Home
Artists
Search Recent Random
Posts
Search DMs Popular Hash Lookup Tags Random
Importer
Import FAQ
Register
Partychan Telegram ThePornDude
Home Artists Posts Import Register

Bonus Material: Marvel's Spider-Man RTX 3060 DLSS vs PS5 IGTI (Patreon)

Published:
2022-09-04 08:30:24
Imported:
2023-01

Content

Hi everyone, Rich here. I mentioned last week that a great deal of background data was generated for the Marvel's Spider-Man Remastered performance video. The idea was to fully investigate how the game runs, concentrating on quality of data rather than quantity - the basis of the bar chart presentation, big on hardware tested, but necessarily extremely limited in terms of what can actually be tested.

Tomorrow, I'll be posting all of the GPU performance data and comparisons used for the graphics elements, but today here's a look at the RTX 3060 running DLSS with dynamic resolution, RT and effects broadly in line with PS5 - but exact settings matches are not possible, therefore performance comparisons are mostly meaningless, especially with both tweaking resolution on the fly in an attempt to maintain 60fps.

Still, I think the comparison is interesting in that two GPUs with very different hardware balances have very different ways of addressing a 4K output and both look good on a 4K screen. PS5 will be running at 1080p to 1440p with temporal injection upscaling to 4K. The 3060 will be operating in a similar window I would imagine, but using using DLSS for its ultra HD output.

Of course, PS5 can unlock performance via its VRR modes and frame-rates in the 70fps range and higher have been observed. So how does that work? If performance is unlocked and so variable, that suggests that dynamic resolution is disabled - otherwise what target would the PS5 be adjusting its resolution to meet? We grabbed some VRR footage using an Elgato HD60X and all shots we looked at were native 1080p, which explains how that likely works - presumably with IGTI scaling back to 4K. So I guess to achieve a similar effect on the RTX 3060, you'd simply lock to DLSS performance mode.

I've got a lot of data to upload - 29.3GB to be precise - and I'll post all of the GPU comparison files tomorrow.

Downloads: https://www.digitalfoundry.net/bonus-materials-marvels-spider-man-rtx-3060-dlss-vs-ps5-igti

Files

Bonus Materials: Marvel's Spider-Man RTX 3060 DLSS vs PS5 IGTI

The PlayStation 5 GPU has significantly more raster performance than RTX 2060, but significantly less RT power - so what happens if we compare dynamic resolution DRS vs dynamic resolution IGTI? Settings can't be entirely matched but the comparison is interesting nonetheless.

Comments

Anonymous

The games looks beautiful on a 3060, unbelievable

digitalfoundry
>>90965961

Per the main video, all graphics tests were done with a 12900K. To lock to 60fps with RT, I would be recommending a 12600 or 5600X. This game is VERY CPU heavy, though I understand Nixxes is still working on this.

Anonymous

Any plans for a DLSS vs IGTI image quality video?

d0x360

Should force enable resizable bar with nvprofileinspector. I did and I was able to stay above 60 at max settings & RT at 4k HDR and DLSS was able to go from balanced to quality. This is before the patches so all I had was launch day code and the nVidia drivers installed via nvcleaninstall I'd guess I gained about 15% in performance fps wise. Loading times were already fast but they seemed quicker by just a bit on initial load from main menu. Also no hitches anymore. They weren't bad before but they would happen every now and then but this stopped them. My specs.. -Asus ROG Crosshair VIII Formula -Ryzen 5800x no OC aside from PBO Auto -32 (2x16) gigs DDR4 3800mhz CL14-14-14-34 I find even though it's stable at 4000 it runs better at 3800. Best guess is the infinity fabric is at it's limits when set to 2ghz -3080ti FTW3 Ultra Gaming +148 core +920 mem in afterburner. -Samsung 980 Pro g4 m.2 (Windows install only) -WB Black 2TB g4 m.2 for just games so I can disable defender and it's extra protection options -WD Black SATA SSD this is where my swap file lives and just my swap file. Normally I wouldn't do that last one lol, I had an extra drive because I carried over drives from a previous build and figured it's gotta be slightly better than using a regular HDD. I like to keep the games on there own drive.. I really wish prices would come down on storage. Everyone who can should give it a try. It helps having the extra VRAM for sure but RT at 4k is brutal

OpenMind

It looks like DLSS is over-sharpening, boosting contrast in a way I don't find completely pleasant, a good example of this is the plant at around 14 seconds, the highlights look way too pronounced, Spider-man also seems to have lots of bright grainy highlights in the material of his suit, while some NPCs seem to have very dark grainy pores - but otherwise it seems to do a great job - as does the PS5 with IGTI. Another thing I picked up on is that (and this is likely a problem with the PC version not simply DLSS) but DoF seems to be incorrectly applied in some of the opening cut scene. The scribbled notes at 20 seconds seem to be out of focus on PC until they're almost off-screen, while more in focus on PS5, the circuit boards at 27 and the table at 30. I thought the table may have simply loaded a lower quality asset, but the edges appear blurred and at the top of the screen, where you can see the table surface behind the magazines, it seems to resolve with a high quality implying there is a high quality asset loaded. I also noted at 1:34 the objects behind Peter seem to be slightly out of focus on PS5 but not PC - this last one may be a difference in upscaling but I think the softer look makes the scene more natural. Glad to hear my 5600X shouldn't stand in the way of targeting 60FPS with RT though!

d0x360
>>90966478

IGTI looks terrible compared to DLSS (or FSR2 injection) it's essentially like the difference between DLSS quality and DLSS performance but DLSS gives a bigger boost in performance.

digitalfoundry
>>90969301

I think the DOF thing is simply that I may have left it on low settings, which is the optimised setting. Also, while both versions were set to full motion blur, full motion blur is far more pronounced on PC which may explain some of your other observations. Textures were also on high per Alex's optimised settings, but the 3060 could - and should - have had very high enabled, what with the 12GB of VRAM there. The sharpening point is interesting - subsequent to my testing, a sharpening slider was added for both DLSS and FSR 2.0.

alpha54

Are we still getting a video from Alex comparing all the various upscaling techniques in Spider-Man? Or am I misremembering whether this content was talked about it - would be super interesting

d0x360
>>90969301

Playing at 4k I found the image unstable looking and by that I mean I could see.. I don't know how to explain it but it looked like everything 15+ feet beyond Spiderman looked like it was covered in a slight heat filter and was kind of jittery. It wasn't just a lower resolution look it was that combined with some kind of odd movement. This was on launch day, I haven't looked since but when I tried it I had everything maxed. I also disable film grain and was using SSAO but now I'm using HBAO because of the fix which helped quite a bit.