Hi all,
Last night I did some testing to try and determine how to build content for VR/ the vive. Since I don't actually have one to test with I mocked up what I thought was a good approximation. Hopefully other people find this helpful.
Rough overview of my scene:
This is Unity 5 running deferred rendering with enlighten providing dynamic GI and baked reflection probes + box mapping. The cameras are set to HDR and I am using a tone mapping script. I also disabled all batching to keep things simple.
In order to approximate a Vive I made two 110 degree FOV cameras and then place + rendered them side by side. I set my game resolution to the Vive render target size I found in a Valve slide (3024x1680). My system is i7 4770 + a gtx 970.
To keep things above 90 FPS I was able to use:
~6 million triangles
~1100 set pass calls (draw calls)
I as also able to use SSAO but had to keep the samples low. This is super expensive at this resolution (5 megapixels).
Replies
Aside from finding some some specific limits I was looking to confirm that the right way to build art content for VR is to throw a lot of triangles at it.