• Hey, guest user. Hope you're enjoying NeoGAF! Have you considered registering for an account? Come join us and add your take to the daily discourse.

How AMD is Upping the Game With FidelityFX Super Resolution

Amiga

Member
Techradar talks with Nick Thibieroz, Director of Game Engineering at AMD, about FidelityFX Super Resolution and how it came about.

Nick Thibieroz: We wanted something that was accessible for multiple generations of GPUs, including past GPUs, to be able to run it. So, we could not set the bar too high. As to why we chose a spatial technique instead of anything else, that is a longer story that I can explain.

Essentially, once we decided that we wanted to develop an upscaling technology, we actually had multiple teams at AMD researching what kind of technology we could provide that ticked all the boxes I mentioned earlier, in terms of goals. So, we had several teams working on this. And then one of those teams was actually led by Timothy Lottes.

He's quite well known in the industry, he now works for Unity, and he actually presented on FSR a few weeks ago at SIGGRAPH. But Timothy is known for FXAA and CAS and LPN, like a bunch of technologies. So, essentially, he led the research for one of the upscaling teams at AMD.

..So, one day, Timothy provided a version of upscaling technology, and we knew straightaway that we had a winner, because we saw the edge quality, which is very important for resolution, right? The quality of the edges produced has to be similar to what you get when you're rendering natively at high resolution.

..But, initially, we did not start the project saying, 'okay, we need to do spatial.' No, we looked at different solutions, and to be honest, we are still looking at different solutions. But we felt we had a breakthrough with a spatial solution, especially with regard to performance and meeting all those goals. We feel the balance is just right for FSR, so we went with it.

 
Last edited:

ethomaz

Banned
Maybe they can successes with v2.
Right it adds nothing to developers.
 
Last edited:

Amiga

Member
Maybe they can successes with v2.
Right it adds nothing to developers.
point of this FSR technique is simple application on almost any engine.

did you check what moders are doing after the source was released? it's a big thing on Linux and right now and that means big things foe Steam Deck, and that in turn means big things for Linux again. there is serious momentum here.

 

Pagusas

Elden Member
I wish we’d see this starting to get back ported to older AMD partnered games like many of the Assassin's Creed entries.
 

M1chl

Currently Gif and Meme Champion
Bit too ambitions, sure. After all making post process filter, which is worse than existing solutions is hard...
 

Reallink

Member
That sure is a lot of meaningless rambling to say we didn't have the hardware or engineering know how to do a real AI upsampler, so we just shat out a modified CA sharpener that doesn't even come close to the competition.
 

Holammer

Member
Emulators start implementing it. RetroArch & RPCS3 afaik. Will be interesting to see it in Ryujinx for example where my computer is juuuuuust a bit too old for most games.
 
Maybe they can successes with v2.
Right it adds nothing to developers.

"We all" knew that this version 1.0 would be a temporary more simpler solution because doing the upscale using the temporal component and ML is harder (Intel just hired the same engineer responsible for DLSS from Nvidia), and also, AMD needs to think carefully on a different strategy for it that requires hardware support.
For now it's actually very good for everyone that AMD made this Version 1.0 the way they did. There are it's uses and cases, it can be used universally. It's extending the life of old and low end hardware there were not expected to be blessed with these new techs.

EDIT: it arrived in the lest version of RetroArch.



Ys_Seven_undub-fsr.png


E-Xp7KZWEBEuUhq.png

E-Xp73AXIAsqCX3.jpg
 
Last edited:

RoadHazard

Gold Member
"We all" knew that this version 1.0 would be a temporary more simpler solution because doing the upscale using the temporal component and ML is harder (Intel just hired the same engineer responsible for DLSS from Nvidia), and also, AMD needs to think carefully on a different strategy for it that requires hardware support.
For now it's actually very good for everyone that AMD made this Version 1.0 the way they did. There are it's uses and cases, it can be used universally. It's extending the life of old and low end hardware there were not expected to be blessed with these new techs.

EDIT: it arrived in the lest version of RetroArch.



Ys_Seven_undub-fsr.png


E-Xp7KZWEBEuUhq.png

E-Xp73AXIAsqCX3.jpg


It just doesn't look better to me. Sure, the pixels are smaller, but to me this looks more like the terrible smoothing filters some people use in emulators than any kind of useful upscaling. It's not intelligently adding any detail like actual AI upscaling can do.
 

Rikkori

Member
It's not intelligently adding any detail like actual AI upscaling can do.
Only DLSS 1.0 somewhat did this but that's dead. So no one today does what you want.

Maybe in the future Facebook will release theirs:

figure1-dancestudio-input-output-GIF-1.gif
 
Last edited:

supernova8

Banned
point of this FSR technique is simple application on almost any engine.

did you check what moders are doing after the source was released? it's a big thing on Linux and right now and that means big things foe Steam Deck, and that in turn means big things for Linux again. there is serious momentum here.



Yeah those results are insane. Sure it might look a bit blurry/flickery in places but if I'm going from like 30fps to way above 60 (or above 100 for RE Village) I think I'll take it.
 

RoadHazard

Gold Member
Only DLSS 1.0 somewhat did this but that's dead. So no one today does what you want.

Maybe in the future Facebook will release theirs:

figure1-dancestudio-input-output-GIF-1.gif

The latest DLSS doesn't? Well, whatever it does the results are much more impressive than AMD's solution. DLSS looks like the game is actually rendering at a higher resolution than it is, this doesn't.
 

Xyphie

Member
Only DLSS 1.0 somewhat did this but that's dead. So no one today does what you want.

The whole point of temporal upsampling is to accumulate samples over multiple frames to try and make an image with more detail than a single native res frame. So no, it's exactly what DLSS 2.x (or CB/TAAU/XeSS) tries to do.
 

Techies

Member
It's a game changer for VR, hope the Oculus software manages to implement it natively.

Someone released it for SteamVr games and now I can finally play Fallout 4 VR on my 1080ti @ 90fps with crisp image, wirelessly.

Wiithout it my game is blurry and still unplayable. This is with an old directx11 game.

You get a better image and around 30 percent performance boost.
 

Spukc

always chasing the next thrill
Only DLSS 1.0 somewhat did this but that's dead. So no one today does what you want.

Maybe in the future Facebook will release theirs:

figure1-dancestudio-input-output-GIF-1.gif
What kind of fake ass bullshit is this example lol.. look at the text above the couch.

fuck off with this shit 🤣
 

assurdum

Banned
Can be Ubisoft has updated WD Legions in the quality mode with FSR? I could swear on ps5 lately the IQ is more clean and with less shimmering especially compared the performance mode though the resolution is clearly lower.
 
Last edited:

ZywyPL

Banned
This video popped up yesterday on my YT recommendations:




And I have to say those are some pretty damn impressive results, basically double the framerate for nearly no image quality loss.
 
Top Bottom