OnRenderImage conversion to URP ScriptableRendererFeature

I am working on a project that makes use of a third party library that has a script that should be attached to cameras in the scene. The library uses several functions from precompiled DLLs that require references to native texture pointers. The main problem that I have is that the script attached to the cameras uses the OnRenderImage callback which is not available in the Universal Render Pipeline. As far as I know I should be able to achieve the same functionality using ScriptableRenderPass features and I have been working on a solution but I can't figure out how to get references to the native texture pointers in a similar way as the original script.

Here is the OnRenderImage function from the original script:

Code (CSharp):

1.     private void OnRenderImage( RenderTexture source, RenderTexture destination )
2.     {
3.         RenderTexture.active = destination;
4.         if (-1 != viosoID)
5.         {
6.             IntPtr dst;
7.             if (!texMap.TryGetValue(source, out dst))
8.             {
9.                 dst = source.GetNativeTexturePtr();
10.                 texMap[source] = dst;
11.             }
12.             UpdateTex(viosoID, dst, IntPtr.Zero);
13.             SetTimeFromUnity(Time.timeSinceLevelLoad);
14.             GL.IssuePluginEvent(GetRenderEventFunc(), viosoID);
15.         }
16.     }

Could anyone help me with converting this functionality to Universal Render Pipeline logic? Currently, I have implemented a custom ScriptableRenderFeature and ScriptableRenderPass, but can't figure out how to get a native texture pointer instance from
a RenderTargetIdentifier during the ScriptableRenderPass Execute method. Is there an easier way to approach this?

 

Did you try RenderPipelineManager.endCameraRendering delegate with Camera.targetTexture.GetNativeTexturePtr() ?

 

I did, but unfortunately this did not work. I am not sure why. I guess it must be because the camera target texture is null as it is not rendering to a render texture but rather the screen. I guess I could get it to work somewhat by adding another layer of indirection by making the camera render to a render texture and rendering that render texture again e.g. through UI image. However, this sounds quite heavy performance wise, especially as my screen resolution is 4k x 4k. Is there anyway to avoid the indirection of having yet another render texture + camera combination in the scene?