AMD Potentially Preparing to Announce FSR 2.0 at GDC 2022

[Punkenjoy]

to the shift of everything being temporal. Faster paced games now can give me motion sickness.

FSR being spatial, whilst not as good IQ, at least gave an option that wasn't temporal based. Now it seems thats coming to an end. RIP.

You won't notice anything if the temporal AA is implemented correctly. No added latency, no ghosting, and even no shimering. The problem is it's not simple than turning a switch to do good TAA or TAAU.

Well I don't know about the upscaling that GPU makers use, but Adobe has an AI upscaling technology called super-resolution and I can attest as a photographer, it definitely increases IQ beyond the original. The results can be staggeringly good. It also corrects a lot of lens aberrations like distortion and can greatly enhamnce very fine detail. I have a 42MP camera and using super-resolution which upscales by 2x and then down scaling back to original resoltuion gives me files that are far more detailed than even the newer version of my camera that has 61MP.

I honestly could not believe what I was seeing. I will be reprocessing some of better landscape photos now.

So it might make you laugh now I guarantee it is possible.

Another example is noisy images. Topaz Denoise AI can actually restore features lost to noise because the AI knows what should be there.

AI image processing is already doing amazing things and it's it's still early days.

Now one way you can slightly improve IQ is to upscale an image (no AI) do some image processing and then resample back to native resolution. So I expect they mean if you ran FSR 2.0 without down-scaling, the resultant image could look better than native, but not by much.

AI can do very good, but this is not the same thing as DLSS. In DLSS, the upscaled pixel come from previous frame and the neural network is only used to select witch pixel to reuse. There is a limit on what you can do in real time.

The added details come from slight difference between frame (the renderer will shift the grid slightly each frame to resolve details and it will push further away the mipmap level so texture get more details than they would the backend resolution. It's clever and that do the job very well. If you don't have to use raw power to do something, don't. It's how to do effective real time rendering.

Will you ever say anything good about AMD's GPUs and software? BTW FSR allows you to play AAA Games at 1080P with Pascal GPUs. It is interesting that DLSS's expansion directly came right after FSR 1.0 launch. So you can actually thank AMD for your sentiment.

Don't hold your breath ! Some people there really have a shills. Let them be, they don't hurt anyone really.

 

[wolf]

It is interesting that DLSS's expansion directly came right after FSR 1.0 launch. So you can actually thank AMD for your sentiment.

It's interesting that FSR 1.0 came right after DLSS, so you can thank Nvidia for your sentiment too.

fwiw, I think FSR 1.0 is great, for what it is. No issue with more choices for people, and I have high hopes for 2.0.

 

[nguyen]

Don't hold your breath ! Some people there really have a shills. Let them be, they don't hurt anyone really.

Kinda funny that you keep wishing for the best and most popular upscaler atm to die and an inferior upscaler to flourish . Meanwhile I don't care about FSR/XeSS co-existing with DLSS, more choices are always better.

 

[Punkenjoy]

Kinda funny that you keep wishing for the best and most popular upscaler atm to die and an inferior upscaler to flourish . Meanwhile I don't care about FSR/XeSS co-existing with DLSS, more choices are always better.

I just can't cheer for closed solutions. If FSR or XeSS would only work on AMD or Intel hardware, i wouldn't cheer for them either.

So I could cheer for Nvidia if they open up DLSS but they will never do that. I agree the technology is good, and I have a RTX GPU in my gaming laptop to see it with my own eyes.

Nvidia made DLSS to capture market share and hold player into their ecosystem and it's succeeding. That is just not good for the GPU market. Who want another Intel 2010-2020.

So anything that can bring competition, from either AMD or Intel, is very welcome in my opinion. There is no company to cheer for here, they are all here to maximize profit. What we can cheer for is more competition.

 

[nguyen]

I just can't cheer for closed solutions. If FSR or XeSS would only work on AMD or Intel hardware, i wouldn't cheer for them either.

So I could cheer for Nvidia if they open up DLSS but they will never do that. I agree the technology is good, and I have a RTX GPU in my gaming laptop to see it with my own eyes.

Nvidia made DLSS to capture market share and hold player into their ecosystem and it's succeeding. That is just not good for the GPU market. Who want another Intel 2010-2020.

So anything that can bring competition, from either AMD or Intel, is very welcome in my opinion. There is no company to cheer for here, they are all here to maximize profit. What we can cheer for is more competition.

I don't care about any corporate greed or what not (good old capitalism), just the products they are selling. Heck Apple should be the scummiest corp on earth IMO but I have no wish for them to go down under.

Anyways as you said more competition is always good; FSR, DLSS and XeSS should co-exist and gamers can pick out the one they prefer (well not all gamers)

 

[Punkenjoy]

I don't care about any corporate greed or what not (good old capitalism), just the products they are selling. Heck Apple should be the scummiest corp on earth IMO but I have no wish for them to go down under.

Anyways as you said more competition is always good; FSR, DLSS and XeSS should co-exist and gamers can pick out the one they prefer (well not all gamers)

If all game have both DLSS, FSR2.0 and/or XeSS, i have no problem with DLSS staying as a technology long term and being closed.

The things is i strongly suspect nvidia will pay a lot of money to be the only upscaller on big titles. This is why i prefer that an open tech (if good enough) win and get the standard. No matter what PC i would use or what brand i would buy, i would be able to use the technology.

 

[Camm]

You won't notice anything if the temporal AA is implemented correctly. No added latency, no ghosting, and even no shimering. The problem is it's not simple than turning a switch to do good TAA or TAAU.

How condescending. We've had about 5 years of Temporal being in everything, I have my doubts in anything faster paced than most 3rd Person Action Adventure Titles when I enable anything temporal.

 

[wolf]

Nvidia made DLSS to capture market share and hold player into their ecosystem and it's succeeding. That is just not good for the GPU market. Who want another Intel 2010-2020.

I think that's such a different situation that it's not really comparable, for so many reasons it's hard to get into.

I don't see it as a bad thing necessarily for the market, Nvidia is pioneered this tech for modern gaming and now innovation in the market is moving at a crazy pace, naturally they will do what they can to make their GPU's more desirable than the competitions. But, I don't think it's up to them to simply make it open and work on everything for everyone's sake, it's up to the market to catch up and perhaps make super compelling features of their own, whether in software or hardware. And as we've seen, when you are a follower in the market, often your only choice will be to make the tech open, especially if the results are broadly worse. Like for some people, because I certainly hear this a lot, an oversized (relative to what would generally be seen as needed today) VRAM buffer is eminently desirable and virtually no amount of RT performance, DLSS, CUDA apps etc could sway someone to take an 8/10 or maybe even 12gb card over a 16gb card. This is, people who would actually consider any brand too, because there are clearly people for whom buying products from some companies is not even a consideration.

 

[ToTTenTranz]

Not a fan to the shift of everything being temporal. Faster paced games now can give me motion sickness.

FSR being spatial, whilst not as good IQ, at least gave an option that wasn't temporal based. Now it seems thats coming to an end. RIP.

FSR 1.0 isn't going anywhere.
Tomorrow AMD is releasing their first GPU driver with RSR, which is FSR enabled at a driver level. Just lower the resolution to sub-native and RSR will do its job.

https://www.amd.com/en/technologies/radeon-super-resolution

 

[Minus Infinity]

You won't notice anything if the temporal AA is implemented correctly. No added latency, no ghosting, and even no shimering. The problem is it's not simple than turning a switch to do good TAA or TAAU.


AI can do very good, but this is not the same thing as DLSS. In DLSS, the upscaled pixel come from previous frame and the neural network is only used to select witch pixel to reuse. There is a limit on what you can do in real time.

The added details come from slight difference between frame (the renderer will shift the grid slightly each frame to resolve details and it will push further away the mipmap level so texture get more details than they would the backend resolution. It's clever and that do the job very well. If you don't have to use raw power to do something, don't. It's how to do effective real time rendering.



Don't hold your breath ! Some people there really have a shills. Let them be, they don't hurt anyone really.

That is similar to pixel shift technology in cameras, you take upto 16 photos with each one shifted by 1 pixel. This allows you to overcome the limitations of the Bayer filter. I wasn't trying to compare Adobe's approach to DLSS, but pointing out to the the other person how AI can indeed improve IQ beyond native.

I have also preformed sub-pixel image processing for a project I was one and the results were very impressive.