Scriptable render pass ConfigureTarget and the depth buffer.

Unity Engine
, ,
1

I’ve created a scriptable render pass to support multiple render textures (MRT).
Here is code snippet:

RenderTargetIdentifier[] rb = {
        new RenderTargetIdentifier(proximityBufferId),
        new RenderTargetIdentifier(renderViewID),
};
public override void Configure(CommandBuffer cmd, RenderTextureDescriptor texture)
    {
        cmd.GetTemporaryRT(
            renderViewID, texture.width, texture.height,
            texture.depthBufferBits, FilterMode.Bilinear, RenderTextureFormat.Default
        );
        cmd.GetTemporaryRT(
            proximityBufferId, texture.width, texture.height,
            texture.depthBufferBits, FilterMode.Bilinear, RenderTextureFormat.Default
        );
        ConfigureTarget(rb);
        ConfigureClear(ClearFlag.None, Color.clear);
    }

public override void Execute(ScriptableRenderContext context, ref RenderingData renderingData)
    {
        var camera = renderingData.cameraData.camera;
        if (camera.cameraType != CameraType.Game)
            return;

        if (m_Material == null)
            return;

        CommandBuffer cmd = CommandBufferPool.Get();
        var drawingSettings = CreateDrawingSettings(proximityShaderTagId, ref renderingData, SortingCriteria.None);
        using (new ProfilingScope(cmd, m_ProfilingSampler))
        {
            context.DrawRenderers(
                renderingData.cullResults, ref drawingSettings, ref _filteringSettings
            );
        }
        context.ExecuteCommandBuffer(cmd);
        cmd.Clear();

        CommandBufferPool.Release(cmd);
    }

It seems to be working fine except it doesn’t write to depth buffer when I view it in the frame debugger.
The shader has ZWrite on.
I was initially trying ConfigureTarget(rb, depthTexID). I am wondering, what is the depthAttachment parameter for in the function? I assume it is for passing in a depth render texture and not the “depth buffer”. But I’m confused because since I’m changing the render target to some other render texture (not the usual _CameraColorTexture), does the same depth buffer get written to or do I need to specify it somehow?

How can I write to the camera’s depth buffer and also to my own render texture at the same time?

2

I think I understand where the problem is but haven’t yet found a solution.
I think I need to pass in the depth buffer with ConfigureTarget(colorTargets, depthBuffer)

Where with the built in render pipeline you could get the depth buffer via:
RenderTexture.depthBuffer

But IDK how to get the depth buffer from the ScriptableRenderer.cameraColorTarget, which is a RenderTargetIdentifier.

3 
    public override void OnCameraSetup(CommandBuffer cmd, ref RenderingData renderingData){
        ConfigureTarget(rb, renderingData.cameraData.renderer.cameraDepthTarget);
    }

I thought maybe cameraDepthTarget would be it but it doesn’t seem so.

4

It seems that changing the render target to something other than default (i.e. the screen buffer), it doesn’t write depth. Can someone confirm this?

5

Did you ever find a solution? I cannot figure out how the new RTHandle system handles depth. I also am having issues with depth when trying to make a custom pass. If the temp RT uses depthbufferbits of 0, then depth does not work in the custom pass - if I use anything more than 0, it terms my temp RT into a depth texture format and renders a depth pass instead of color

6

Hi, color and depth render texture cannot be combined together in a RTHandle. You should create a color RTHandle and a depth RTHandle and use ConfigureTarget(colorHandle, depthHandle); to set the render target.

If you don’t need to write depth, you should set the target with ConfigureTarget(colorHandle, colorHandle); instead of ConfigureTarget(colorHandle);.

To create a depth RTHandle, you can try:

RTHandle depthHandle;

// Get the camera target's render texture descriptor
var desc = renderingData.cameraData.cameraTargetDescriptor;

// No need to modify the depth buffer bits, it seems that creating a depth RTHandle has a higher priority.

// Allocate a RTHandle
RenderingUtils.ReAllocateIfNeeded(ref depthHandle, desc, name: "_MyDepthTexture");

// Set render target to depth only
ConfigureTarget(depthHandle, depthHandle);
// Or provide a color RTHandle
ConfigureTarget(xxx, depthHandle);

Example to create a color RTHandle:

RTHandle colorHandle;

// Get the camera target's render texture descriptor
var desc = renderingData.cameraData.cameraTargetDescriptor;

// Modify the descriptor.
desc.depthBufferBits = 0;

// Allocate a RTHandle
RenderingUtils.ReAllocateIfNeeded(ref colorHandle, desc, name: "_MyColorTexture");

// Set render target to color only
ConfigureTarget(colorHandle, colorHandle);
// Or provide a depth RTHandle
ConfigureTarget(colorHandle, xxx);
4 Likes
7

@wwWwwwW1 thanks! Will test when i get home from work

1 Like
8

@wwWwwwW1 Yes your solution worked for me. You are a life saver. I appreciate it!

1 Like