Unity自定义渲染管线（Scriptable Render Pipeline）架构设计与实现指南

一、SRP技术体系概述

1. 核心设计理念

• 全托管渲染控制：通过C#脚本完全掌控渲染流程

• 模块化架构：将渲染流程拆分为可组合的RenderPass

• GPU友好设计：支持CommandBuffer与Compute Shader混合编程

• 对惹，这里有一个游戏开发交流小组，希望大家可以点击进来一起交流一下开发经验呀

2. 与传统管线对比

特性	内置管线	SRP
流程控制	黑盒模式	全脚本可编程
渲染策略	固定前向/延迟	自由组合多Pass
性能优化	有限批处理	深度控制资源生命周期
跨平台支持	自动适配	需手动实现变体

---

二、核心架构设计

1. 分层架构图

graph TBA[SRP Asset] -->|配置参数| B[RenderPipeline]B -->|调度| C[CameraRenderer]C -->|组织| D[RenderPass]D -->|执行| E[CommandBuffer]E -->|提交| F[GPU]

2. 核心组件职责

• SRP Asset：管线全局配置（质量等级、Shader变体等）

• RenderPipeline：主入口类，负责每帧调度

• CameraRenderer：单相机渲染流程控制器

• RenderPass：最小渲染单元（如阴影Pass、光照Pass）

---

三、基础管线实现

1. SRP Asset创建

[CreateAssetMenu(menuName = "Rendering/Custom RP")]
public class CustomRenderPipelineAsset : RenderPipelineAsset {public bool useDynamicBatching = true;public bool useGPUInstancing = true;protected override RenderPipeline CreatePipeline() {return new CustomRenderPipeline(this);}
}

2. 主渲染管线类

public class CustomRenderPipeline : RenderPipeline {private CameraRenderer renderer = new CameraRenderer();private bool useDynamicBatching;private bool useGPUInstancing;public CustomRenderPipeline(CustomRenderPipelineAsset asset) {useDynamicBatching = asset.useDynamicBatching;useGPUInstancing = asset.useGPUInstancing;}protected override void Render(ScriptableRenderContext context, Camera[] cameras) {foreach (var camera in cameras) {renderer.Render(context, camera, useDynamicBatching, useGPUInstancing);}}
}

3. 相机渲染器

public class CameraRenderer {static ShaderTagId unlitShaderTagId = new ShaderTagId("SRPDefaultUnlit");public void Render(ScriptableRenderContext context, Camera camera,bool useDynamicBatching,bool useGPUInstancing) {// 设置相机矩阵context.SetupCameraProperties(camera);// 清除渲染目标var cmd = new CommandBuffer { name = camera.name };cmd.ClearRenderTarget(true, true, Color.clear);context.ExecuteCommandBuffer(cmd);cmd.Clear();// 物体排序与绘制var sortingSettings = new SortingSettings(camera) {criteria = SortingCriteria.CommonOpaque};var drawingSettings = new DrawingSettings(unlitShaderTagId, sortingSettings) {enableDynamicBatching = useDynamicBatching,enableInstancing = useGPUInstancing};// 执行绘制var filterSettings = new FilteringSettings(RenderQueueRange.opaque);context.DrawRenderers(cullingResults, ref drawingSettings, ref filterSettings);context.Submit();}
}

---

四、高级渲染功能实现

1. 多Pass渲染架构

public class LightingPass : ScriptableRenderPass {public override void Configure(CommandBuffer cmd, RenderTextureDescriptor cameraTextureDescriptor) {// 创建临时RTcmd.GetTemporaryRT(Shader.PropertyToID("_LightBuffer"),1024, 1024, 24, FilterMode.Bilinear);}public override void Execute(ScriptableRenderContext context, ref RenderingData renderingData) {// 绘制光照几何体cmd.DrawMesh(lightMesh, matrix, lightMaterial);context.ExecuteCommandBuffer(cmd);}public override void FrameCleanup(CommandBuffer cmd) {// 释放临时RTcmd.ReleaseTemporaryRT(Shader.PropertyToID("_LightBuffer"));}
}

2. 延迟渲染路径实现

G-Buffer生成Shader

struct GBufferOutput {float4 albedo   : SV_Target0; // RGB:BaseColor, A:Smoothnessfloat4 normal   : SV_Target1; // RGB:WorldNormal, A:Metallicfloat4 emission : SV_Target2; // RGB:Emission, A:Occlusion
};GBufferOutput FragGBuffer(Varyings input) {GBufferOutput output;// 采样材质属性...output.albedo = float4(baseColor, smoothness);output.normal = float4(normalWS * 0.5 + 0.5, metallic);output.emission = float4(emission, occlusion);return output;
}

光照计算Pass

public class DeferredLightingPass : ScriptableRenderPass {private Material lightingMat;public override void Execute(...) {// 设置G-Buffer纹理cmd.SetGlobalTexture("_AlbedoBuffer", albedoRT);cmd.SetGlobalTexture("_NormalBuffer", normalRT);// 全屏四边形绘制光照cmd.DrawProcedural(Matrix4x4.identity, lightingMat, 0,MeshTopology.Triangles, 3, 1);}
}

---

五、性能优化技术

1. SRP Batcher优化

// Shader中声明PerObject数据
CBUFFER_START(UnityPerMaterial)
float4 _BaseColor;
float _Metallic;
CBUFFER_END// C#端启用SRP Batcher
GraphicsSettings.useScriptableRenderPipelineBatching = true;

2. GPU Driven Rendering

ComputeBuffer argsBuffer = new ComputeBuffer(5, sizeof(uint), ComputeBufferType.IndirectArguments
);// 通过Compute Shader生成绘制参数
computeShader.Dispatch(kernel, groupCount, 1, 1);// 间接绘制
cmd.DrawMeshInstancedIndirect(mesh, 0, material, bounds, argsBuffer
);

3. 多线程渲染支持

// 创建并行上下文
var renderThread = new RenderThread();
renderThread.Start();// 提交渲染任务
renderThread.Schedule(() => {context.SetupCameraProperties(camera);context.DrawRenderers(...);context.Submit();
});

---

六、实战案例：卡通渲染管线

1. 架构设计

功能模块	技术方案
轮廓描边	法线扩展+深度检测
色阶着色	颜色量化+LUT映射
高光处理	各向异性高光模型

2. 关键Shader代码

// 轮廓Pass
Pass {Cull FrontHLSLPROGRAMfloat _OutlineWidth;Varyings VertOutline(Varyings input) {float3 normal = input.normal * _OutlineWidth;input.positionOS += normal;return input;}ENDHLSL
}// 色阶处理
float3 QuantizeColor(float3 color) {float levels = 3.0;return floor(color * levels) / levels;
}

---

七、调试与优化工具

1. 帧调试器扩展

[MenuItem("Custom RP/Debug/Show Light Count")]
static void ToggleLightCountDebug() {Shader.EnableKeyword("_DEBUG_LIGHT_COUNT");
}

2. 性能分析标记

using (new ProfilingScope(cmd, new ProfilingSampler("Lighting Pass"))) 
{// 光照计算代码...
}

---

八、完整项目参考

Unlit渲染管线全流程详解

---

通过自定义渲染管线，开发者可突破Unity默认渲染的限制，实现从移动端到主机的全平台优化方案。核心在于：1) 合理设计RenderPass执行顺序；2) 利用CommandBuffer精细控制GPU资源；3) 结合Compute Shader实现GPU Driven架构。建议将常用渲染功能（如阴影、后处理）模块化封装，便于不同项目间复用。