They have devoted more time downplaying DLSS than showing how it can help PC gamers. So, nope, they dont get any credit for telling us what we already knew - since January, 15th 2020:
Current Generation Games Analysis Technical Discussion [2023] [XBSX|S, PS5, PC]
- Thread starter BRiT
- Start date
- Status
That's just a straight up lie. Absolutely devoid of any truth at all.
I mean, here's another video from a year ago where in the very beginning, he says this:
"DLSS of course has many strengths and we recommend that Nvidia GPU owners use it more often than not".
Again, y'all are so eager to hate on them, you're literally inventing criticisms.
Last edited:
Flappy Pannus
Veteran
That's one game, and the vast majority of the video is focused on ray tracing and vrs, not dlss.
Youngblood also has severe issues with motion blur and DLSS, which while not a significant regression as you can disable it easily, gets back to my point about low-res buffer effects being missed in these assessments often. However, more to the point, motion artifacting in general was not analyzed thoroughly with these early DLSS comparisons at all, as they usually focused on comparing still scenes, as is the case in this video - one scene where the camera is almost completely still is utilized to highlight the effectiveness of DLSS resolve. Of course with later DF videos this was greatly expanded to more thoroughly compare motion resolve.
Every channel has looked at how DLSS performs in individual games from time to time, I'm talking of a compressive assessment of how it is progressing over a suite of games. I also said they have given it unreasonably short-shrift at points in the past too, so your reply has little to do with my post.
Last edited:
Flappy Pannus
Veteran
I definitely use 4k DLSS perf over native 1080p every time too, it's no contest - but yeah - as your screenshots show, native 1080p has a significant performance advantage, and it can be greater in other games. DLSS has a cost, so it's only really relevant to compare it against bilinear upscaled modes that deliver the same performance, it's internal rendering res really isn't that relevant.
Also to properly judge image quality you need to compare shots with some camera motion, as that's where lower levels of DLSS can break down. My gripe with DLSS at these lower settings is not that when it produces artifacts, they're similar to what you get at the native resolution it's starting from - they can actually be worse. DLSS can amplify these at times, if it was just producing them at a quality of what you would get with native 1440p/1080p I wouldn't care much, they're a little blurrier, big whoop. It's only noticeable when they product artifacts you never see with regular upscaling, no matter what the starting res.
Also bear in mind these hair/fur artifacts only really occur with the High hair setting in Spiderman, DLSS just doesn't play well with its implementation, but it even affects the PS5 in Quality mode too, albeit not nearly to the same degree - but hair in performance mode at a lower res is more stable than the hair in quality more, precisely because it's using something close to PC's Medium setting performance vs. High (or slightly less) inn Quality. You just get far more noticable shimmer with it, to the point I don't think I'd ever go over Medium hair regardless of my GPU power. Medium's less fine detail is more than compensated for its image stability imo.
Flappy Pannus
Veteran
We're talking about Hardware Unboxed's coverage of DLSS, not the superiority of the PC platform in this thread, albeit I get for some posters that's their raison d'etre here.
Like why are you trying to educate me about the history and benefits of DLSS? I've looked at it in all games extensively, hell it's pretty much the only videos I ever post on my channel, and I pretty much always use it when available. Outside of bitching about shader compiling there is not a single facet of PC rendering tech I don't devote more time to.
Thanks but my comparisons are usually in movement as well (you can see Kratos running, or Ellie moving. That's why my comparisons are never properly aligned). I even did video comparisons. (Make sure to download them, as Dropbox preview view will be highly compressed)
davis.anthony
Veteran
I tried this last night (hair on highest setting) with the DLSS 4k down sample trick and it matched the quality of native 1440p with the games TAA which I was little surprised by as normally it offers better AA than native, but it was a huge improvement over plain old DLSS quality mode at 1440p.
Not true, Youngblood was the second game to get DLSS 2.0. Deliver us the Moon got it first in December 2019.
PS. the HUB hate here is absolutely ridiculous.
davis.anthony
Veteran
Hate is a strong word, dislike maybe better.
And they have bought a certain amount of it on to themselves with some of their testing and words.
Last edited:
Both games got the update in December. And there is no hate for AMDunboxed. They have never considered DLSS as a relevant feature for a GPU decision in the high end. Here is their conclusion from the 6800XT review:
2 1/2 years later this "questionable" feature looks in most cases as good as or even better than native resolution.
Do you have a source for this?
Official Betshesda patch notes are January 9th 2020. SteamDB shows the exact same date for the game adding nvngx_dlss.dll.
Nvidia introduced the RTX features a couple of days earlier during their CES 2020 presentation. I don't see anything pointing to December 2019.
It's exactly this kind of childish bullshit that I mean. Hate, name-calling, call it whatever you want, doesn't really matter. The dogpiling on HUB on this forum has been going on for a good while and it's pretty distasteful imo.
DavidGraham
Veteran
Yeah, they flip flopped a lot, and their messaging weren't consistent at all.
First they claimed DLSS2 is good, but not that good, then they insisted it doesn't factor in purchasing decisions, then claimed FSR1 is close to DLSS2, then as DLSSs was getting better and getting wider support, they said DLSS2 is important and a key selling feature of RTX GPUs then claimed FSR2 is also close to DLSS2 and is a DLSS killer, then said they are only going to test games with FSR2, then backpedalled on that decision, and now they say DLSS2 is far superior.
They were doing the same with RT too, first refusing to acknowledge its importance, and not factoring it in purchasing decisions, and ignoring testing it (outside of some curiosity tests), then started featuring it bit by bit, in their main reviews, and now they are factoring it in the overall results and conclusions.
They just need to be consistent, and not let the trends take them up and down.
Last edited:
D
Deleted member 2197
Guest
Yeah, they do have refurbished 3080's on amazon (and other sites for much less). Lowest price should be considered if price/performance comparison reviews are to be taken seriously.
Maybe because they called something "questionable" which has influenced a huge wave of new upscaling techniques to improve the life of a PC gamer. DLSS doesnt matter, it was the push towards better reconstruction techniques to make the most of the hardware. And we have already learned in January 2020 that native rendering will not the future anymore.
HUB was one channel who was against upscaling because it would have given nVidia a huge push. So they downplayed it as long as possible - see the 6800XT review.
The bolded part is blatant conjecture on your part, right? Or did they actually say they are against upscaling because they are anti-Nvidia?
Also, still waiting on your source for Youngblood DLSS 2.0 December 2019.
DavidGraham
Veteran
The test shows an interesting aspect, while the 10GB of the 3080 was not able to play Hogwarts Legacy at 4K with RT on because it ran out of VRAM, the 6800XT wasn't able either, as it lacked the processing power to do it. So 16GB of VRAM isn't always the winning forumula when it comes to these troubled games.
- Status
Similar threads
