Digital Foundry Article Technical Discussion [2021]

Status
Not open for further replies.
Saw similar artifact in the DF GodFall video where falling leaves floating across the metal disk the character is standing on ... like a comet trail appearing behind leaf. Timestamped below.
FSR uses TAA's gradients to reconstruct geometry edges in higher resolution. Gradients essentially show the distance to the geometry edge hence you can use them to build an edge plane and then clamp pixels to this plane in higher resolution.
TAA produces temporary stable gradients, hence FSR relies heavily upon it to upres the edges. When TAA fails, so does the FSR, and all temporal artifacts like ghosting, shimmering, etc are inherited and amplified too (since by increasing 1 shimmering pixel 4x spatially you will get 4x more shimmering pixels in output res)
 
I think TAA U is just as good an option as tsr is, if not better. Re-inventing the wheel for no reason here.
If only every engine had a quality TAAU solution. AMD just gave a "better than nothing" option away for those engines/games that don't. I don't disagree that TAAU looks better. Especially Unreal's in engine TAAU. But not every game uses Unreal, or an engine that has a built in TAAU. That's why this wheel was invented.
 
Could it be added to a game w/o TAA, like Nioh 2?
I wish there was a way to add it via a control panel toggle, even if it caused hud elements to get blurry. Mostly for older games that don't have high resolution support and for benchmarking/IQ analysis for games that support other forms of upscaling. Or maybe this could be achieved with reshade.
 
1440pupscaler9jvj.png

I made one with Quality mode in the end for comparison as I was going to make more, but the game required an internet connexion to their server to test - and halfway through our testing FSR, the game stopped working and did not connect to the server. In the end I limited my comparison to the 1080p result in the video because it proves the point that TAA U is superior to FSR at a similar cost - something the screenshot above also shows for a higher quality mode if you just look at it.

Thanks for pointing out the comparison alex. The reason i asked was because i was curious as to the improvement fsr gave to games in a vacuum. Not as good as specialized solutions definately. But is it fair, like others have mentioned, that its a useful technique for upscaling even if not as good as the others?

I guess i was confused because you seemed really down on the tech when we are supposedly expecting devs to be utilizing it like luminous productions on ps5 and i didn't get a sense in your video if a technique like this was even worth it based on the results

I guess that was mostly my fault coming at the analysis from a console perspective hoping to see what the tech would do compared to current games as opposed to a strictly pc based analysis with other competing upscaling tech.

Great video btw.
 
From yesterdays pcworld podcast with AMD we learned the current version of FSR exists to address the current state of the market by trying to reach as many gpus as it can. There will be FSR versions likely proprietary to RDNA and with different methods and capabilities, as per AMD in the podcast.

So this is phase 1 it seems.
 
I agree with DF the biggest advantage of FSR is the number of GPU supported and in theory older games without TAA can use it. Imo the biggest weakness is dev need to patch the game to use it.

The quality is not as good as temporal but it is a compromise.

I think for the moment this is more interesting for older PC GPU. On consoles many teams have their own solution.

EDIT: Forget the fact this is an Open source solution maybe PC community can do something for it on older games
 
Last edited:
when we are supposedly expecting devs to be utilizing it like luminous productions on ps5
I'm not very bullish on this. There are tons of upsampling techniques out there and a very large number of custom ones: insomniac et al.

I don't see this breaking into the console market outside of any sort of exception case; it has an uphill battle with existing upsampling technique which may work well for your title/engine.
 
Last edited:
Thanks - these last few days have been a whirlwind. First doing all the extra hours to get Metro PS5 done and start FSR, make FSR video and article, then have to freak out as I read a post on reddit saying DOF changes its aperture.
Long story short:
tumblr_p378y8BgaB1txtch1o2_500.gif
lol yea I was going to say, you've had a rough week with dealing with all of this. Definitely take a well deserved break. People don't understand how much time is spent setting up all this stuff for analysis. Really easy to armchair things. You could spend forever trying to get the exact pixel coordinates for screenshot analysis. Across so many settings as well.
 
Thanks - these last few days have been a whirlwind. First doing all the extra hours to get Metro PS5 done and start FSR, make FSR video and article, then have to freak out as I read a post on reddit saying DOF changes its aperture.
Long story short:
tumblr_p378y8BgaB1txtch1o2_500.gif
:LOL: chin up! You are young, you are a good looker and smart. You will do fine.

Hope you get your vacation soon and can bid farewell until returning from vacation to your work colleagues with a resounding...

775e60e30b0ee2bfe00992f5f29b032a.gif
 
I'd be curious as to know why the game seems less performant on the upgrade versus bc. Especially since they are supposedly similar visuals and feature wise outside of loading.

Also DF should hire this guy. Rich John Alex tom..he seems a right fit with those guys
 
I'd be curious as to know why the game seems less performant on the upgrade versus bc. Especially since they are supposedly similar visuals and feature wise outside of loading.

Also DF should hire this guy. Rich John Alex tom..he seems a right fit with those guys
Pretty sure he already has a job... For himself.
 
Status
Not open for further replies.
Back
Top