So, a few questions for 
In the stream I’ve already asked about the Async Post Process since that’s the feature I’m most hyped for in 4.12. Its this card on the roadmap:
It’s marked as [DONE 4.12] so I would think it’s in 4.12. I think it’s probably something like AMD showed here:
And since Post Process is quite heavy in VR I would love to know if it’s really improving stuff as much as showed there.
The other question I have is about the “compute based triangle occlusion and backface culling” on the roadmap. It’s marked for May/June, so it should be added to the master branch soon. I would like to know what exactly that means, my guess would be that it means implementing this stuff from AMDs GeometryFX they made free to use with their GPUOpen stuff:
Do you plan to integrate that code from AMD or is the roadmap talking about something else?
Next question would be about the Oculus Forward Renderer. You said you at Epic are working on your own Forward Rendering solution that would not support all features but be more designed for desktop than the current mobile forward renderer. How would your implementation differ from the implementation Oculus did? The way Oculus blog post sounds it supports all current desktop rendering features, which is great. When would the implementation from Epic be usable? VR games that are currently in development or released want to have such performance improvements as soon as possible.
I would love to see the path from oculus integrated as an experimental feature in the engine. The 73% performance increase they get are really significant. I can’t really use their code since I run a modified version of the engine and I want to be able to use 4.12 and released hotfixes and Oculus said they don’t plan to maintain their forward fork. Please take their code and implement it so that it’s not a “use theirs and stay at 4.11 forever” thing. 73% better performance is significant, but being forced to stay at 4.11 or having to constantly manually merge new versions and hotfixes is not nice.
Are you considering implementing oculus path or will you only focus on your own forward implementation?