[unity渲染管线]延迟渲染

想了想还是写下电子日志做备忘，之后陆续将以前做过的龙书的笔记也码过来好了，顺便复习。

准备工作

把camera的clear flag改为don’t clear；culling mask改为nothing。把color space改为linear。

延迟渲染

原理已经有很多大佬写的很清楚了，这里就不细讲了，Gbuffer摆放是参照虚幻的。

Gbuffer pass

延迟渲染的第一步是Gbuffer pass:
首先定义rendertexture的结构

void Start()
    {
        rt = new RenderTexture(Screen.width, Screen.height, 0, RenderTextureFormat.ARGB32, RenderTextureReadWrite.Linear);
        GBufferTexture = new RenderTexture[]
        {
            new RenderTexture(Screen.width,Screen.height,0,RenderTextureFormat.ARGB32,RenderTextureReadWrite.Linear),
            new RenderTexture(Screen.width,Screen.height,0,RenderTextureFormat.ARGB32,RenderTextureReadWrite.Linear),
            new RenderTexture(Screen.width,Screen.height,0,RenderTextureFormat.ARGB2101010,RenderTextureReadWrite.Linear),
            new RenderTexture(Screen.width,Screen.height,0,RenderTextureFormat.ARGB2101010,RenderTextureReadWrite.Linear)
         };
        depthTexture = new RenderTexture(Screen.width, Screen.height, 24, RenderTextureFormat.Depth, RenderTextureReadWrite.Linear);
        GBuffers = new RenderBuffer[GBufferTexture.Length];
        for (int i = 0; i < GBuffers.Length; i++)
        {
            GBuffers[i] = GBufferTexture[i].colorBuffer;
        }
        gbufferIDs = new int[] {
            Shader.PropertyToID("outGBuffer0"),
            Shader.PropertyToID("outGBuffer1"),
            Shader.PropertyToID("outGBuffer2"),
            Shader.PropertyToID("outEmission")
        };

    }

在shader里只需要简单地将当面帧的屏幕空间信息（漫反射颜色，金属度，法线等）写进多个rendertexture就行。

struct VertexInput
{
	float4 vertex   : POSITION;
	float2 uv0      : TEXCOORD0;
	float2 uv1      : TEXCOORD1;
	float2 uv2      : TEXCOORD2;
	float2 uv3      : TEXCOORD3;
	float4 tangent  : TANGENT;
	half3 normal    : NORMAL;
	
	
};


	    
	struct VertexOutputDeferred
	{
		float4 pos                            : SV_POSITION;
		float4 uv0                            : TEXCOORD0;
		float4 uv1                            : TEXCOORD1;
		float4 TtoW0                          : TEXCOORD2;
		float4 TtoW1                          : TEXCOORD3;
		float4 TtoW2                          : TEXCOORD4;
	};

	sampler2D _MainTex;
	float4 _MainTex_ST;
	sampler2D _BumpMap;
	float4 _BumpMap_ST;
	sampler2D _AORoughnessMetallic;
	float4 _AORoughnessMetallic_ST;
	sampler2D  _EmissionMap;
	float4 _EmissionMap_ST;

	VertexOutputDeferred vertDeferred(VertexInput v)
	{
		
		VertexOutputDeferred o;
		o.uv0.xy = TRANSFORM_TEX(v.uv0, _MainTex);
		o.uv0.zw = TRANSFORM_TEX(v.uv1, _BumpMap);
		o.uv1.xy = TRANSFORM_TEX(v.uv2, _AORoughnessMetallic);
		o.uv1.zw = TRANSFORM_TEX(v.uv3, _EmissionMap);
		
		o.pos = UnityObjectToClipPos(v.vertex);
		float3 worldPos = mul(unity_ObjectToWorld, v.vertex).xyz;
		float3 worldNormal = UnityObjectToWorldNormal(v.normal);
		float3 worldTangent = UnityObjectToWorldDir(v.tangent.xyz);
		float3 worldBinormal = cross(worldNormal, worldTangent)*v.tangent.w;

		o.TtoW0 = float4(worldTangent.x, worldBinormal.x, worldNormal.x, worldPos.x);
		o.TtoW1 = float4(worldTangent.y, worldBinormal.y, worldNormal.y, worldPos.y);
		o.TtoW2 = float4(worldTangent.z, worldBinormal.z, worldNormal.z, worldPos.z);

		return o;
	}

	void fragDeferred(
		VertexOutputDeferred i,
		out half4 outGBuffer0 : SV_Target0,
		out half4 outGBuffer1 : SV_Target1,
		out half4 outGBuffer2 : SV_Target2,
		out half4 outEmission : SV_Target3         

	)
	{
		float3 bump = UnpackNormal(tex2D(_BumpMap, i.uv0.zw));
		bump.z = sqrt(1.0 - saturate(dot(bump.xy, bump.xy)));
		bump = normalize(float3(dot(i.TtoW0.xyz, bump), dot(i.TtoW1.xyz, bump), dot(i.TtoW2.xyz, bump)));
		half3 diffuse = tex2D(_MainTex, i.uv0.xy).xyz;
		half ao = tex2D(_AORoughnessMetallic, i.uv1.xy).x;
		half roughness = tex2D(_AORoughnessMetallic, i.uv1.xy).y;
		half metallic = tex2D(_AORoughnessMetallic, i.uv1.xy).z;
		half3 emissive = tex2D(_EmissionMap, i.uv1.zw).xyz;
		outGBuffer0 = half4(diffuse, ao);
		outGBuffer1 = half4(1.0, 1.0, 1.0, roughness);
		outGBuffer2 = half4(bump*0.5f + 0.5f, metallic);
		outEmission = half4(emissive, 1.0f);

	
		
	}

这里为了以后修改方便在顶点着色器里我还另外存了worldpos。
材质的属性有：
漫反射贴图，法线贴图，遮蔽粗糙金属贴图，自发光贴图。
gbuffer的结构为：
RT0:RGB(diffuse color) A(AO)
RT1:RG(motion vector) A(roughness)
RT2:RGB(world normal) A(metallic)
RT3:RGB(emissive)
AO和motion vector和emissve之后再实现。

光照计算

当前帧的各种信息存入RT之后，计算光照的时候直接对它们采样。这里在同学大佬的建议下没有用unity的PBR而是抄了下虚幻的做法。
关于PBR理论和解释可以看毛神的博客。

基础概念

光通量：单位时间内某一表面接受到的光子总数量，要注意的是光通量不是亮度，亮度应该是光通量的密度，也就是单位面积的光通量。
立体角：立体角在球面上覆盖的范围是就是球面积分里的微元。计算光照时，