spt贴图，以及两张瞳孔高光，以及反应环境的matcapture贴图。
|
||||
虽然很多动画风格的渲染中会省略掉瞳孔中的虹彩部分，但是本作为了提高角色靠近时的效果，进行了详细的绘制，同时为了体现环境的变化与matcap的贴图进行叠加。
|
||||
高光贴图有两张，分别使用不同的UV动画进行控制，用于表现眼睛的湿润感。虽然是很细微的操作，但是对于表现角色的感情非常的有用。
|
||||
|
||||
|
||||
SunnySideUp UnityChan
|
||||
|
||||
### 其他效果实现
|
||||
#### 眼睛高光效果
|
||||
1. 贴图高光。使用事先绘制的高光形状贴图，贴到最外面的。并且使用ViewDirection来控制。设定4个UV Coord, 根据 View=》眼睛的本地坐标系=》Normalize后的向量进行插值。
|
||||
2. ~~PBR的思路~~
|
||||
|
||||
#### Matcap反射效果
|
||||
Matcap材质+球形法线贴图
|
||||
```c++
|
||||
float3 NormalBlend_MatcapUV_Detail = viewNormal.rgb * float3(-1,-1,1);
|
||||
float3 NormalBlend_MatcapUV_Base = (mul( UNITY_MATRIX_V, float4(viewDirection,0)).rgb*float3(-1,-1,1)) + float3(0,0,1);
|
||||
float3 noSknewViewNormal = NormalBlend_MatcapUV_Base * dot(NormalBlend_MatcapUV_Base, NormalBlend_MatcapUV_Detail) / NormalBlend_MatcapUV_Base.b - NormalBlend_MatcapUV_Detail;
|
||||
float2 ViewNormalAsMatCapUV = noSknewViewNormal.rg * 0.5 + 0.5;
|
||||
```
|
||||
|
||||
#### 焦散效果
|
||||
>焦散的表现反倒简单了，直接画在眼睛贴图上都可以，考虑到卡通表达的自由性，焦散是否存在与焦散的形状都可以没有限制，只要好看就行。
|
||||
下图也是miHoYo的分享，可以简单的理解为直接贴张Mask上去，然后用光照方向和菲涅尔去影响强度变化。
|
||||
|
||||
|
||||
使用Mask贴图、NoL与菲尼尔来控制
|
||||
|
||||
# 其他参考整理
|
||||
![[ToonEye.png|400]]
|
||||
|
||||
![[ToonEyes.png|400]]
|
||||
|
||||
## PBR Refrection（有问题）
|
||||
```
|
||||
//input anteriorChamberDepth 贴图 使用HeightMap
|
||||
//input radius
|
||||
//input mask 贴图
|
||||
//input texcoord
|
||||
//input n 折射率比值
|
||||
//input frontNormalW 贴图float3(0,0,1) Local=>World
|
||||
|
||||
// 角膜区域突起的模型 height = max(-positionL.z – eyeIrisDepth, 0.0);
|
||||
// 球形模型 Plot[Max[1.0 - 18.4 * r * r, 0.0], {r, 0, 0.3}]
|
||||
float height = anteriorChamberDepth ;//* saturate( 1.0 - 18.4 * radius * radius );
|
||||
|
||||
float3 normalW=Parameters.WorldNormal;
|
||||
float3 viewW= mul(float3(0,0,1),(float3x3)View.ViewToTranslatedWorld);//CameraViewToTranslatedWorld
|
||||
|
||||
// refractedW
|
||||
float w = n * dot( normalW, viewW );
|
||||
float k = sqrt( 1.0 + ( w - n ) * ( w + n ) );
|
||||
float3 refractedW = ( w - k ) * normalW - n * viewW;
|
||||
refractedW=-normalize(refractedW);
|
||||
|
||||
//float3 frontNormalW=mul(float(0,0,1),float(3x3)GetPrimitiveData(Parameters.PrimitiveId).LocalToWorld) * -1;
|
||||
float cosAlpha = dot(frontNormalW, refractedW);
|
||||
float dist = height / cosAlpha;
|
||||
float3 offsetW = dist * refractedW;
|
||||
float2 offsetL = mul(offsetW, (float3x2)GetPrimitiveData(Parameters.PrimitiveId).WorldToLocal);
|
||||
texcoord += float2(mask, -mask) * offsetL;
|
||||
|
||||
return texcoord;
|
||||
```
|
||||
### 原始代码
|
||||
```
|
||||
// 角膜区域突起的模型
|
||||
// Alternatively, use a displacement map
|
||||
// height = max(-positionL.z – eyeIrisDepth, 0.0);
|
||||
|
||||
// 球形模型
|
||||
// Plot[Max[1.0 - 18.4 * r * r, 0.0], {r, 0, 0.3}]
|
||||
height = anteriorChamberDepth * saturate( 1.0 - 18.4 * radius * radius );
|
||||
|
||||
// refractedW
|
||||
float w = n * dot( normalW, viewW );
|
||||
float k = sqrt( 1.0 + ( w - n ) * ( w + n ) );
|
||||
float3 refractedW = ( w - k ) * normalW - n * viewW;
|
||||
|
||||
float cosAlpha = dot(frontNormalW, -refractedW);
|
||||
float dist = height / cosAlpha;
|
||||
float3 offsetW = dist * refractedW;
|
||||
float2 offsetL = mul(offsetW, (float3x2) worldInverse);
|
||||
texcoord += float2(mask, -mask) * offsetL;
|
||||
```
|
||||
## Matcap反射效果
|
||||
```
|
||||
float2 CalcMatcapUV(FMaterialPixelParameters Parameters,float3 normalTexture)
|
||||
{
|
||||
float3 ViewVector=Parameters.CameraVector * -1;
|
||||
float3 ViewSpaceRightVector=normalize(cross(ViewVector , mul( float3(0.0,1.0,0.0) , ResolvedView.ViewToTranslatedWorld )));
|
||||
float3 ViewSpaceUpVector=cross(ViewVector ,ViewSpaceRightVector);
|
||||
|
||||
float3x3 Matrix=float3x3(
|
||||
ViewSpaceRightVector.x,ViewSpaceUpVector.x,ViewVector.x,
|
||||
ViewSpaceRightVector.y,ViewSpaceUpVector.y,ViewVector.y,
|
||||
ViewSpaceRightVector.z,ViewSpaceUpVector.z,ViewVector.z
|
||||
);
|
||||
|
||||
float3 ZeroOneNormal=mul(normalize(Parameters.WorldNormal+normalTexture),Matrix)*0.5+0.5;
|
||||
return float2(ZeroOneNormal.x,ZeroOneNormal.y*-1);
|
||||
}
|
||||
```
|
||||
|
||||
### 这个不太行
|
||||
```
|
||||
//input normalTexture 法线贴图
|
||||
float3 TangentWS=Parameters.TangentToWorld[0];
|
||||
float3 NormalWS=Parameters.TangentToWorld[2];
|
||||
float3 BinormalWS=cross(NormalWS, TangentWS);
|
||||
|
||||
float3 worldNormal;
|
||||
worldNormal.x = dot(float3(TangentWS.x,BinormalWS.x,NormalWS.x), normalTexture);
|
||||
worldNormal.y = dot(float3(TangentWS.y,BinormalWS.y,NormalWS.y), normalTexture);
|
||||
worldNormal.z = dot(float3(TangentWS.z,BinormalWS.z,NormalWS.z), normalTexture);
|
||||
worldNormal = normalize(worldNormal);
|
||||
|
||||
float3 e = normalize(GetWorldPosition(Parameters) - ResolvedView.WorldCameraOrigin);
|
||||
float3 reflectVector = reflect(e, worldNormal);
|
||||
float3 reflectVectorVS = normalize(mul(reflectVector,ResolvedView.TranslatedWorldToView));
|
||||
float m = 2.82842712474619 * sqrt(reflectVectorVS.z + 1.0);
|
||||
float2 cap = reflectVectorVS.xy / m + 0.5;
|
||||
cap=cap*0.5 + 0.5;
|
||||
return cap;
|
||||
```
|
||||
|
||||
## 高光贴图位置调整
|
||||
```
|
||||
//input float2 LeftUp,LeftDown,RightUp,RightDown;
|
||||
|
||||
float3 viewW = mul(float3(0,0,1),(float3x3)View.ViewToTranslatedWorld);
|
||||
float3 viewL = mul(viewW, (float3x2)GetPrimitiveData(Parameters.PrimitiveId).WorldToLocal);
|
||||
float2 viewL2D = normalize(viewL.xy);
|
||||
|
||||
float2 Left=lerp(LeftUp,LeftDown,viewL2D.y);
|
||||
float2 Right=lerp(RightUp,RightDown,viewL2D.y);
|
||||
return lerp(Left,Right,viewL2D.x);
|
||||
```
|
||||
## 焦散
|
||||
- dot(Parameters.WorldNormal,View.DirectionalLightDirection);
|
||||
- Fresnel
|
||||
|
||||
乘以焦散Mask即可
|
||||
89
03-UnrealEngine/卡通渲染相关资料/渲染功能/描边/描边.md
Normal file
89
03-UnrealEngine/卡通渲染相关资料/渲染功能/描边/描边.md
Normal file
@@ -0,0 +1,89 @@
|
||||
---
|
||||
title: 描边
|
||||
date: 2023-12-08 16:49:45
|
||||
excerpt:
|
||||
tags:
|
||||
rating: ⭐
|
||||
---
|
||||
# 实现功能
|
||||
- 后处理描边
|
||||
- MeshDraw描边
|
||||
|
||||
# 杂项
|
||||
## 李兄实现Outline思路
|
||||
### Depth与Normal描边
|
||||
ToonOutlineMain()
|
||||
```c++
|
||||
float3x3 laplacianOperator = float3x3(-1, -1, -1,
|
||||
-1, 8, -1,
|
||||
float3x3 Gx = float3x3( -1, +0, +1,
|
||||
-1, -1, -1);
|
||||
-2, +0, +2,
|
||||
-1, +0, +1);
|
||||
float3x3 Gy = float3x3( +1, +2, +1,
|
||||
+0, +0, +0,
|
||||
-1, -2, -1);
|
||||
```
|
||||
使用GetPixelValue()取得Normal与Depth，之后使用拉普拉斯算子与Sobel算子进行边缘检测：
|
||||
```c++
|
||||
float4 fs0 = s0 * laplacianOperator[0][0];
|
||||
float4 fs1 = s1 * laplacianOperator[0][1];
|
||||
float4 fs2 = s2 * laplacianOperator[0][2];
|
||||
float4 fs3 = s3 * laplacianOperator[1][0];
|
||||
float4 fs4 = s4 * laplacianOperator[1][1];
|
||||
float4 fs5 = s5 * laplacianOperator[1][2];
|
||||
float4 fs6 = s6 * laplacianOperator[2][0];
|
||||
float4 fs7 = s7 * laplacianOperator[2][1];
|
||||
float4 fs8 = s8 * laplacianOperator[2][2];
|
||||
|
||||
float4 sampledValue = fs0 + fs1 + fs2 + fs3 + fs4 + fs5 + fs6 + fs7 + fs8;
|
||||
OutlineMask0 = saturate(1.0 - length(sampledValue)); //Line is black
|
||||
```
|
||||
```c++
|
||||
float4 ds0x = s0 * Gx[0][0];
|
||||
float4 ds1x = s1 * Gx[0][1];
|
||||
float4 ds2x = s2 * Gx[0][2];
|
||||
float4 ds3x = s3 * Gx[1][0];
|
||||
float4 ds4x = s4 * Gx[1][1];
|
||||
float4 ds5x = s5 * Gx[1][2];
|
||||
float4 ds6x = s6 * Gx[2][0];
|
||||
float4 ds7x = s7 * Gx[2][1];
|
||||
float4 ds8x = s8 * Gx[2][2];
|
||||
float4 SGX = ds0x + ds1x + ds2x + ds3x + ds4x + ds5x + ds6x + ds7x + ds8x;
|
||||
|
||||
float4 ds0y = s0 * Gy[0][0];
|
||||
float4 ds1y = s1 * Gy[0][1];
|
||||
float4 ds2y = s2 * Gy[0][2];
|
||||
float4 ds3y = s3 * Gy[1][0];
|
||||
float4 ds4y = s4 * Gy[1][1];
|
||||
float4 ds5y = s5 * Gy[1][2];
|
||||
float4 ds6y = s6 * Gy[2][0];
|
||||
float4 ds7y = s7 * Gy[2][1];
|
||||
float4 ds8y = s8 * Gy[2][2];
|
||||
float4 SGY = ds0y + ds1y + ds2y + ds3y + ds4y + ds5y + ds6y + ds7y + ds8y;
|
||||
|
||||
OutlineMask1 = saturate(2.0 - step(0.9, length(sqrt(SGX * SGX + SGY * SGY))));
|
||||
```
|
||||
这个算法巧妙的地方在于对计算卷积核之后使用length(float4(Normal,Depth))来取得结果；Sobel也是：length(sqrt(SGX * SGX + SGY * SGY)。
|
||||
|
||||
最后使用```OutColor.rgba = OutlineMask0;//lerp(OutlineMask0, OutlineMask1, 0.5);```进行混合。使用OutlineIDMap.a作为宽度控制项，并乘以通过Depth重映射后的变量作为宽度Fix因子。
|
||||
|
||||
### ID描边
|
||||
对ToonIDTexture进行Sobel描边。（只使用了Sobel）
|
||||
|
||||
### 混合结果
|
||||
|
||||
## 蓝色协议的做法
|
||||
### 模型外扩
|
||||
|
||||
### 后处理
|
||||
1. 使用Sobel算子进行深度检测，**只勾数值差别较大的区域**：脸部顶点色定义区域 与 模型外部轮廓。
|
||||
2. 使用Sobel进行ID贴图检测。**只勾数值差别较大的区域**。
|
||||
3. 进行法线点积（dot）检测。**在 深度差异小 以及 同一个ID区域内进行检测**：手指区域。
|
||||
|
||||
1. 使用Sobel过滤器进行深度检测描边。
|
||||
2. 使用Sobel过滤器进行Id图检测描边。
|
||||
3. 使用Sobel过滤器进行Normal检测描边。用于处理一些难以分ID，深度差又很小的地方，通过获取周围点法线求点乘的方式判断出轮廓。![[08-Assets/Images/ImageBag/UrealEngineNPR/蓝色协议_Normal检测描边.png)
|
||||
4. 预先画好的轮廓（GBuffer）。
|
||||
|
||||
所以使用需要 OutlineId、OutlineWidth（感觉可以传递一个全局Outline信息贴图再通过ID查表来获取，但只能在角色较少时使用）、OutlinePaint 、OutlineZShift（个人感觉不需要）
|
||||
@@ -1,114 +0,0 @@
|
||||
![[ToonEye.png]]
|
||||
|
||||
![[ToonEyes.png]]
|
||||
|
||||
## PBR Refrection（有问题）
|
||||
```
|
||||
//input anteriorChamberDepth 贴图 使用HeightMap
|
||||
//input radius
|
||||
//input mask 贴图
|
||||
//input texcoord
|
||||
//input n 折射率比值
|
||||
//input frontNormalW 贴图float3(0,0,1) Local=>World
|
||||
|
||||
// 角膜区域突起的模型 height = max(-positionL.z – eyeIrisDepth, 0.0);
|
||||
// 球形模型 Plot[Max[1.0 - 18.4 * r * r, 0.0], {r, 0, 0.3}]
|
||||
float height = anteriorChamberDepth ;//* saturate( 1.0 - 18.4 * radius * radius );
|
||||
|
||||
float3 normalW=Parameters.WorldNormal;
|
||||
float3 viewW= mul(float3(0,0,1),(float3x3)View.ViewToTranslatedWorld);//CameraViewToTranslatedWorld
|
||||
|
||||
// refractedW
|
||||
float w = n * dot( normalW, viewW );
|
||||
float k = sqrt( 1.0 + ( w - n ) * ( w + n ) );
|
||||
float3 refractedW = ( w - k ) * normalW - n * viewW;
|
||||
refractedW=-normalize(refractedW);
|
||||
|
||||
//float3 frontNormalW=mul(float(0,0,1),float(3x3)GetPrimitiveData(Parameters.PrimitiveId).LocalToWorld) * -1;
|
||||
float cosAlpha = dot(frontNormalW, refractedW);
|
||||
float dist = height / cosAlpha;
|
||||
float3 offsetW = dist * refractedW;
|
||||
float2 offsetL = mul(offsetW, (float3x2)GetPrimitiveData(Parameters.PrimitiveId).WorldToLocal);
|
||||
texcoord += float2(mask, -mask) * offsetL;
|
||||
|
||||
return texcoord;
|
||||
```
|
||||
### 原始代码
|
||||
```
|
||||
// 角膜区域突起的模型
|
||||
// Alternatively, use a displacement map
|
||||
// height = max(-positionL.z – eyeIrisDepth, 0.0);
|
||||
|
||||
// 球形模型
|
||||
// Plot[Max[1.0 - 18.4 * r * r, 0.0], {r, 0, 0.3}]
|
||||
height = anteriorChamberDepth * saturate( 1.0 - 18.4 * radius * radius );
|
||||
|
||||
// refractedW
|
||||
float w = n * dot( normalW, viewW );
|
||||
float k = sqrt( 1.0 + ( w - n ) * ( w + n ) );
|
||||
float3 refractedW = ( w - k ) * normalW - n * viewW;
|
||||
|
||||
float cosAlpha = dot(frontNormalW, -refractedW);
|
||||
float dist = height / cosAlpha;
|
||||
float3 offsetW = dist * refractedW;
|
||||
float2 offsetL = mul(offsetW, (float3x2) worldInverse);
|
||||
texcoord += float2(mask, -mask) * offsetL;
|
||||
```
|
||||
## Matcap反射效果
|
||||
```
|
||||
float2 CalcMatcapUV(FMaterialPixelParameters Parameters,float3 normalTexture)
|
||||
{
|
||||
float3 ViewVector=Parameters.CameraVector * -1;
|
||||
float3 ViewSpaceRightVector=normalize(cross(ViewVector , mul( float3(0.0,1.0,0.0) , ResolvedView.ViewToTranslatedWorld )));
|
||||
float3 ViewSpaceUpVector=cross(ViewVector ,ViewSpaceRightVector);
|
||||
|
||||
float3x3 Matrix=float3x3(
|
||||
ViewSpaceRightVector.x,ViewSpaceUpVector.x,ViewVector.x,
|
||||
ViewSpaceRightVector.y,ViewSpaceUpVector.y,ViewVector.y,
|
||||
ViewSpaceRightVector.z,ViewSpaceUpVector.z,ViewVector.z
|
||||
);
|
||||
|
||||
float3 ZeroOneNormal=mul(normalize(Parameters.WorldNormal+normalTexture),Matrix)*0.5+0.5;
|
||||
return float2(ZeroOneNormal.x,ZeroOneNormal.y*-1);
|
||||
}
|
||||
```
|
||||
|
||||
### 这个不太行
|
||||
```
|
||||
//input normalTexture 法线贴图
|
||||
float3 TangentWS=Parameters.TangentToWorld[0];
|
||||
float3 NormalWS=Parameters.TangentToWorld[2];
|
||||
float3 BinormalWS=cross(NormalWS, TangentWS);
|
||||
|
||||
float3 worldNormal;
|
||||
worldNormal.x = dot(float3(TangentWS.x,BinormalWS.x,NormalWS.x), normalTexture);
|
||||
worldNormal.y = dot(float3(TangentWS.y,BinormalWS.y,NormalWS.y), normalTexture);
|
||||
worldNormal.z = dot(float3(TangentWS.z,BinormalWS.z,NormalWS.z), normalTexture);
|
||||
worldNormal = normalize(worldNormal);
|
||||
|
||||
float3 e = normalize(GetWorldPosition(Parameters) - ResolvedView.WorldCameraOrigin);
|
||||
float3 reflectVector = reflect(e, worldNormal);
|
||||
float3 reflectVectorVS = normalize(mul(reflectVector,ResolvedView.TranslatedWorldToView));
|
||||
float m = 2.82842712474619 * sqrt(reflectVectorVS.z + 1.0);
|
||||
float2 cap = reflectVectorVS.xy / m + 0.5;
|
||||
cap=cap*0.5 + 0.5;
|
||||
return cap;
|
||||
```
|
||||
|
||||
## 高光贴图位置调整
|
||||
```
|
||||
//input float2 LeftUp,LeftDown,RightUp,RightDown;
|
||||
|
||||
float3 viewW = mul(float3(0,0,1),(float3x3)View.ViewToTranslatedWorld);
|
||||
float3 viewL = mul(viewW, (float3x2)GetPrimitiveData(Parameters.PrimitiveId).WorldToLocal);
|
||||
float2 viewL2D = normalize(viewL.xy);
|
||||
|
||||
float2 Left=lerp(LeftUp,LeftDown,viewL2D.y);
|
||||
float2 Right=lerp(RightUp,RightDown,viewL2D.y);
|
||||
return lerp(Left,Right,viewL2D.x);
|
||||
```
|
||||
## 焦散
|
||||
- dot(Parameters.WorldNormal,View.DirectionalLightDirection);
|
||||
- Fresnel
|
||||
|
||||
乘以焦散Mask即可
|
||||
Reference in New Issue
Block a user