AMD Fidelity FX Super Resolution 2.0 Gets 5 New Games

Published by

Click here to post a comment for AMD Fidelity FX Super Resolution 2.0 Gets 5 New Games on our message forum
https://forums.guru3d.com/data/avatars/m/198/198862.jpg
Good to see fsr2.0 more games support. It needs to catch on.
https://forums.guru3d.com/data/avatars/m/56/56004.jpg
Great to see the adoption of FSR 2.0 by more game developers, looking forward to The Calisto Protocol. Alex of DF must be pretty upset with more games supporting FSR 2.0
https://forums.guru3d.com/data/avatars/m/251/251189.jpg
mikeysg:

Alex of DF must be pretty upset with more games supporting FSR 2.0
Don't think so, as it will offer magnitudes of opportunities to demonstrate that DLSS is better. 😀
https://forums.guru3d.com/data/avatars/m/132/132389.jpg
aufkrawall2:

Don't think so, as it will offer magnitudes of opportunities to demonstrate that DLSS is better. 😀
Is it really better if it's locked down to one asshole company and requires dedicated die space?
https://forums.guru3d.com/data/avatars/m/251/251189.jpg
Neo Cyrus:

Is it really better if it's locked down to one asshole company and requires dedicated die space?
If AMD don't solve the transparency/occlusion issues of FSR 2.0: yes. 2.5 years have passed since DLSS 2.0 in Youngblood, how much more time do you want to give AMD? (Well, you apparently already went RTX card yourself...)
https://forums.guru3d.com/data/avatars/m/132/132389.jpg
aufkrawall2:

If AMD don't solve the transparency/occlusion issues of FSR 2.0: yes. 2.5 years have passed since DLSS 2.0 in Youngblood, how much more time do you want to give AMD? (Well, you apparently already went RTX card yourself...)
You're missing the part where we all lose if it's proprietary to one dickbag company that makes it require specialized hardware. I'd rather use FSR 2 on everything than have to permanently eat nVidia's shit.
https://forums.guru3d.com/data/avatars/m/251/251189.jpg
Dedicated hardware makes inference part of DLSS faster, probably by a large margin. By how much we'll see with XeSS, but neither you nor me do know that yet. Criticizing them for investing silicon to make things faster is just silly. Also, Nvidia can only commit as much assery as their competitors allow them to do. FSR 2.0 having issues and you buying their high-end cars despite you hating them just proves their point to keep things closed...
https://forums.guru3d.com/data/avatars/m/132/132389.jpg
aufkrawall2:

Dedicated hardware makes inference part of DLSS faster, probably by a large margin. By how much we'll see with XeSS, but neither you nor me do know that yet. Criticizing them for investing silicon to make things faster is just silly.
No, and no. The benchmarks so far show DLSS barely being faster than FSR 2 despite having dedicated hardware. Criticizing them for having dedicated hardware for it is far from "silly" considering it was never meant for gaming to begin with. RT and Tensor cores were meant for pro workloads, potential uses in gaming was a complete afterthought, not at all close to the primary goal. nVidia's approach to AI upscaling is the same as their asshole approach to VRR, they're looking for a marketing advantage that they know can't possibly last forever. Good upscaling being locked to nVidia will die off just as VRR being locked to nVidia died.
aufkrawall2:

Also, Nvidia can only commit as much assery as their competitors allow them to do. FSR 2.0 having issues and you buying their high-end cars despite you hating them just proves their point to keep things closed...
I bought my 3080 before FSR 1.0 even existed, and during a time that there was literally 0 stock of AMD cards, I've had it for almost 2 years now. The days of DLSS being added value are over, no one is going to buy nVidia cards just because of DLSS anymore.
https://forums.guru3d.com/data/avatars/m/108/108389.jpg
Neo Cyrus:

No, and no. The benchmarks so far show DLSS barely being faster than FSR 2 despite having dedicated hardware. Criticizing them for having dedicated hardware for it is far from "silly" considering it was never meant for gaming to begin with. RT and Tensor cores were meant for pro workloads, potential uses in gaming was a complete afterthought, not at all close to the primary goal. nVidia's approach to AI upscaling is the same as their asshole approach to VRR, they're looking for a marketing advantage that they know can't possibly last forever. Good upscaling being locked to nVidia will die off just as VRR being locked to nVidia died. I bought my 3080 before FSR 1.0 even existed, and during a time that there was literally 0 stock of AMD cards, I've had it for almost 2 years now. The days of DLSS being added value are over, no one is going to buy nVidia cards just because of DLSS anymore.
Lol just don't buy Nvidia if you don't like them, make it easier for others to buy Nvidia then 😀
https://forums.guru3d.com/data/avatars/m/251/251189.jpg
Neo Cyrus:

No, and no. The benchmarks so far show DLSS barely being faster than FSR 2 despite having dedicated hardware.
Because FSR 2.0 is not DL based, like multiple other TAAUs before. You likely have some weird misconception what TAAU does, as many unfortunately have. If you think it looks good enough, fine. But speak for yourself. My impression after testing both DLSS and FSR 2.0 extensively in GoW is that FSR 2.0 looks to be flawed at a fundamental design level when it comes to sudden screen content changes. And I say that as somebody who criticized DLSS smearing issues a lot. But DLSS smearing in GoW is completely harmless vs. the issues FSR 2.0 has in GoW (or rather with its dynamic content). You must have low expectations for image quality if you think this TAAU in its current form should completely replace DLSS.
https://forums.guru3d.com/data/avatars/m/108/108389.jpg
aufkrawall2:

Because FSR 2.0 is not DL based, like multiple other TAAUs before. You likely have some weird misconception what TAAU does, as many unfortunately have. If you think it looks good enough, fine. But speak for yourself. My impression after testing both DLSS and FSR 2.0 extensively in GoW is that FSR 2.0 looks to be flawed at a fundamental design level when it comes to sudden screen content changes. And I say that as somebody who criticized DLSS smearing issues a lot. But DLSS smearing in GoW is completely harmless vs. the issues FSR 2.0 has in GoW (or rather with its dynamic content). You must have low expectations for image quality if you think this TAAU in its current form should completely replace DLSS.
FSR1.0 probably looks better than 2.0 in GOW in 4K (UQ vs Q mode)
data/avatar/default/avatar37.webp
What about Fortnite ?
https://forums.guru3d.com/data/avatars/m/251/251189.jpg
Seth-01:

What about Fortnite ?
There is only UE TAA, UE upscaling (not TAAU) and DLSS (which looks sharper in still image and in motion than UE TAA).
https://forums.guru3d.com/data/avatars/m/121/121885.jpg
aufkrawall2:

Because FSR 2.0 is not DL based, like multiple other TAAUs before. You likely have some weird misconception what TAAU does, as many unfortunately have. If you think it looks good enough, fine. But speak for yourself. My impression after testing both DLSS and FSR 2.0 extensively in GoW is that FSR 2.0 looks to be flawed at a fundamental design level when it comes to sudden screen content changes. And I say that as somebody who criticized DLSS smearing issues a lot. But DLSS smearing in GoW is completely harmless vs. the issues FSR 2.0 has in GoW (or rather with its dynamic content). You must have low expectations for image quality if you think this TAAU in its current form should completely replace DLSS.
Most of the issues present FSR 2.0 are solvable in the same way the same the issues where in TAA itself. It's going to be interesting to see it's refinement over time. Because really none of the issues we are seeing in FSR 2.0 are stuff we haven't seen resolved in TAA before (Even in GOW's own TAA). Also I think Alex is probably excited. As he generally finds temporal upscaling interesting. And as we get more temporal methods, it's going to be awesome for the type of comparisons he likes doing dives into.
Seth-01:

What about Fortnite ?
Fortnite is on UE5 now right? Although I'd love to see FSR 2.0 come to Fortnite, I'd also love to see it showcase EPIC's own TemporalSuperResolution (TSR).
https://forums.guru3d.com/data/avatars/m/282/282473.jpg
Trunks0:

Most of the issues present FSR 2.0 are solvable in the same way the same the issues where in TAA itself.
of course they are but they require updates from developers it's not as simple as dropping a dlss dll into the folder and having contrails and ghosting reduced.
https://forums.guru3d.com/data/avatars/m/121/121885.jpg
cucaulay malkin:

of course they are but they require updates from developers it's not as simple as dropping a dlss dll into the folder and having contrails and ghosting reduced.
Absolutely true. Although we didn't even know we could do that with DLSS 2 till later after release, further still till nVidia officially accepted that and implemented that swap ability driver side. And it's still not universal, as its not always an improvement for some games and can't be done at all with some DLSS 2 games(I think it's Anti-Cheat systems that prevent it in some cases?). I think people need to stop thinking of different scalers as something that will kill off one or the other.
https://forums.guru3d.com/data/avatars/m/251/251189.jpg
Trunks0:

(I think it's Anti-Cheat systems that prevent it in some cases?).
Those are very rare cases, though it can be a hassle to always replace the DLL before starting the game when the launcher always replaces it.