The x2 shows what the initial FPS was as x2 is generating initial frames from the base framerate, thus ADDING that 50ms. So...roughly 18fps base FPS before frame generation kicks in. What is nice is that they know adding more frames to an already laggy feature isn't compounding the problem but exponentially reducing the hit you take from it. +50ms(x2) < +5ms(x3) < +2ms(x4).
I bet you can see PiSSeR though
James
nope he was specifically talking about the old CNN algorithm vs the new transformers algorithm (yes that is the marketing term I know). Hence why a big deal its being backwards compatible to the 2xxx generation. Frame generation is specific to each generation of card.
They can't use Indiana Jones because no one at DF has Gamepass
Yes the enhancements are more subtle. Hence you would need a direct feed if you don't want to rely on zooming. You'd likely notice the difference then side by side.
That screenshot is from nVidia's keynote, not DF footage, and it's also notably not a game, so it isn't something where latency would matter a great deal.
As always, these DF guys are complete sell outs. With this multi FG, the problem is not just the simple input latency. It's the input interval. You're basically getting 4 times the frame per each action you perform. Say if you have moved for 1/60 sec, your guy on the screen will move for next 4/60 sec. Which means if the base frame rate is 22, it still is effectively a 22 fps game when it comes to the control response. You're just getting 4 x the pictures. This is just crazy, and it's unbelievable to see these DF guys who's been saying how FG should've never been used in a low base frame rate games (say 30 to 60 like they did on PS5 Wukong) and now they keep their mouth shut for something like 20 to 80 LOL
It really is crazy how the 5080's raw raster performance is basically the same (or even arguably worse) as 4080 Super if you turn the 4x frame gen multiplier down to 2x which 4080S is using. So all we are getting with RTX5000 is more fake frames with the same dogshit latency, got it.
As always, these DF guys are complete sell outs. With this multi FG, the problem is not just the simple input latency. It's the input interval. You're basically getting 4 times the frame per each action you perform. Say if you have moved for 1/60 sec, your guy on the screen will move for next 4/60 sec. Which means if the base frame rate is 22, it still is effectively a 22 fps game when it comes to the control response. You're just getting 4 x the pictures. This is just crazy, and it's unbelievable to see these DF guys who's been saying how FG should've never been used in a low base frame rate games (say 30 to 60 like they did on PS5 Wukong) and now they keep their mouth shut for something like 20 to 80 LOL
Would you prefer he be a sexier older gent
I wanna see the game not some bald UK bloke
His rose tinted glasses are off for this one.I bet you can see PiSSeR though
BS. DF didn't argue for 22fps. They talked about how this could be helpful those who want to utilize the most of their ~4K ~240Hz monitors. Except some crazy edge cases no one is going to play games at 22 fps on 50XX anyway. At ~60fps and above f.ex. the ~16ms + ~10-15ms latency is low enough for most people. And the few remaining professional deniers can just turn it off it if makes their universe fall apart.
The generated frames are inserted between the real frames, so there is no extra movement.
Wait till end of jan/early feb for independend reviews, mate, i would love to buy 5090, even at 3k usd/euro(obviously something with beefed up cooling, this monster has 575w tdp after all ;p) but performance has to be there, and i dont mean dlss4 performance, i want proper raw raster performance at 25-30% above 4090 and around 2x vs my current 3080ti.
First they complain that every girl in gaming isn't fuckable enough, now they've come for the weirdos analyzing frame times?
The moment you stop moving the game realizes that without having to wait for any frame to be generated.
Hoes before balds.
Dude, no. If that were the case playing any game with that setting would feel like swimming on butter. Do you think that's what's going to happen?
That's not really how it works, the whole process runs behind, so it renders one frame, the next frame, and then generates the in-betweens, all before displaying any of them.
Doesn't work like that, at best you'll get a 50% FPS increase in Cyberpunk by using Fgen with a 4000 series, it won't just double your framerate.It really is crazy how the 5080's raw raster performance is basically the same (or even arguably worse) as 4080 Super if you turn the 4x frame gen multiplier down to 2x which 4080S is using. So all we are getting with RTX5000 is more fake frames with the same dogshit latency, got it.
Bro, delta time. The time between frames.
Do you seriously believe the 5080 will be worse than a 4080S? Basically equal to a 4080? Do you really believe what you just typed?It really is crazy how the 5080's raw raster performance is basically the same (or even arguably worse) as 4080 Super if you turn the 4x frame gen multiplier down to 2x which 4080S is using. So all we are getting with RTX5000 is more fake frames with the same dogshit latency, got it.
There are ways to intervene between frames in ways that respond to player input. VR games have had frame gen forever that responds to headset telemetry so it can generate frames according to the current position rather than just tweening like DLSS. This doesn't make the game logic update faster, of course, but it can make the game look and feel more responsive, perceptually.
You're still not getting it. Frame gen like Reflex 2 and ASW can consider player input and update in between fully rendered frames. While this doesn't necessarily change the internal game logic (which can also update or consider input between frames in many cases), it allows the visual output on screen to respond to changes in input between fully rendered frames.
