FSR 3.0
- Thread starter Zer
- Start date
mandazi39
Well-known member
Would love this! CPU always overworked compared to GPU for me on this game.
Really hope someone with some know how would respond!
Really hope someone with some know how would respond!
Because VAM uses OpenGL and not DirectX it would (in theory) not be compatible, I don't think any other FSR versions work with VAM actually, so we'll have to wait and see if 3.0 somehow does.
For frame generation I don't see why it wouldn't work on VR games that use DirectX.
For frame generation I don't see why it wouldn't work on VR games that use DirectX.
VAM 1.x uses DX11.
On a side note I made a thread in relation to this (performance, in general, including the possibility ((or not)) for DLSS and/or FSR):
Any plans for a 1.x update to implement DLSS / FSR support?
Hi. So, I'm aware of some unsupported / third-party methods for using DLSS / AMD FSR for the current version of VAM. I've had mostly failed attempts, one worked... but produced such bad image quality that just running the game at 480p might have looked better still. But, beyond the subject of...
hub.virtamate.com
DLSS implementation is planned for VAM 2.x, there's that much. But for VAM 1.x I think MeshedVR did probably everything he could, maybe.
The reason it's currently not possible to use FSR 2 and 3 is because VaM has incomplete motion vectors, if I ever find a way to "repair" the motion vectors that get generated, then I plan to add FSR 2 to https://hub.virtamate.com/resources/post-processing.37521/
FSR 1 would definitely be possible as is, but personally I don't think it looks very good, it has a lot of image instability because it doesn't make use of temporal data, it's purely a spatial upscaler, so although I'm sort of curious to add it, I'm much more interested in fixing the motion vectors because it will unlock a lot more possibilities.
If I ever get 2 working, 3 should be easy also because they use the same data.
FSR 1 would definitely be possible as is, but personally I don't think it looks very good, it has a lot of image instability because it doesn't make use of temporal data, it's purely a spatial upscaler, so although I'm sort of curious to add it, I'm much more interested in fixing the motion vectors because it will unlock a lot more possibilities.
If I ever get 2 working, 3 should be easy also because they use the same data.
This is great news, ty for looking for a possible 'repair' and FSR 2.0 implementation. If that does work it will help tremendously for general VAM performance (I would argue VAM is in such a dire need of FSR that the slight reduction in image quality is something nobody should be worried about, considering the boost in performance we would gain; basically we have a lot more to gain than to 'lose' from FSR implementation).
