Digital Foundry Article Technical Discussion [2021]

Status
Not open for further replies.
Is it the UI? I cannot imagine playing any RTS game with a controller although I know quite a few of them are really popular.
Consoles have full M&K support.

Could even make it mandatory.
I don't think anyone who really wanted to play it would take offence to that.
 
I liked the UE5 demo by The Coallition, they gave good insight about how te see VRS. According to them what VRS brings to table is predictable performance savings, they think in categories of frame budgets costs in XXX ms, and they like VRS because it is easier to plan rendering costs in a predictable way. Every technology needs to mature, every implementation can be better.
 
A hypothetical dynamic and clever VRS (apply VRS only to prevent the framerate of going under 60fps / 120fps) would have being an interesting tool if game engines didn't have mature DRS / TAA. Those techs are much better solutions overall as they do their job without destroying the artists work and overall sharpness (or much less) in the process.

I liked the UE5 demo by The Coallition, they gave good insight about how te see VRS. According to them what VRS brings to table is predictable performance savings, they think in categories of frame budgets costs in XXX ms, and they like VRS because it is easier to plan rendering costs in a predictable way. Every technology needs to mature, every implementation can be better.
This is a Microsoft studio. Why is (RDNA2) VRS mainly praised by Microsoft studios? This is a politized opinion, not an objective one.

AFAIK the only big studio that uses VRS is Activision and they carefully use their own software VRS that doesn't destroy the textures by notably making them blocky, as they immediately understood that it was not possible to ship a game with those artefacts / IQ. So they created their own VRS with a target (probably not 100% reached) of a 'win fps' 'same sharpness / IQ' solution as they clearly weren't interested into the standard 'win fps' 'lose sharpness / IQ' solution.
 
AFAIK the only big studio that uses VRS is Activision and they carefully use their own software VRS that doesn't destroy the textures by notably making them blocky, as they immediately understood that it was not possible to ship a game with those artefacts / IQ. So they created their own VRS with a target (probably not 100% reached) of a 'win fps' 'same sharpness / IQ' solution as they clearly weren't interested into the standard 'win fps' 'lose sharpness / IQ' solution.

Activsion employ a de-blocking process after their VRS pass to remove all blocking artefacts, the same blocking artefacts visible in Halo Infinite.
 
VRS, as with any tech used in rendering games, will always depend on the implementation. In Gears 5 and Doom, especially Gears 5, it is virtually unnoticeable unless you are told that it is been used. With Doom, you'd have to really zoom in to find it. That it is more easily identifiable in Halo: Infinite doesn't make it a bad tech. It just means that it wasn't implemented well enough, or it's more obvious because of the overall technical makeup of the game, etc. As with other innovations and tools in this space, it is used to make faster or prettier pixels where and when it matters.
Wolfenstein Youngblood has a pretty great VRS implementation that I couldn't see any noticeable decrease in image quality when it was enabled. Unfortunately for VRS, it also has an excellent DLSS implementation (that also disabled VRS) that looks better than native most of the time and runs even better. I am curious, since I couldn't find any benchmarks, if VRS has a more substantial impact if used on an RDNA2 card with RT on in Youngblood.
 
You have completely missed my point.

A lot of your examples are worth the effort as the performance (for any given visible artefact) increase is more substantial then what VRS has offered in any game thus far.

Is it really worth degrading the image with VRS for 1-3fps? IMO, no it's not.
You have big dials and you have small dials. No reason not to incorporate both. Everyone talks like DRS is a perfect solution, but it's got it's own issues. In order for users to not notice DRS there has to be a gradual change in resolution to avoid visible quality jumps. That means to ensure you're never going to run into frame rate issues, you must also be conservative with how early you reduce resolution resulting in, ultimately the inability to drive the GPU to its furthest potential.

If you're willing to be aggressive with how close you will lose frame rate with DRS, as in, you're switching very close to the point where you are about to lose frame rate, you still have VRS to smoothen the issues out. If you need more performance, it's there.

We haven't yet bothered to discuss that if the game is running 4K, you don't want it to dip all the way back down to 1080p either if loads get so rough. So to be able to have other avenues to keep the resolution up is still ideal.

VRS is still very much in its infancy, over time it will improve and have further features in how it can be leveraged to greater success.
 
Last edited:
Well in theory it should give a larger performance boost on RDNA2 with RT enabled.
In theory, yes. But I'd like to see if it does.

My little rant on reviews... They tend to stick to a handful of games across sites, and they often turn off features so it's a "fair comparison". That's all well and good, but if want to buy a card that will run Witcher 3 with hairworks cranked all the way up those reviews don't tell you anything. I get that it's objectively the same if you have the settings the same, but both AMD and (mostly) nVidia have exclusive features and sometimes that's the difference you want to see if you are looking to make a purchase. I don't know exactly what the solution is but I sort of liked the old HardOCP (I think it was them) way of benchmarking where instead of running some benchmarks and measuring FPS, they ran the cards at settings that were playable. So instead of Card X gets 47fps in this game and that's all the information you get, it was more "this card has to turn down these settings at this resolution but you can turn up these settings at little performance cost".

"Netease has implemented VRS into their upcoming game: Revelation Mobile (Tianyu). Through this collaboration and the improvements provided by VRS, the game saw a boost of over 30% in power efficiency and 38% in framerate."
I'm weary of any press release claims like this. nVidia claims something like 20-30% performance also IIRC, but most games that support VRS are in the 5-10% range.
 


Alex found the same performance issues on nvidia cards that computerbase found here:

https://www.computerbase.de/2021-12...hnitt_benchmarks_in_full_hd_wqhd_und_ultra_hd

The game is broken on nvidia cards, the same way as Valhalla and Dirt 5 was (at least during the time when rdna2 cards launched and every outlet benchmarked this game, after several months it was patched and the 30% lead of rdna2 cards magically evaporated and now nvidia is ahead in the game)
 


Alex found the same performance issues on nvidia cards that computerbase found here:

https://www.computerbase.de/2021-12...hnitt_benchmarks_in_full_hd_wqhd_und_ultra_hd

The game is broken on nvidia cards, the same way as Valhalla and Dirt 5 was (at least during the time when rdna2 cards launched and every outlet benchmarked this game, after several months it was patched and the 30% lead of rdna2 cards magically evaporated and now nvidia is ahead in the game)
the game could use another year of cooking likely but a game must be released. It's an ambitious project that went through development hell.

Hopefully in the coming years, a great deal of many things will improve and not just performance.
 


Alex found the same performance issues on nvidia cards that computerbase found here:

https://www.computerbase.de/2021-12...hnitt_benchmarks_in_full_hd_wqhd_und_ultra_hd

The game is broken on nvidia cards, the same way as Valhalla and Dirt 5 was (at least during the time when rdna2 cards launched and every outlet benchmarked this game, after several months it was patched and the 30% lead of rdna2 cards magically evaporated and now nvidia is ahead in the game)
GTX 1080 performing quite a bit worse than an Xbox One X.
 
That's why Halo Infinite at launch is an 86 and not a 92 like Forza Horizon 5. It needs some polishing.

Some of the review scores seemed to factor in the Multiplayer and deducted points because of the cosmetics and battlepass and playlist, the delay of forge, and the delay of co-op. I'd say the HI review scores are more impressive than what the overall numbers show.
 
Status
Not open for further replies.
Back
Top