#can we get the fps capped to monitor refresh rate back? Or at least get an option to enable it?
40 messages · Page 1 of 1 (latest)
YO MOM I AM FAMOUS
XDDDDDD
you are a really good and practical example of one good reason people wanted uncapped FPS
although most just wanted to flex their ultra high fps
@solar quail isn't there a little quality bump since you are generating from higher FPS?
I can notice it a little
no, because when you framegen for 144 and had 90 fps, LS would do 72 real and 72 fake or 48 real and 96 fake
isn't that what vsync option does or you mean something else
yeah but it adds latency
vsync prevent rendering excess frames and that add latency
normal thing
what version 2.9 is doing ?
i exaplined if you look at the first screenshot
but it worked like this:
let's say you have a 60hz monitor.
You want to do framegen from 35 or 40 fps to have a lower much better latency.
When you activated framegen your game would keep running at 40 fps, while Lossless scaling would work by only taking 30 of those frames and applying framegeneration.
Basically, your game would run at 40 fp(lower input latency), while lossless scaling would do 30 real frames and 30 fake frames, thus consuming less gpu resources.
and so what it did for let's say your MONITORS REFRESH RATE IS 60 HZ, no matter your in game framerate LOSSLESS SCALING WOULD DO:
for 2x => 60 / 2 (30 real and 30 fake) frames
for 3x => 60 / 3 (20 real and 40 fake) frames
While the game was ran at 40 fps or more.
Same logic applies for all refresh rates of the monitor
for a 144hz monitor, while running the game at 90 fps
for 2x => 144 / 2 (72 real and 72 fake) frames
for 3x => 144 / 3 (48 real and 96 fake) frames
While the game was ran at 90 fps or more(lower input latency, game responsiveness for 90 fps instead of 72 or 48).
and so there was no need v sync and you could have allow tearing turned on for even less latency, mix that up with VRR from amd or intel and you had the best possible results
it used to do what everyone does now when they framegen over refresh rate, but better and more optimized, because you didn't consume as many gpu resources.
better input latency for less consumed gpu resources.
you would also get higher quality results, because the app had more headroom to process image data and return higher quality images and better frametimes/latency
now it is much easier to bottleneck if you are looking for the same results and get worse results overall
ohhhhhhh thats deep af XD , but how did you managed to know that v2.9 was working like that ? also the lsfg use alot of resources already and if they did what you ask this will result in bad quality since framegen use less frames 🤔
that is also a good point but applies only for 60hz monitors. once you reach 144 hz a min of 48 or 72 is enough to not get artifacts or low quality results.
hell on this new update i don't get artifacts for 45x3 on cyberpunk
with the fps sometiems going to 35
or they are so small i don't notice them
i had a brinstorming session with someone guessing how it worked and then asked ths if i was right and he confirmed XDDD
we need a dev here to explain missing pieces and possibilities
splits captured fps based on monitor refresh rate
like i mentioned above, but you can ping and ask the dev on his opinion here
i personally found it crazy smart that he did what he did on 2.8 and it's sad that it wasn't appreciated by people
umm i like the way you thinking
Thanks
there is no swapchain buffer in my config now in latest version , max frame latency is in software itself now i know but no buffer count option 🤔