Image Effects with Single Pass Stereo Rendering for VR

There seems to be a long standing issue with image effects and SPSR for VR.

Everything is working great in 2017.1.1f1

However after trying out 2017.2.0f2, I am having a lot of issues with them again. SSAO Pro results in a completely black screen, and a Highlighting effect does not work correctly either.

Were there any breaking changes I should know about that might help me adapt these full screen image effects to work with 2017.2.0f2?

I hope you can get an answer…

Please look at my older post being unanswered: https://discussions.unity.com/t/634990/17

This issue still exists: the camera width and height is never consistent with render texture resolution used internally and using command buffers there’s sadly no way to figure it out.
One could force render texture resolution once it is known for each HMD device/SDK used, that’s what I’m going to do.

The worst thing with this issue is that the width of the render texture is severely downscaled, creating the illusion to users that they are getting a huge improvement with SPSR.
This is non-sense… Single Pass doesn’t means an image effect have less pixels to handle compared to Multi Pass.

Actually, in 2017.2 onward, you can fetch the RenderTextureDescriptor from the eye texture manager via XRSettings.eyeTextureDesc. You can then use the resultant descriptor in CommandBuffer.GetTemporaryRT.

2 Likes

If you had a sample project I could look at, I’ll try to get a few cycles to see what’s going on. I’ve been working with the new Post Processing Stack v2 myself, and working on updates, but I haven’t taken a look at effects from the Asset Store.

To respond to your original point, there shouldn’t be anything in there breaking bits between 2017.1 and 2017.2, but then, we aren’t testing these effects from the Asset Store.

I can confirm that putting a post processing profile to a camera without turning on any image effect will still split the screen into two in 2017.1.2f1 with single pass. This does not happen in 2017.1.1p4 or before.

Could you please file a bug report with a minimal reproduction case and reply in here with the case #?

(Case #958473)
When I created the test project, I found that the first time I created a post processing profile (no image effect ticked) and attach it to the camera, it works. But once I ticked any image effect, and even I unticked all of them, the screen split into two forever until I removed the post processing profile from the camera.

1 Like