Digital Foundry Article Technical Discussion [2023]

[Cyan]

Video is up.

colour me surprised, I didnt know this game features XeSS support. I haven't watched the full video yet, but I got to the point where Alex mentions the pixel reconstruction techniques.

 

[Dr. Nick]

What I promised earlier. The video is unlisted so that only people who have the link can see it.
Notice how much Green fills the screen when the MC turns. That is what I feel is proper usage, I think it may have been possible to make it work properly in Deep Space Remake with a bit more fine tuning.

 

[Below2D]

colour me surprised, I didnt know this game features XeSS support. I haven't watched the full video yet, but I got to the point where Alex mentions the pixel reconstruction techniques.

They're all subpar. XeSS has a weird issue where it flickers like hell with DOF turned on.

Overall, it's a terrible job.

 

[MCharta]

Video is up.

I still hope they patch the "beast" to fix texture streaming for 8 and even 6gb vram gpus.

 

[Cyan]

They're all subpar. XeSS has a weird issue where it flickers like hell with DOF turned on.

Overall, it's a terrible job.

tbh I have like 10 games that support XeSS natively and for instance Judgment and Lost Judgment, while looking amazing overall, have some issues with DOF in certain situations where the image looks like in the DF video, but for some parts of the scene, it's odd in that case. Here it affects everywhere where DOF is applied. With other games I don't remember having trouble with DOF on using XeSS although I usually disable it, along with motion blur and similar stuff.

 

[Inuhanyou]

Even solutions like DLSS and FSR2 and xess have problems when implemented poorly. I still don't know how forespoken is such a technical mess of a game to this degree

 

[yamaci17]

So the actual game is worse than the demo. Noted. On demo, 8 GB loaded textures fine at 1080p/standard settings, but they poofed away with ray tracing.
Seems like in actual game, the issue becomes more pronounced.

This also happens when the game literally only uses 6-6.2 GB VRAM out of total budget. Glad to see this behaviour being replicated on Alex's video. I hope he can raise the issue to developers. I think his exposition on Spiderman helped Nixxes to see the problem for what it is and increased the game's VRAM consumption which helped a lot.

However this begs the question: will this turn into a rat chase? Why they leave so much VRAM unused? Why they cannot give a simple texture quality preset? Not everyone plays at 4K, not everyone needs 4K tailored textures. What gives?

 

[techuse]

The Gears ones are pretty solid, especially given how long ago they were implemented.

Does hardware VRS work with compute shaders yet? If not it's still a big downside.

 

[eastmen]

nice video. Other than that, I'd love to see Elden Ring running on the Deck and see how it struggles to keep 60fps or maybe 30fps. It's not a super demanding game but it has a bit of a erratic behaviour. The engine is kinda special, it's the first engine I've seen in a game where the caves and interiors are seamlessly integrated with the rest of the open world. My hats off to From Software.

It runs fine ?

I played through it at 40hz

 

[pjbliverpool]

Video is up.

Brilliant video. As ever the content is so dense I feel like I need to watch it at 1/4 speed to take it all in. Covers pretty much every major aspect of the game in a great level of detail.

I found the DirectStorage comparisons particularly interesting. I'll probably have another stab at the demo with RT off as well. See how close I can get to a locked 60.

 

[mr magoo]

Depends how you plumb it in. If you’re doing VRS at the very end or if you are running it on selective buffers

It’s also noted in Martin Fuller presentation under performance gains. Adding VRS at very end after applying all effects like bloom, motion blur etc gives best results.

As long as it's applied smartly on places you as the viewer won't see it, I don't think anyone will mind VRS. DRS already does similar thing when the GPU is stressed ( in busy scenarios with a lot of movement ) with the entire image and the actual amount of times people notice it is relatively small unless the AA solution is bad especially now with FSR2 recleaning the image back up.

Yes true. However with VRS you gain one massive benefit, predictable performance gain. Let’s say you shave off 10ms from your rendering pipeline. Now you have constant value of 10ms to spend elsewhere. Afaik the Coallition guys said this is why they like it so much. Easier to decide on what to spend your rendering budget.

 

[iroboto]

It’s also noted in Martin Fuller presentation under performance gains. Adding VRS at very end after applying all effects like bloom, motion blur etc gives best results.



Yes true. However with VRS you gain one massive benefit, predictable performance gain. Let’s say you shave off 10ms from your rendering pipeline. Now you have constant value of 10ms to spend elsewhere. Afaik the Coallition guys said this is why they like it so much. Easier to decide on what to spend your rendering budget.

The nuance on how to leverage it makes for an interesting discussion as to why we see the implementations we’ve seen today.

Good discussion

 

[davis.anthony]

Let’s say you shave off 10ms from your rendering pipeline. Now you have constant value of 10ms to spend elsewhere.

I remember their presentation having a range of performance savings that varied wildly and not a fixed 10ms saving?

 

[cwjs]

I remember their presentation having a range of performance savings that varied wildly and not a fixed 10ms saving?

I don't think I trust the 10ms number, but it's (mostly) a fact that it scales down pixel shader time by a flat % -- how that actually manifests in framerate will vary depending how bottlenecked you are by pixel shaders in a given scene, but it's a great piece of information to know when you're deciding how many materials with how complex of shaders to include in the scene.

 

[iroboto]

Does hardware VRS work with compute shaders yet? If not it's still a big downside.

Nope. I don’t think there is any plan for it. Only on 3D pipeline

 

[Inuhanyou]

Yes true. However with VRS you gain one massive benefit, predictable performance gain. Let’s say you shave off 10ms from your rendering pipeline. Now you have constant value of 10ms to spend elsewhere. Afaik the Coallition guys said this is why they like it so much. Easier to decide on what to spend your rendering budget.

I wonder how much Infinity ward saves on performance. From the DF video on the modern warfare reboot it seems like a good chunk of performance

 

[mr magoo]

I remember their presentation having a range of performance savings that varied wildly and not a fixed 10ms saving?

https://devblogs.microsoft.com/directx/gears-vrs-tier2/

Total performance savings was between 3-4 ms across all rendering passes. 10ms was just arbitrary number for example purpose.

 

[mr magoo]

I wonder how much Infinity ward saves on performance. From the DF video on the modern warfare reboot it seems like a good chunk of performance

I think you can find your answers here https://research.activision.com/pub...able-rate-shading-in-call-of-duty--modern-war

Haven’t read it yet.

 

[Remij]

None of this stuff matters until they fix stuttering.

 

[Globalisateur]

Doom overall is significantly higher resolution on Xbox and manages to maintain much more consistency in resolution vs PlayStation. We have this in the thread where I look at spectrum analysis.

Ultimately we don’t play games in screenshots. And during developer interview John does agree that without 3x zoom it’s neigh impossible to notice. We just don’t play games like that. Our implementations for AA and upscaling Algol’s are just the same, trade offs in which we get temporary artifacts and edge artifacts in exchange for higher overall resolution, or slightly less better AA.

Native 4K with no temporal component will always look the same as 4K TAAU at a standstill screen shot, but motion is a completely different animal. Alex covers a lot of ghosting artifacts in many videos caused by occlusion or just looking left and right.

VRS will look worse in standstill to any trained eye, but in motion assuredly the overall higher resolution is going to be more noticeable.

The challenge is that as we continue to mix all sorts of technologies together, one has to ask if it continues to make sense. Compounding techniques could be an issue that requires additional navigation.

The reality is, this is still one of few (less than 10ish?) VRS titles ever released versus the hundreds of games that are using DRS, temporal AA and upscaling. There are more games using FSR than and RT than VRS. So this idea that we’ve seen peak VRS implementation is unlikely. People aren’t going to engage in it until they need it, and apparently it’s not needed yet as there are so many other areas in the game engine that needs upgrading prior to freeing up more compute. Wave 3/4 games may require it though, when the limits of these consoles are pushed at edge : you will want an additional tools to smooth out issues.

It’s important to note, VRS can also be dynamic, and no developer has implemented such a system yet. Right now it’s a constant application, but it doesn’t have to be. VRS if it ever becomes mainstream will have better implementations as times go on.

None of that stuff matters if most of the assets actually look higher res on the machine having the lowest resolution!

This is a old Doom Eternal comparison:

And I'd disagree about the spectrum analysis as it detects only one part of the problem. Keeping the native high frequency details is also important. With VRS you degrade the original assets and you get those blocky textures everywhere (you also get some ugly artifacts on reflections in Doom Eternal BTW). Ugh. This is my take after playing Metro Exodus, most of the times those degraded textures don't look correct at all. I would have prefered native textures in an otherwise lower resolution game (not noticeable because of the DRS anyway). I bet most of those blocky artefacts wouldn't be really detected by your tool. For instance in the case of Dead Zone super low aliasing on PS5, I doubt your tool would detect those 540p like assets (when TAA couldn't).

but in motion assuredly the overall higher resolution is going to be more noticeable.

Why would that be? COD developers do it in a much more scientific way with an ambitious target of retaining high frequency details with clever solutions. Which evidently matter very much for them even in this very fast paced game!

Hopefully after that Dead Space fiasco we won't see many multiplatform developers trying the technique in such an intrusive way again.