You can manually do this by enabling DSR and setting the in game resolution to higher than your monitor's resolution and then activating DLSS in the game. If you have the game output 4k, and the internal DLSS resolution is 1440p, then you get a really clean image. Performance is worse than native 1440, though.
Nvidia DLSS 1 and 2 antialiasing discussion *spawn*
- Thread starter DavidGraham
- Start date
-
- Tags
- nvidia
Flappy Pannus
Veteran
What other 1440p games on the PS5 are you comparing it to? Control does have a pretty strong TAA (one of the reasons DLSS works so well for this game, it can make it considerably sharper at higher resolutions), and the console versions have some settings like reflection quality/LOD on low compared to the PC, but I can't think of other 1440p games which look much better.
Edit: One aspect might be the complete lack of MSAA on the console versions, the way Remedy's engine works is that MSAA is kind of integral to the upscaling (you couldn't even disable it in Quantum Break), aspects like hair are far more artifacted vs the PC ime.
I'm not sure how this conclusion has been reached? The PC is being tested here at max quality settings which are most closely comparable tot he PS5's Quality mode (although an exact match doesn't seem possible). At that mode the PS5 is rendering below 4K almost all of the time and can drop significant resolution to deal with the bigger performance drops. The 2060 on the other hand is outputting at a fixed 4K (with DLSS). So surely this isn't a valid comparison?
Not fixed 4K: reconstructed 4K. But we'll never know how PS5 is really performing against Nvidia in that game as Digital Foundry decided not do that comparison.
Flappy Pannus
Veteran
That's DLSS in performance mode though, which means it's rendering at 1080p 'native'. Now the final image output quality may actually be superior to the PS5's quality mode (well, aside from LOD) due to the lack of pixel shimmering - which is what matters the most of course from the gaming perspective, but just from a 'raw' rendering power comparison the PS5 certainly seems to be punching well above the 2060 here. It's perhaps academic, but this is Beyond3D.
It sure as hell should.
Flappy Pannus
Veteran
Which is why I said as much.
D
Deleted member 86764
Guest
Miles Morales in RT Perfomance mode (1440p). This game even drops to 1080p in this mode and the image quality is *significantly* nicer than Control PS5.
Demon's Souls looks much cleaner in its 60hz (1440p) mode too, but admittedly I've only seen that in videos.
Last edited by a moderator:
Flappy Pannus
Veteran
Yeah fair enough. Insomniac's temporal upscaling is extremely good, as well Control does have some pretty low res buffers.
D
Deleted member 86764
Guest
There's something weird happening with the texture filtering in Control too. I can't tell if it's the denoising or not, but some aspects of the image look very noisy to my eyes. The close up cutscenes look decent, just the in-game stuff can look odd.
I think it's also got maybe a screenspace shadowing effect going on, the system seems to cull shadows around the main character and it causes a weird halo effect.
... but yeah, maybe just on consoles, I dunno...
I understand that which is why I said fixed 4k (with DLSS). The whole premise of the comparison snc made was that the 2060 'even with DLSS on' is struggling to hit 4k60 which has some bearing on PS5 performance. But my point is that I'm not sure it's comparable to what the PS5 is doing because the DLSS output resolution on the PC is 4k fixed, while on the PS5 it's less than that, sometimes significantly so.
Naturally the PS5 should be beating a 2060 in a non RT game at native resolution with ease.
Yes there's no doubt the PS5 is performing better than a 2060 as we'd expect, but as you point out above I don't think the native DLSS input resolution is a good comparison point as the DLSS step itself adds rendering time. I think the bottom line with this game is that it's just a really poor comparison point for PC and console performance. The settings don't match and the resolution seems near impossible to equalise.
snc
Veteran
nah its not comparable but still dlss performance mode is not stable 60fps on 2060 and isnt base resolution for performance dlss mode just 1080p ? also cant keep 60 at 1440p (ps5 60fps dynamic resolution highest 3840x2160 and the lowest resolution found being 2560x1440 median at 3072x1728 according to vgtech)
Last edited:
Keep in mind that DLSS has costs and that cost is fixed.
The less tensor cores the longer that fixed cost will be so it doesn’t matter how little or how much is happening on the screen really, it has little control over that portion. to meet 16.6ms you’re going to need to be significantly less than that to make up for DLSS time.
If we assume DLSS has a fixed time of 5-6ms (it is likely less). That frame time needs to be below or around 9ms to make it.
I’m just not sure if the 2060 is the card to do that, its nearly like asking for it to make 100+ fps. Which is tough already for most cards in the 6TF range.
Last edited:
I'm a bit perplexed by this statement.
the 2060 is 6.5TF. It's outputting a better an fixed 4K buffer that looks better than native while being under powered and with less bandwidth. It's not holding a locked 60 sure, I think that is a bit of a stretch anyway for it, but I don't understand the claim about the PS5 punching above it's weight, and in this case well above a 2060 here. 16 GB vs 6GBish. 448GB/s vs 336 GB/s, less clockspeed, less rops, less compute... PS5 by default weighs more than a 2060.
Just because the internal rendering resolution for DLSS is 1080p before it's boosted to 4K, doesn't make its output inferior. We literally just watched a video about how it's superior to native.
People need to get out of the mindset of trying to make apples to apples comparisons especially across architectures and vendors and rendering methods. There's just no value here.
Flappy Pannus
Veteran
A claim I never made, so I join you in the perplexed category.
Yes. And I said exactly that previously:
Flappy Pannus said:
And I said exactly that as well, l mean you literally quoted it:iroboto said:
Flappy Pannus said:
My initial response was to a poster that was seemingly impressed how much better the PS5 was doing next to a stock 2060 with performance DLSS at 4k, and I said outside of RTX uses, this is pretty much par for the course.
The interesting comparison points would be the tiers from 2070 to 3060ti IMO.
One issue I take with DLSS testing is that IQ is always tested at 4k where DLSS works the best. IQ suffers very noticeably as you enable DLSS at 1440p and especially 1080p. This is never factored in when people talk about performance on weaker GPUs.
Last edited:
DavidGraham
Veteran
Nope, I remember them pretty well. You also listed "pretty much every TXAA game" which is a false sweeping statement.
DavidGraham
Veteran
In this case, the 2060 uses it's machine learning cores to close the gap in compute and bandwidth vs the PS5.
I went back and checked Syndicate and it's indeed broken.
