URP Render Feature depth mask and _CameraDepthTexture

Hi, I am trying to create a render feature that renders a mask and depth value of an object to a texture. So this render feature does a pass after rendering pre pass. I render select objects to this texture using layermask. The shader for the render feature material writes this color to a texture rgba(1,depth,0,0). The graph is shown below.



I created another pass to test that the depth was properly being stored in the texture. The below image shows the G value.



The problem here is that the depth values here are different from what is found in the texture
_CameraDepthTexture. Here is the R values from _CameraDepthTexture. (Other objects in the scene are included)



As you can see these values very different and I can't use the two to compare depths of objects in one layermask and the depths found in the _CameraDepthTexture. Might anybody know what I am getting wrong here? I thought one might have been the inverse of the other but I tried that and it didn't work. Is there something that I can fix in the shadergraph? Maybe the way I am calculating the depth in the shadergraph is off? Or maybe there is some built in unity function I can call when doing the second pass?

This is the code I am working on right that I will use to do the depth comparison. For now, I used it to output the above two sample images.

Code (CSharp):

1. void OutlinePass_float(float2 UV, out float4 Out)
2. {
3.     float4 sceneColor = SAMPLE_TEXTURE2D_X(_CameraColorTexture, sampler_CameraColorTexture, UnityStereoTransformScreenSpaceTex(UV));
4.     float4 sceneDepth = SAMPLE_TEXTURE2D_X(_CameraDepthTexture, sampler_CameraDepthTexture, UnityStereoTransformScreenSpaceTex(UV));
5.     float4 outlineMask = SAMPLE_TEXTURE2D_X(_OutlineMaskTexture, sampler_OutlineMaskTexture, UnityStereoTransformScreenSpaceTex(UV));
6.    
7.     //Out = outlineMask.g;
8.     Out = sceneDepth.r;
9. }
10.  

 

Found the solution. So using scene depth node with sampling on "Raw" was the trick. Also encoding the float in a float2 when rendering to the texture helps with the precision.

Code (CSharp):

1. void OutlinePass_float(float2 UV, out float4 Out)
2. {
3.     float4 sceneColor = SAMPLE_TEXTURE2D_X(_CameraColorTexture, sampler_CameraColorTexture, UnityStereoTransformScreenSpaceTex(UV));
4.     float4 sceneDepth = SAMPLE_TEXTURE2D_X(_CameraDepthTexture, sampler_CameraDepthTexture, UnityStereoTransformScreenSpaceTex(UV)).r;
5.     float4 outlineMask = SAMPLE_TEXTURE2D_X(_OutlineMaskTexture, sampler_OutlineMaskTexture, UnityStereoTransformScreenSpaceTex(UV));
6.  
7.     float outlineMaskDepth = DecodeFloatRG(outlineMask.gb);
8.  
9.     Out = sceneDepth;
10. }

The encoding and decoding helper functions can be found in UnityCG.cginc.

Edit:

I found that the scene depth node doesn't work as I wanted to when I uncheck the layer under the "Opaque Layer Mask" filter setting in my URP renderer asset. This is because the scene depth node samples from the texture, it doesn't calculate the depth of the fragment unless it is rendered to the depth texture. To solve this I replaced the above "Scene Depth" node with a custom subgraph node "Object Depth Raw".



The Z Buffer Params is also a custom subgraph which just uses the Unity's built-in _ZBufferParams.

Code (CSharp):

1. void ZBufferParams_float(out float4 Out)
2. {
3.     Out = _ZBufferParams;
4. }

This is what I was originally trying to look for - a way to get the depth of a fragment without sample the depth texture. Now it would be great if I can find some type of function that converts any object/world space coordinate into that float4 "Screen Position". But for now, this works for what I need

 

Do you know how to make a renderer feature only apply to a specific camera in the stack?

 

I know it's a late repy but:
You can create multiple renderers (one per camera) with different render features.