tests on a AMD GPU would as be good here. Its hilarious you expected and improvement in you upsacle 720 to 1440p using FSR. It needs at least upscale 1080p to 1440p to see the improvement
Do you really think that the industry has that many good developers? This is not a problem of just Game Development its generally a problem of all Developing Industries, they even hire college students. People want Games released fast This is what you get.
@@kCenk LOL the prosumer's fault?👀👀👀👀👀👀 The industry does it because they can get away with it. The cash cow erupted in 2020 and they'll never look back as long as people keep coughing up bucks a transactions. edit- "prepurchase today!" 🤪
Specially if the new products its Garbo and the marketing have convinced people play on PC like on console its the way to go. Nvidia will continued down this route : 90-80 series good gen jump , 70-60 garbo products with terrify prices and a software let people play like on console ( its upscaling from ages). The people keep cheer on .... sadge
@@CJTheTokay its a "60 series" by Die size with a pathetic memory bandwith will last no more than a year. But thats its all Nvidia want it : make you change GPU every year.
They are already charging for it semi annually. DLSS is great, FG is awesome, except on the card I bought two years ago, it isn't worth a damn on that. I really think comparing the two is sort of apples to oranges. FSR is great in that it can keep older cards relevant, DLSS mostly works on cards that shouldn't need it.
Sorry Vex, I disagree with you on this (Personal opinion), I think Ancient Gameplays did a better comparison, FSR 3.1 is definitely better than 2.2 and I CAN TOTALLY SEE the difference while playing Horizon Forbidden West, it's much more stable. 1440p FSR Balance used to look terrible and now is decent and in QUALITY MODE I hardly even notice the difference from native res. Also, without zooming in and slowing down you won't even notice anything whatsoever because people just don't do that. Don't get me wrong, FSR 3.1 is not a game changer (It's not even using dedicated hardware), it still has flaws, DLSS is still better but it is becoming insignificant at this point.
If it does end up using dedicated hardware then you can expect most cards that are supported now to suddenly become unsupported, then people will really hate AMD and almost fully abandon them.
@@ArtisChronicles That's not true, XESS has a fallback code and it looks good without dedicated hardware, FSR could do that and move on with machine learning, it's really impresive how far in quality FSR has come with broad compatibility.
The zooming in part and slowing down is clearly for purpose of making a video and relaying what he sees to the viewer so they can see it. It's an academic approach. It can also make up for RUclips's compression quality losses. Out of all criticisms I see the gripes with slowing down footage and zooming in is the strangest I have to admit. I mean the goal is to nitpick /review.
@@silvio351thing is what intel did with XeSS is much more complicated than nvidia DLSS. and AMD push FSR because they did not want to deal with ML stuff like nvidia did with DLSS. Now you're expecting AMD to take even more complicated method like intel XeSS?
@@TheBlackAnomaly it doesn't matter what you mean,dlss is perfect solution to gain performance with minimal hit on image quality,fsr is a short term solution from AMD until they find the resources to add full ai hardware acceleration which it will happen in rdn 5
@@Jaba-hs6vd it really shows how little AMD understands what is happening. Nvidia already had AI acceleration back when AMD didn't even have DP4a and mesh shaders support. it was other way around back in GCN1-2 architecture days
Honestly, yes, DLSS is still better, but the difference in quality, playing at least at 1440p Quality or even 1440p Balanced, is negligible. DLSS was the last thing that still kept me take Nvidia into consideration. Now that FSR is in practice at the same level of DLSS, Nvidia to me is irrelevant.
Yep. Indeed dlss is better but let's be honest.. for the same price you can run a game at native with AMD and quality upscaling with Nvidia and get similar performance. Only when you have a mid range GPU on a 4k screen running the latest games slare you required to upscale and upscaling looks better at 4k by far. It's a tossup as to whether Nvidia GPUs are better overall. Tradeoffs are made unless you're rich.
FSR is not anywhere better than DLSS if not in really specific scenes, conditions in specific games; to say something similar you're doing a really forced cherry picking. As he says, FSR 3.1 barely holds up to the worse version of XeSS 1.3 (22:18). FSR can be passable at something like 4k quality, maybe 1440p quality sometimes. It of course depends on what you consider to be a satisfying visual quality and on how much you notice visual breakups.
Because in youtube 60fps lock and low bitrate, normal gameplay would not be as telling of the details he wants to point out. Wich all the those details add up to be noticeable in real life gameplay.
Because he’s trying to see the total difference, quality is less easy to tell. Sure most people use quality I’m sure, but this is the best way to do this.
Not true, I play at 1440p and I do use performance mode (720p->1440p) a lot, depends on a game tho. In some games I am using performance mode only, like cp2077 - can't tell difference, so why not have more fps. In some games even quality mode looks noticeably worse then native, so no upscale there. Best thing tho is to use DLSSTweaks mod and choose base resolution for upscale yourself, in my opinion.
@@agentnukaz1715 then you are mistaken not every 4k gamer has a top tier gpu and also 4k performance is still better than 1440p quality. 4k is a really good resolution for using upscalers.
The moment you have to slow down to show the oh so 'noticeable' differences you can ask yourself if dlss is really worth the additional tax we pay to Nvidia and you smartly avoid that part of the equasion. . Both seemed fine to me and as you said may come up to personal preference here and there. This kind of nitpicking is certainly not worth it here where the difference in price between nvidia and amd is a LOT bigger than in the US. (up to 200 Euro for similar performance) . Maybe also mention that the RTX 3 series get left behind
You don't need to slow it down. You can see the differences in person easily. Bitrate on RUclips makes it harder to notice and slowing it down will make it even easier to see the difference
Месяц назад+6
And it's worst case scenario, in balanced or quality mode is so much better
DLSS its just the chiky trick Nvidia use to cut the cost of productions make more profit , and keep a reasonable ( in some case fake) framerate. Basically PC gamer do what the Console player ( and PC player always has hated ..ahaha) from ages.
@anrypettit3112 what? The entire point is that you don't need to slow it down or zoom if you play it yourself. I didn't bother with the rest as it makes no difference to the argument.
cool comparison video, vex. appreciate the time and effort you put into collecting all this photage/data. personally i would only ever consider using "quality" mode with upscalers, mostly cause i think with "performance" mode they just haven't got there yet
It's absolutely ridiculous that people paying 500$+ have to use upscaling in some poorly optimized games. There was a point in time when budget builds could handle 1080p no question not too long ago. We have the hardware still that can do this, people need to be alot more criticle of the crap we're being fed from developers in the ironically anti-consumer quest for more "shareholder value".
yeh missed those times man... dlss and fsr before is just a means for low end gpus to gain additional fps, it really be like that before... many are so happy that we can now afford a significant increase in fps without worrying about the quality that much but now, fookin hell 😂
@@MaloryHovell Honestly if I've had a time machine and could go back in time to get rid of one person, the fucker that thought of and implemented TAA would make top of the list... the shit is SO BAD...
FSR 3.1 Frame Generation with NO upscaling is freakin awesome. I’m running it this way in Ratchet & Clank: Rift Apart without all the FSR upscaling artifacts on my 6800M laptop GPU 😊. I also tested FSR 3.1 in Spider-Man Remastered, with upscaling and raytracing, and it’s excellent.
Please mate, real world applications if you test graphics. This slow mo effect matters to exactly no one. People use these upscaling techs to get higher fps, sometimes 100+ fps, Slowing down a video and zooming in is not what people are using this tech for. Most people want higher FPS with demanding games. Show effects and gameplay without slowing down the video. Many new graphical tech advancements use visual illusions and brain triggers because the eyes can be fooled. The brain can be fooled. One good example of this is how many frames per second a screen updates. Showing slow motion of a 200+ FPS game does exactly nothing. However a game looks in slow motion does not matter in an upscaling tech, since no games offers time warps. When we learn how to slow down time, this video may be more factual...until then, nothing in the video is what you can expect from any of upscaling tech.. If anyone want to play witcher 3 in 120 fps instead of 85 fps, use these upscaling techs. If you wonder how the grains and weird ghosting looks when sped up to 150fps, its not noticeable. these techs are used to fool the eye and they work just fine. @ 24:11 is a good example of how these techs look when used.
@@untrainedprofessionals2374I agree, and i think for the masses that are not buying gpus that cost like cars. These upscaling techs vill be as easy to use for gaming as the vsync technology. Especially for consoles that will benefit even more than pcs from better software implementation.
for me though it is to curb down power consumption and temps when playing at flat 60 FPS.. these techs are doing wonders to keep my PC from overheating!
@@Cosmicdonut6764 It being hard to tell at full speed is the real point though. It's not meant to be scrutinized to such an extent, but be passable in gameplay while offering better performance. The tech really isn't designed to be used with absolutely zero sharpening either.
This man used to promote amd perf/dollar so much that so many amd fans came to him and now when he is saying something bad about them they are triggered so much
I couldn't care less for both technologies. I seldom use them as I play at 1080p. To me they're both fine and I will always pick the better price to performance card.
Yep same here bought a RX 6800 about a year or so back since it was on sale. I use it for 1080p high FPS, I dont use upscaling at all. Once i get the performance i want for a reasonable price i am fine.
FSR 3.1 IS INCREDIBLE THOUGH. Now with 3.1 you can just use framegeb without terrible upscaling and TAA which sucks ass. Free FPS. My rx 580 plays this amazing horizon forbidden west game without issue at 60fps
@@DrLogic_ Obviously... I agree that is what you use it for. But the more commonly used settings are 1080P Quality mode and 1440P Balanced or Quality. At least for those I know that use it regardless of which brand card they own.
Well, FSR is hardware agnostic so... it shouldn't matter. Maybe FSR4.0 will require hardware component and work only with series 7000 and up (since AMD has AI accelerator cores, just not using it for anything...) and then there would be tangible benefits of using AMD card with FSR specifically. Most people own some low-mid range Nvidia GPU that do not support DLSS so it's literally the largest target audience for FSR - bigger than AMD users since we are in the fringe minority.
@@Micromation It may in theory be agnostic, but still, games are supposed to be agnostic too, but one game may perform better on an AMD than on a Nvidia or vice versa. I just read about a monitor (Samsung G8 I believe) that worked better on Nvidia, though that has been solved as far as I know. So to be sure I think you should test it on the hardware it was developed on, there may be a (small) difference, so it would be more fair to test it on an AMD IMHO.
@@nitrogenius You are thinking the stock price, market cap. I specifically said revenue. You know, the thing that actually matters. Can the company actually make money by selling products. Not the fantasy land of Wall St valuation and investor hype. Can the company actually make money. And Intel makes more than Nvidia and AMD combined.
@@halycon404 I think you have no idea that how many gpu is sold by Nvdia to Datacenters including azure,aws and cloud by AI hype. In 2023 Nvdia has more revenue than intel 60b vs 54b. Intel may have more stacked cash but it does not long much. AI is real thing that evolves everything and intel has nothing to do with their cpu's.
@@nitrogenius No. In financial year 2023 by Nvidia's own investor call they grossed 26.9 billion with a net income of 8.3 billion. Those are Nvidia's own numbers which you can pull from Nvidia's earnings report for shareholders. It's a downloadable PDF. Intel had 54.2 billion revenue, a net revenue of only 1.1 billion but 2023 was also a reinvestment year. They spent over 10 billion in capital investment outlays for retooling their fabrication plants. That's also Intel's own numbers which are a downloadable PDF off Intel's website. That 60 billion is not an Nvidia number. It's what investors hope fiscal year 2024 looks like.
If you're at the point where you need to slow things down, zoom way in, put it side by side, are you really playing the game at all? I think that's a huge win for FSR and XeSS in terms of all GPU's being able to utilize it lol.
Nah on 1080p you can tell a lot of difference especially when moving fsr has shimmering more when moving around. I play on 4k and I can tell even at that resolution when moving left and right with the camera, which feels smoother, and it was dlss most of the time. FSR either oversharpens or is not sharp enough, I'd rather have anti-aliasing that I don't have to worry about and it's already been tested by hardware unboxed which one is smoother and stable, Spoiler it's dlss. I'm no fanboy but it's a fact. Majority of games with dlss tested runs better on dlss. In terms of image stability mostly and at times performance.
@@cloud380 Huh, don't you know RUclips has bitrate compression? I said that I can't see any big difference solely based on how it looks on this RUclips video I've tested both and yeah, obviously DLSS has better reconstruction on fine details, but higher ghosting than FSR, it all depends on the use scenario, so no need to get all salty and come saying "bs" lol
I still believe upscalling is to get the jump from 1440p to 4k gamming with your new shiny 4k monitor, but most graphics card can't handle 4k native yet. If you are using it to upscale in performance 1080p then you realy can't complain, there are no miracles there and quality will suffer badly. Other use case is to get raytracing to more acceptable performance in good cards. Yes i know most people want to get their old graphics card to live a couple more years, but it's realy all about pushing 4k to the mainstream market...
Gonna be honest, it's probably another instance of the marketing team screwing the development team over again. They said by the end of Q2 and the engineers were like "y'all, it's still not ready yet!" and AMD was like "sorry, can't hear you over the sound of us jettisoning the crap out the door!". Honestly, what the engineers have managed to achieve without AI IS genuinely impressive, let no one take that away from them. But... I just don't think it's ever going to compare. They shot themselves in both feet when they, for literally no reason, decided to not use AI for upscaling purposes, and now I'm just assuming they're flopping around until the 8000 series comes out, which is just supposed to be 7000 series, but like... actually good. Most likely they'll have AI cores that actually do shit, FSR will finally actually be worth more than my spare toenail clippings, and then the races can FINALLY begin for real.
The difference are NULL AND VOID while actually gaming... You had to PAUSE and ZOOM IN to find the issues and nobody thats gaming is noticing any of that....
@dorion9111 this isn't true. You can tell them apart when playing easily. The reason for zoom in and slow down is due to compression and bitrate of RUclips. Unless you have poor vision you can tell straight away in game
@@lilpain1997 I mean kinda? At 1440P FSR 2.1 Quality mode in many games you can't always tell when it's even on outside of getting higher FPS. 3.1 is just that much better now. I think nvidia's last stronghold in this area is aggressive upscaling on lower end cards at this point.
@charlesearl4321 disagree, I use 3440x1440 and standard 2560x1440 for anything that doesn't support my 21:9. I can tell fsr2.1 is on instantly even at quality. It's poor overall. 3.1 is a step in the right direction though and better.
agreed, fsrQ still is pretty bad, just not as bad as before, but its still pretty jarring to see while in game, it was the reason i sold my 7900XT. Eventually we will get to a point where there is no difference in image quality, but for that amd needs to invest in ML to close that gap, and that wont happen until their gpu's have the hardware for it.
@@ishiddddd4783The 7000 series actually has hardware AI acceleration, AMD just can't really build their upscaler based on hardware that maybe a third of their active cards have.
I'm still excited for it. Mainly for the vulkan support, and decoupled frame gen. This means that one mod that would let you convert dlss framegen to fsr framegen should now be possible for many more games such as Portal RTX, allowing my 3090ti to get framegen in it.
Your explanation of what these technologies do is very amateur. FSR and DLSS are super sampling methods. Taking a low res. image and scaling it to fit whatever res. screen you have is just that, scaling. It happens when I connect an Xbox 360 (or any non native res.) to my 4k TV by the display. Not very impressive although there are better and worse scaling techniques. FSR and DLSS upsample the image, adding visual information to the original render and outputting to a higher resolution. DLSS stands for deep learning (machine learning) super sampling. Not trying to be rude but educating your viewers on how these things work in greater detail would add value to your channel. Cheers.
Not really, DLSS will always be ahead. AMD is doing what they alway do, see nvidia do something new then follow it. Issue is they are always behind doing this.
The thing too is they need FSR to be viable because all the modern consoles (PS5 and XSX) run on AMD chips and they wanna push the framerates while the games look good :b
I wish you had compared the higher settings too. The performance mode looks like crap on all upscalers. It's when you get to about 75% (i.e. 1080p->1440p) that it starts looking acceptable. For 720p->1440p I would probably prefer integer scaling over any of these upscalers and just pretend that I have a 720p display. It's only the most insane setup that would not be CPU bottlenecked at 720p, which would defeat the purpose of upscaling, since lowering resolution doesn't help your CPU.
Regardless of the output quality by all upscaling tgechnologies whether it be DLSS, FSR, XeSS or Lossless Scaling etc ... they all have their benefits or drawbacks! Having a choice of upscaler is a great thing for the consumer regardless as choice gives each user their desired outcome which imo is a great thing. I generally use DLSS .... but FSR 3.1 has got me contemplating Radeons upscaling solution as an alternative as had Lossless Scaling :D Great video Vex and thank you for covering these technologies.
Generally they only have drawbacks... the only difference is how big they are. The only benefit DLSS has is that it can be combined with DLDSR which has no equivalent on AMD or Intel cards but it's basically a workaround to a problem that doesn't need to exist (really bad console oriented anti-aliasing in modern games with no alternatives)
As an AMD user, I always go for Xess. It just looks better than FSR in most games. Sometimes FSR looks worse than even TSR. It's really dissapointing. AMD seriously need to up their game in terms of upscaling. Maybe, like Intel, go the AI route, and make 2 versions, one open-source and one strictly for AMD GPUs?
As long as you have to compare "performance" modes of the upscalers to spot the difference, the quality of upscaling is good enough for me. Because "performance" mode is there only as a last option for really desperate usecases. Most of the time, balance or even quality presets are being used by players.
Great video, I like how you literally zoom and point out stuff. No one else does this, make sit easier to understand and see stuff that I usually miss. Keep it up! Subscribed
I feel like what is happening with FSR is that whomever is working on it needs to learn to rest their eyes when working on their upscaler Its the same thing when you are mixing and mastering a song. Sometimes you need to let your ears rest so you can get the best mix possible I say this because i feel like its the same type of scenario with the devs working on FSR. If they take time in between and rest their eyes, they could better see if a change is an improvement or not. Im sure looking at the same ghosting images and editing code alongside that can tire your eyes out very quickly from comparing before and after constantly. To the point that it all starts looking the same to the person working on it. It would help to explain why there is such little differences between the two, and how in some scenarios FSR 2 looks better. I know its a weird comparison, but im sure someone is tracking what im trying to say
AMD should start working on AI based upscaling since, lot of handhelds use AMD APUs. Also switch 2 comes with DLSS and would crush all AMD handhelds in visual quality. With all the AI stuff being stuffed onto everyone, may be it is time for AMD to put good use of AI and use it in upscaling the most. RT improvements are also required.
The amount of detail that FSR 3.1 brings is way better than fsr 2.2 not sure why vex is double guessing himself here cause the difference is obvious to me however some of my friends have a hard time telling 1440p from 4k so I guess I understand were some people come from because they cannot tell
In practice the difference between upscalers is pretty marginal, no one plays games in super slow motion at high magnifications. This video is nothing more than making a mountain out of an ant hill.
@@arenzricodexd4409 Upscaling will always have artifacts, you are after all inventing pixels, you shouldn't expect miracles from it and it's not gonna get much better than this because the cost/quality just gets too high, simply rendering the actual pixels becomes cheaper.
@@utahnl that's why this war for "which upscaling is better" is stupid in the first place. There are other ways of improving the performance without introducing those artifacts. Funny thing is people are now obsessed with FPS number again with things like FG. i thought VRR was supposed to make those high FPS to be irrelevent for most gamer other than competitive games where they want that high FPS for latency reason.
If dlss fsr xess are all turned on performance mode, the most important function should be to increase fps. The purpose of players turning on this function is because the gpu is not powerful enough. Why not compare fps? And there is no comparison between balance and quality mode. Just stare at the performance mode to see the picture quality. Does anyone slow down and zoom in to see the picture? ? ?
Performance mode with FSR is extremely noticeable at 1440p and 4k, meanwhile DLSS manages to make a surprisingly coherent image, image quality does matter and so does fps.
in my opinion DLSS FSR etc they only make sense when there is a lot of information so tv/monitor 2160p and only quality option (1440p>2160p) This can look even better than TAA DLSS FSR on 1080p native has completly zero sense, and 1440p... ok maybe quality option...
I've always said this. The only reasonable use case of upscaling to 1080p is really old hardware and handhelds like the steam deck or rog ally. Otherwise it should be 1440p quality or in 4k quality/balanced. People act like DLSS is magic but using performance mode at 1080p destroys the image too so.. What's the point?
Yea, but you will have zealots telling you upscalling is the best thing since sliced bread since it can give your old GPU a second life. Like, homie, these features look like garbage on budget hardware. If you REALLY want to play Helldivers 2 and you can only afford RX6600, FSR performance will allow you to have stable 60fps at 1080p. It will also give you headache, nausea and after few hours you will want to gouge your eyes out. Anything less than 4k and upscalling is falling off the cliff and introduces weird artifacts. Same with frame generation - literally unplayable if you're not getting 60-70fps MINIMUM native... but it's time and time again, people saying shit on the internet. Just like recommending 3060 over 6600XT because MUH RAY TRACING... no homie, you ain't using RT in anything but Minecraft (and maybe Portal IF you combine it with upscalling at 1080p LMAO), calm down. I honestly hate this timeline... it feels like console wars in PS360 era and bunch of morons cheering for their corporate overlords o.0 TL;DR these are NOT the features for poor people and developers already leaned on it way too hard (Remnant 2 runs like GARBAGE on top end hardware...) to compensate their ineptitude.
@@thomaswhite9160 when PPI is good then yes scalling can help like in Steamdeck, but imho steamdeck is not a hardware to play AAA games, its shines in emulation retro games, metroidvania like hollow knight etc etc, and maybe older AAA games from ps360 era, where those tech like FSR is no needed. people swallow marketing and that's sad
@@V1lkas I mean, you can play AAA games, not amazingly, but you can.. I wouldn't do a full playthrough of a game there, but to play a bit a way from home is quite ok. I have an ROG Ally and during a trip I played a bit of Elden ring, sure not the whole playthrough, but a bit on a weekend when I was away from my desktop PC, and just recently I also played a bit of Ghost of Tsushima on the ally, and it's serviciable... Again, not to have as your ONLY means of playing games, but to play bits and pieces, it isn't bad.
All of the talk around DLSS/FRS, framegen, RTX and stuff confuse the hell out of me. I can kind of see it for 4k if you need it to hit some fps "bracket" right? but what's the point of using DLSS if say you're already at like 120 fps at 1440p? Doesn't native look better? does going from 120 fps native to 150 fps DLSS/FSR make a difference in the fluidity of single player games? do we even notice it when actually playing instead of looking at screenshots or videos at 25% speed? is the performance gain so big that it justifies losing graphical fidelity when you're already over 60/120/240 fps? I feel like we're running in circles now just measuring fps and pixels for the hell of it.
Maybe. These upscalers can be used for anything to your liking. So if your running the game at 50fps you could turn on upscaling to get to 60fps with hopefully with little to no difference. Or if you already have a good frame rate, you could get extra performance to then increase settings like raytracing. The point is more performance with good image quality. So for testing this you just look closer at the image to find out the differences. Which is why people zoom in and slow down the game to better see the differences. Also because of youtube compression.
Ofc one of the problem is that you need a higher hz monitor to benefit from upscaling. most older monitors are 60 hz and newest tech gamer monitors goes up to 540 hz. That would, if you had a gpu that could drive such a beast, be 540 different frames rendered every second. The most important thing is to utilize as much of the monitors hz as possible, if your monitor has a cap of 120 hz...you will see it used to its maximum @ 120 fps. moving the mouse cursor very fast on the screen is a perfect example of how fps looks. the trails of the mouse becomes alot more fluid on higher frames per second. and if on a lower hz monitor, it will be choppy and the mouse cursor will lag and pop up randomly everywhere.
I consider it to be the least important and something that should be given the soviet treatment. Can I just get raw performance and companies releasing properly ready and complete games instead of beta tests with a years patching?
In my situation I can hardly even tell. Like even when slowed down and zoomed in I have to focus so hard that my eyes were hurting throughout this vid. Sure there were some moments where it was noticeable but it's not as big of a difference as Vex makes it out to be imo.
I can tell the difference about 50% of the time. And I don’t know about you, but only managing 50% distinction of the time for a test that was frankly, very unfair and one-sided considering the level of up scaling he was doing (which no one will be doing) and the graphics card he was using for both tests (definitely not an AMD card), I would say that you may pretty much conclude it was the same thing.
It's kinda sad that you didn't try other presets as well and just went with performance. I mean I get it you wanted to try the worst case scenario to make differentiating easier but I for one never use anything but FSR quality since it's kinda pointless going lower than that with considering fsr capabilities.
I keep thinking that there is some agreement that amd and Nvidia made that amd would keep their upscaler from competing with Nvidia's. Nvidia is now the most valuable company in the world.. agreements are made outside of the public eyes.. you better believe that is the truth. Evem intel comes out from the start with a better upscaler and amd has been working on it as long as nvidia. It just makes no ducking sense guys.
AMD uses a "dumb" upscaling approach, there's nothing more to it than static math deciding how it upscales compared to a more sophisticated (but computationally costlier) approach where an AI algorithm trained on desirable outcomes analyses frames & decides which approach will likely result in the desirable outcome. You can't really compete with that unless the costlier approach makes a gigantic blunder.
@@MLWJ1993 would you say that Intel uses something similar to Nvidia but the instruction uses less die space and less energy to run the algorithm? I suppose you're right with reguards to the dumb approach... AMD could potentially get better image quality but it would be more GPU intensive making it less frame and thus defeating the purpose. If AMD did not use an instruction to accelerate upscaling next gen.. 8000 will be a fail.
@christophermullins7163 Intel uses 2 different approaches within XeSS. One only works on Intel Arc GPU's & is the XMX version, which is similar to Nvidia's approach. Then there's the DP4A mode that's a thinned out version of XMX that will run on any GPU capable of using DP4A (which is all fairly recent GPU's). However DP4A instructions cost you quite a bit more performance to get a result inbetween DLSS/XeSS XMX & FSR.
Honestly, dlss just have a strong case, especially for people who likes to play on higher resolution, or like to plug their pc to a 4k TV and use upscaling. However, the visual difference between fsr and dlss in "quality" mode is really really minor, so using fsr in 4k quality mode or even 1440p quality mode is something that is very reasonable, if you don't have the budget to get a 4070ti or a 4080, of if you just prefer amd and go with the mid to high end gpus beginning with rx 6700xt up to 7900 xtx.
Like who cares about slow motion just to see the difference? as long as it performs like DLSS with a bit of flaws it does not matter. AMD is getting better and better in making FSR competitive to DLSS I have to complaints the fact that low to medium budget graphics card can now enjoy the game with higher fps.
your brain isn't braining is it? he is slowing down the motion so you can see the difference he is seeing in a normal gaming session because of youtube is making the video quality worse so he has to slow motion the difference for you to be able to see it.
I would love to see a review from a different pov. Instead of both rendered at the same low res and upscaled at 1440p, try to make their fps on par. For example if fsr3.1 have a higher fps, then afjust dlss to match that one. And it goes the same with fsr3.1 if it has a lower fps. Then compare their qualities at that fps.
Since XeSS has an AI upscaler open for every GPU (DP4a version), the next step for FSR *should* be integrating machine learning, because i think there isn't too much improvement window left from 3.1. And let's not forget that RDNA3 have it's own AI accelerating cores that AMD still don't use for anything, there is room for an "XMX" version of FSR to help them sell Radeon (even iGPUs).
Would this not make a difference using an AMD card it was designed to work with? It looks like you are just using a 4070 ti for both, which was designed to work with DLSS.
Maybe everyone should start doing good optimization again just like old times because it looks like everyone just forget about such a words and now simple aiming for some"technologies" to improve overall gameplay but to be honest more and more of latest games have a bad optimization and a tons of bugs with graphics especially with lights
While I accept that DLSS probably still has the edge, I really don't think there is as much difference as is being made out here. Other RUclips channels have done side by side for FSR3.1 and DLSS across the same games. Straight screen captures at 60fps 1440p/4k and I can't tell the difference. I don't pixel peep and slow my game to 15 fps when I play. So long as the game looks good and I get a stable +60fps I couldn't care less about the minor details you are pointing out here. TBH I couldn't even notice some of the issues that were being discussed in this video.
I've seen comparisons using 4k performance and FSR 3.1 holds up literally very good to DLSS. Also with 1440p Quality it seems to get very close to DLSS on quality too. I wish he tested that.
RUclips compression is butchering these videos hence why slowing down makes more accurate representation for the viewer. For best effect you want to watch video in 4k, even if you have 1080p monitor because 4k videos on YT have higher bitrate (and thus less blockiness and compression related artifacts). So it isn't exactly slowing down and pixel peeping because that's the limitation of watching these comparisons on video. In person it's much more obvious. Personally I can't stand any upscalling... the thing she points out, like ghosting and artifacting are really jarring.
@@Micromation Just for clarification I have both AMD and NVIDIA systems so I'm not a fanboy of either. I buy my systems for best price to performance at any given time. Having used both scalers on their retrospective GPUs I can honestly say I struggle to see much difference in the way DLSS and FSR perform when running both on quality at 1440p or 4k when regular gaming. I really think we're at a point now where the upscalers are so close in performance that it shouldn't be the deciding factor as to whether you go team green or red. Heck, go team blue if you think that offers the best bang for buck to you.
@@madcowgimbo I tend to avoid upscaling, unless absolutely necessary. It's actually one of main reasons why I've opted out of 4k and went into ultrawide OLED 1440p instead. Can't stand DLSS, XeSS or FSR and the artifacts and shimmering they produce so I'd rather have more FPS and native resolution instead. I also do have both AMD and Nvidia cards.
what if entire time that high end users never have any ai cores in 7900xt, 7800xt, 7900xtx as reason amd don't want to talked about it when they already fired radeon group marketing team for just following orders from lisa su?? there should be a lawsuit...people spend over $900 bucks to have filler tiles and amd falsely call it ai cores is real reason AMD have not mention it anymore and not use it due to odd shape die that tsmc use filler tiles to make complete shape for cooler support. AMD fans should sue amd for theft again...since amd fx
I think AMD thought that they could get away with not using AI/Deep learning with their upscale and are now just grabbing staws. While I love AMD CPUs I don't understand their GPU strategy. It's always like this Nvidia releases a new feature like CUDA, RT, or DLSS and then AMD tries to play catchup as cheaply as possible. CUDA is now 17 years old yet for creative types it's still the only worthwhile option. RT and DLSS are 6 years old yet AMD still doesn't put RT cores in their GPUs and their upscaler still doesn't use AI. I can understand them not being able to do as much research as Nvidia and therefore not creating new features but even their catchup game is shit. They got beaten by Intel in RT and upscaled on their first try. Sure Intel GPUs are currently only budget cards but that won't stay the case. If AMD doesn't step up their game and Intel beats them on raster performance for a lower price AMD can pretty much pack their bags.
I mean if you need to slow down the game by half and zoom-in to spot differences and issues there is no issue because in motion while gaming normally you won't see them. Pizel peeping at its finest. "Huge freakin' difference", no it is not, not while actually gaming.
True, and I think that's the main flaw a lot of these test are doing, they are assuming we are playing games with binoculars and in slow motion, clearly if you go out of your way to look for flaws, you will find them in any tech, regardless of how good the tech is. What really matters is the perceived difference in quality under normal play conditions, basically under the conditions that 99.9% of gamers are likely to play their games with, and I suspect most gamers don't care how an image looks when zoomed in or under slow motion, they are more interested in how the game looks under normal play conditions, which a lot of these tech sites and RUclipsrs seems to be getting wrong. DLSS is still overall better, but FSR 3.1 has closed the gap that I suspect for most gamers, it doesn't matter which upscaling tech is used, but with that said, I wouldn't use any upscaling tech unless it's 1440p or above resolution and quality mode, going below that and all these upscaling tech show their flaws, same with frame gen, 60fps or above. With that said, if you are playing on the small screen like with a Steam Deck, you can get away with more in lowering the resolution and frame rate, but more testing is needed on that. What you really want to see with these test is normal zoom level, without any slow motion and a good high quality video stream, that would be far more useful to 99.9% of gamers than all this zooming in and slow motion, which can make any tech look better or worse than another, but what really matters is the perceived difference between the different tech to most gamers, if the gap becomes so small under normal play to most gamers, then the difference between them all doesn't matter much. Honestly, if you really want to test these tech out, you really are better off testing them yourselves, because a lot of tech sites and RUclipsrs are doing it all wrong which doesn't represent normal play conditions, and don't get me wrong, I've no issues with zooming in and slow motion, but more coverage really needs to be shown where it matters under normal play conditions, as I suspect that matters far more to gamers then zooming in and slow motion, to put it another way, these videos are kinda pointless.
As i put before in another comment. You don't need to do eitherof those in person. RUclips isnt great as it compresses the video... the vid will get compressed before as well. Bitrate aint great on RUclips. Slowing and zooming helps with either of those things.
I'm grateful for these recent technological advancements. Many of us grew up playing with 8 blobs of pixels/object then moved on to 480p then 720p before it got okay starting from FHD and now 4K looks great. I don't expect images to be flawless when upscaled from old-time resolutions, I mainly expect better FPS and pay the price with some minor artifacting. It's a fair exchange and much better than a native low res. How I see it for the same price we either buy a stronger AMD card and run chosen resolution natively or buy a weaker Nvidia and run that res with dlss quality. it's roughly the same image for the same money. Better native FPS at AMD, better RT at nVidia.
youtube compression is a b***h
tests on a AMD GPU would as be good here. Its hilarious you expected and improvement in you upsacle 720 to 1440p using FSR. It needs at least upscale 1080p to 1440p to see the improvement
Beach ✅
It is all very obvious anyway. Compression did NOT hide the artifacts 100%
that's why i upscale my video to 4K even though the video itself is lower than that. at least the higher bitrate of 4K can retain some details.
Even premium bitrate doesn't help. Need an extreme zoom for this kind of video
Framegen and upscaling is nice to have and all but i just want devs to actually optimize their fucking games.
💯
"Here are some extra frames the machina thinks you want to see"
True .. the graphic today is not that mindblowing. It just poorly optimized
@@syarifairlangga4608exactly bruh I’m so sick of it
Do you really think that the industry has that many good developers? This is not a problem of just Game Development its generally a problem of all Developing Industries, they even hire college students. People want Games released fast This is what you get.
@@kCenk LOL the prosumer's fault?👀👀👀👀👀👀
The industry does it because they can get away with it. The cash cow erupted in 2020 and they'll never look back as long as people keep coughing up bucks a transactions.
edit- "prepurchase today!" 🤪
1 minute and no "bro you fell off ahh" comment?
The comment section fell off.
The most accurate thing ever from nice vids bro to bro 99 views in 6 hours bro fell oh
8hrs and no "8 hrs and no"8hrs and x views comments fell off" comment? Comments fell off
yeah iirc fh5 is next so, soon i guess
Not a surprise that upscaling and framegen advances so slow, it probably hurts the sales of new and expensive GPUs in the market.
Specially if the new products its Garbo and the marketing have convinced people play on PC like on console its the way to go.
Nvidia will continued down this route : 90-80 series good gen jump , 70-60 garbo products with terrify prices and a software let people play like on console ( its upscaling from ages). The people keep cheer on .... sadge
@@matttiaz7576 glad I'm not the only one that noticed.
@@matttiaz7576i thought the 4070 was good
@@CJTheTokay its a "60 series" by Die size with a pathetic memory bandwith will last no more than a year. But thats its all Nvidia want it : make you change GPU every year.
@@CJTheTokay😂😂😂😂
One day in the future, and that day will come, trust me, Nvidia will make you pay mothly subscribtion for DLSS.
this mark it
Yeah I'm looking forward to the reception for that. Bet people will still cheer it on and gladly pay for it
Right now it’s embedded into the selling price as we’re seeing minimal generational uplift at the cost of featurset
They are already charging for it semi annually. DLSS is great, FG is awesome, except on the card I bought two years ago, it isn't worth a damn on that.
I really think comparing the two is sort of apples to oranges. FSR is great in that it can keep older cards relevant, DLSS mostly works on cards that shouldn't need it.
Bruh 💀
AMDs biggest struggle are transparencies and particle ghosting.
you play whit 500%zoom in and motion whit 25% slowed gameplay?
@@elfabrote No, I play at 4k. I also tested it myself.
You'd think because of the ghosting they wouldn't have transparency problems but
@@elfabroteyou can notice it ingame specially on grass
It doesn't exist with 3.1. framegen can run without upscalinf
I hope they implement FSR 3.1 in the new Frostpunk game, they worked close with AMD and designed limited FP2 style GPU's
I hope AMD just stop to follow this stupid nosense trend Nvidia promote , and launch a good and reasonable priced new gen.
@@matttiaz7576rx 8000 series, plus prices are getting cheaper these days for the 7000 and 6000 series
@@matttiaz7576frame gen is amazing?!?!
With 3.1 there is no longer a need to use fsr2 upscaling. So it's pretty much a 50% boost
Sorry Vex, I disagree with you on this (Personal opinion), I think Ancient Gameplays did a better comparison, FSR 3.1 is definitely better than 2.2 and I CAN TOTALLY SEE the difference while playing Horizon Forbidden West, it's much more stable. 1440p FSR Balance used to look terrible and now is decent and in QUALITY MODE I hardly even notice the difference from native res. Also, without zooming in and slowing down you won't even notice anything whatsoever because people just don't do that. Don't get me wrong, FSR 3.1 is not a game changer (It's not even using dedicated hardware), it still has flaws, DLSS is still better but it is becoming insignificant at this point.
If it does end up using dedicated hardware then you can expect most cards that are supported now to suddenly become unsupported, then people will really hate AMD and almost fully abandon them.
@@ArtisChronicles That's not true, XESS has a fallback code and it looks good without dedicated hardware, FSR could do that and move on with machine learning, it's really impresive how far in quality FSR has come with broad compatibility.
The zooming in part and slowing down is clearly for purpose of making a video and relaying what he sees to the viewer so they can see it. It's an academic approach. It can also make up for RUclips's compression quality losses.
Out of all criticisms I see the gripes with slowing down footage and zooming in is the strangest I have to admit. I mean the goal is to nitpick /review.
Sounds like cope to me knowing full well that Ancient has an extreme AMD bias
@@silvio351thing is what intel did with XeSS is much more complicated than nvidia DLSS. and AMD push FSR because they did not want to deal with ML stuff like nvidia did with DLSS. Now you're expecting AMD to take even more complicated method like intel XeSS?
With this new technologies, games are slowly becoming like video streaming quality.
you mean with fsr because dlss is damn good solution,amd is years behind
@@Jaba-hs6vd No, i mean all this new techs. I don't think this is the right way to improve the gaming future.
@@TheBlackAnomaly it doesn't matter what you mean,dlss is perfect solution to gain performance with minimal hit on image quality,fsr is a short term solution from AMD until they find the resources to add full ai hardware acceleration which it will happen in rdn 5
@@Jaba-hs6vd it really shows how little AMD understands what is happening. Nvidia already had AI acceleration back when AMD didn't even have DP4a and mesh shaders support. it was other way around back in GCN1-2 architecture days
@@TheBlackAnomalythis is towards the good direction. Fsr 3.1 DOES NOT require upscaling, so the quality is native. No antialiasing forced.
Honestly, yes, DLSS is still better, but the difference in quality, playing at least at 1440p Quality or even 1440p Balanced, is negligible. DLSS was the last thing that still kept me take Nvidia into consideration. Now that FSR is in practice at the same level of DLSS, Nvidia to me is irrelevant.
Yep. Indeed dlss is better but let's be honest.. for the same price you can run a game at native with AMD and quality upscaling with Nvidia and get similar performance. Only when you have a mid range GPU on a 4k screen running the latest games slare you required to upscale and upscaling looks better at 4k by far. It's a tossup as to whether Nvidia GPUs are better overall. Tradeoffs are made unless you're rich.
FSR is not anywhere better than DLSS if not in really specific scenes, conditions in specific games; to say something similar you're doing a really forced cherry picking. As he says, FSR 3.1 barely holds up to the worse version of XeSS 1.3 (22:18). FSR can be passable at something like 4k quality, maybe 1440p quality sometimes. It of course depends on what you consider to be a satisfying visual quality and on how much you notice visual breakups.
@@matte_bar So, just use XeSS. Radeons CAN do that you know. 😁
@@AvroBellow it is still far behind dlss dude, don’t act like a dumb amd fanboy, be objective
@@AvroBellow Xess is worse in none arc gpus.
The candles not moving is hilarious.
it made it temporaly stable :D
@@HybOj LITERALLY
Why not test quality mode? And also give us the impression you get when you’re not zoomed in three times and slowed down to 60%?
Because in youtube 60fps lock and low bitrate, normal gameplay would not be as telling of the details he wants to point out. Wich all the those details add up to be noticeable in real life gameplay.
Quality mode won't be noticeable
Because he’s trying to see the total difference, quality is less easy to tell. Sure most people use quality I’m sure, but this is the best way to do this.
@@Cosmicdonut6764 agreed
@@TheFansvideo '' give us the impression you get '' doesn't mean that he needs to show the gameplay 😂
Nobody uses Performance mode unless you're playing in 4K. For 1440p or 1080p try to use Quality or Balance mode atleast.
100%
Not true, I play at 1440p and I do use performance mode (720p->1440p) a lot, depends on a game tho. In some games I am using performance mode only, like cp2077 - can't tell difference, so why not have more fps. In some games even quality mode looks noticeably worse then native, so no upscale there. Best thing tho is to use DLSSTweaks mod and choose base resolution for upscale yourself, in my opinion.
I use DLSS Quality for 4k on a 4070ti Super mostly with ultra settings. I get 90 to 120 fps, and that's perfectly fine for me
I don't think any 4k gamer would settle for less then quality
@@agentnukaz1715 then you are mistaken not every 4k gamer has a top tier gpu and also 4k performance is still better than 1440p quality. 4k is a really good resolution for using upscalers.
The moment you have to slow down to show the oh so 'noticeable' differences you can ask yourself if dlss is really worth the additional tax we pay to Nvidia and you smartly avoid that part of the equasion. .
Both seemed fine to me and as you said may come up to personal preference here and there.
This kind of nitpicking is certainly not worth it here where the difference in price between nvidia and amd is a LOT bigger than in the US. (up to 200 Euro for similar performance) .
Maybe also mention that the RTX 3 series get left behind
You don't need to slow it down. You can see the differences in person easily. Bitrate on RUclips makes it harder to notice and slowing it down will make it even easier to see the difference
And it's worst case scenario, in balanced or quality mode is so much better
DLSS its just the chiky trick Nvidia use to cut the cost of productions make more profit , and keep a reasonable ( in some case fake) framerate.
Basically PC gamer do what the Console player ( and PC player always has hated ..ahaha) from ages.
@@lilpain1997 so... you "smartly" avoid the tax part of the equasion too?
@anrypettit3112 what? The entire point is that you don't need to slow it down or zoom if you play it yourself. I didn't bother with the rest as it makes no difference to the argument.
cool comparison video, vex. appreciate the time and effort you put into collecting all this photage/data. personally i would only ever consider using "quality" mode with upscalers, mostly cause i think with "performance" mode they just haven't got there yet
It's absolutely ridiculous that people paying 500$+ have to use upscaling in some poorly optimized games.
There was a point in time when budget builds could handle 1080p no question not too long ago. We have the hardware still that can do this, people need to be alot more criticle of the crap we're being fed from developers in the ironically anti-consumer quest for more "shareholder value".
yeh missed those times man... dlss and fsr before is just a means for low end gpus to gain additional fps, it really be like that before... many are so happy that we can now afford a significant increase in fps without worrying about the quality that much but now, fookin hell 😂
Moreover even if you don't use upscaler you still get a blurry mess cos devs force you to use TAA in their titles.
Isn't there the rtx 4060, 3060, 3060 ti, rx 6700 xt, 7600xt,7600 ?
Or we just like more fps. I always use DLSS at 1440p with my 4070ti I paid $850 for. Prefer fps over resolution always.
@@MaloryHovell Honestly if I've had a time machine and could go back in time to get rid of one person, the fucker that thought of and implemented TAA would make top of the list... the shit is SO BAD...
FSR 3.1 Frame Generation with NO upscaling is freakin awesome. I’m running it this way in Ratchet & Clank: Rift Apart without all the FSR upscaling artifacts on my 6800M laptop GPU 😊. I also tested FSR 3.1 in Spider-Man Remastered, with upscaling and raytracing, and it’s excellent.
Please mate, real world applications if you test graphics. This slow mo effect matters to exactly no one.
People use these upscaling techs to get higher fps, sometimes 100+ fps, Slowing down a video and zooming in is not what people are using this tech for.
Most people want higher FPS with demanding games.
Show effects and gameplay without slowing down the video. Many new graphical tech advancements use visual illusions and brain triggers because the eyes can be fooled. The brain can be fooled. One good example of this is how many frames per second a screen updates. Showing slow motion of a 200+ FPS game does exactly nothing.
However a game looks in slow motion does not matter in an upscaling tech, since no games offers time warps.
When we learn how to slow down time, this video may be more factual...until then, nothing in the video is what you can expect from any of upscaling tech..
If anyone want to play witcher 3 in 120 fps instead of 85 fps, use these upscaling techs.
If you wonder how the grains and weird ghosting looks when sped up to 150fps, its not noticeable. these techs are used to fool the eye and they work just fine.
@ 24:11 is a good example of how these techs look when used.
I am personally flat out telling the difference. I would think these upscalers are all pretty good and will allow people to jump from 1440 to 4k.
@@untrainedprofessionals2374I agree, and i think for the masses that are not buying gpus that cost like cars. These upscaling techs vill be as easy to use for gaming as the vsync technology. Especially for consoles that will benefit even more than pcs from better software implementation.
This is the correct way to do it, kinda hard to see a difference at 120fps full speed now isn’t it?
for me though it is to curb down power consumption and temps when playing at flat 60 FPS.. these techs are doing wonders to keep my PC from overheating!
@@Cosmicdonut6764 It being hard to tell at full speed is the real point though. It's not meant to be scrutinized to such an extent, but be passable in gameplay while offering better performance. The tech really isn't designed to be used with absolutely zero sharpening either.
But nobody uses FSR or any upscaler on performance mode, so this video really doesn't answer the question "is now FSR 3.1 good enough to use?"
You obviously did not heard of pc handheld devices.
This man used to promote amd perf/dollar so much that so many amd fans came to him and now when he is saying something bad about them they are triggered so much
Amd fanboys are hands down far worse than most fanboys only out done by Sony ponys
AMD is COOKED cooked.
I couldn't care less for both technologies. I seldom use them as I play at 1080p. To me they're both fine and I will always pick the better price to performance card.
Even if I play at 1440p I prefer not to bother with them
I actually would take it one step further and say that I _actively_ _dislike_ this tech. It's just been leading game developers to be lazy.
@@ArtisChronicles
I use them when gaming at 4k other than that I never use them
@@aqualung2000the biggest example of this to me is Ark Survival Ascended that game is unoptimized as heck I have trouble playing it on my 1660 Ti. 😭
Yep same here bought a RX 6800 about a year or so back since it was on sale. I use it for 1080p high FPS, I dont use upscaling at all. Once i get the performance i want for a reasonable price i am fine.
think it was a mistake upscaling from 720 p
Seems like an nvidia commercial
Every time when I hear FSR out done DLSS, I laugh.
FSR 3.1 IS INCREDIBLE THOUGH. Now with 3.1 you can just use framegeb without terrible upscaling and TAA which sucks ass. Free FPS. My rx 580 plays this amazing horizon forbidden west game without issue at 60fps
Wasn't the title ''did amd just destroy dlss3'' or somthing like that lol ?
Why not test quality mode? And also give the impression you get when you’re not zoomed in three times and slow down to 60%?
Performance mode stresses the technology the most
@@DrLogic_ Obviously... I agree that is what you use it for. But the more commonly used settings are 1080P Quality mode and 1440P Balanced or Quality. At least for those I know that use it regardless of which brand card they own.
@@F1neW1ne this video is about showing the differences, it would be harder to show them in quality mode
Shouldn't you test AMD's newest thing on an AMD gpu?
My thoughts
Well, FSR is hardware agnostic so... it shouldn't matter. Maybe FSR4.0 will require hardware component and work only with series 7000 and up (since AMD has AI accelerator cores, just not using it for anything...) and then there would be tangible benefits of using AMD card with FSR specifically. Most people own some low-mid range Nvidia GPU that do not support DLSS so it's literally the largest target audience for FSR - bigger than AMD users since we are in the fringe minority.
@@Micromation It may in theory be agnostic, but still, games are supposed to be agnostic too, but one game may perform better on an AMD than on a Nvidia or vice versa. I just read about a monitor (Samsung G8 I believe) that worked better on Nvidia, though that has been solved as far as I know. So to be sure I think you should test it on the hardware it was developed on, there may be a (small) difference, so it would be more fair to test it on an AMD IMHO.
@@brulaapgaapmeester8052 lol you obviously have 0 clue how any of this works. Just say that 😊. Does NOT matter
XeSS shows how Intel handle the upscaling so well in a short time.
They threw money at it. Intel still has higher revenue than AMD and Nvidia combined. They can afford just to throw money at it.
@@halycon404 lol it is not true. The fact that now nvdia has more money than intel and amd combined
@@nitrogenius You are thinking the stock price, market cap. I specifically said revenue. You know, the thing that actually matters. Can the company actually make money by selling products. Not the fantasy land of Wall St valuation and investor hype. Can the company actually make money. And Intel makes more than Nvidia and AMD combined.
@@halycon404 I think you have no idea that how many gpu is sold by Nvdia to Datacenters including azure,aws and cloud by AI hype. In 2023 Nvdia has more revenue than intel 60b vs 54b. Intel may have more stacked cash but it does not long much. AI is real thing that evolves everything and intel has nothing to do with their cpu's.
@@nitrogenius No. In financial year 2023 by Nvidia's own investor call they grossed 26.9 billion with a net income of 8.3 billion. Those are Nvidia's own numbers which you can pull from Nvidia's earnings report for shareholders. It's a downloadable PDF. Intel had 54.2 billion revenue, a net revenue of only 1.1 billion but 2023 was also a reinvestment year. They spent over 10 billion in capital investment outlays for retooling their fabrication plants. That's also Intel's own numbers which are a downloadable PDF off Intel's website. That 60 billion is not an Nvidia number. It's what investors hope fiscal year 2024 looks like.
It's sad that the upscaling quality doesn't improve, but the decoupling of upscaling and frame generation alone is worth the update.
This is actually the best thing ever. I hate upscaling and TAA
If you're at the point where you need to slow things down, zoom way in, put it side by side, are you really playing the game at all? I think that's a huge win for FSR and XeSS in terms of all GPU's being able to utilize it lol.
Old-school gamers who played video games with unintended pixel art see all of this as a minor issue.
I tested bots and now I cant see any big difference between DLSS and FSR, both have strenghts and weaknesses
Nah on 1080p you can tell a lot of difference especially when moving fsr has shimmering more when moving around. I play on 4k and I can tell even at that resolution when moving left and right with the camera, which feels smoother, and it was dlss most of the time. FSR either oversharpens or is not sharp enough, I'd rather have anti-aliasing that I don't have to worry about and it's already been tested by hardware unboxed which one is smoother and stable, Spoiler it's dlss. I'm no fanboy but it's a fact. Majority of games with dlss tested runs better on dlss. In terms of image stability mostly and at times performance.
bs i have 4070 ti super and i tested both fsr 3 and dlss 3.5 fsr on my 4k native looks worse than dlss 3.5 in 1080 p on my 4k display.
@@cloud380 Huh, don't you know RUclips has bitrate compression? I said that I can't see any big difference solely based on how it looks on this RUclips video
I've tested both and yeah, obviously DLSS has better reconstruction on fine details, but higher ghosting than FSR, it all depends on the use scenario, so no need to get all salty and come saying "bs" lol
DLSS is better, but native rendering is always the best.
I think you should have done 1080p to 1440p comparison because for the new improvements to show, there needs to be enough pixels to work on
agreed.
The great thing about far 3.1 is that you can use Frame Gen with whatever upscaling that you want.
I still believe upscalling is to get the jump from 1440p to 4k gamming with your new shiny 4k monitor, but most graphics card can't handle 4k native yet. If you are using it to upscale in performance 1080p then you realy can't complain, there are no miracles there and quality will suffer badly. Other use case is to get raytracing to more acceptable performance in good cards. Yes i know most people want to get their old graphics card to live a couple more years, but it's realy all about pushing 4k to the mainstream market...
Nvidia is using a hardware solution and AMD is trying a software solution and it isn't competing, not sure if it ever will tbh.
Gonna be honest, it's probably another instance of the marketing team screwing the development team over again. They said by the end of Q2 and the engineers were like "y'all, it's still not ready yet!" and AMD was like "sorry, can't hear you over the sound of us jettisoning the crap out the door!".
Honestly, what the engineers have managed to achieve without AI IS genuinely impressive, let no one take that away from them. But... I just don't think it's ever going to compare. They shot themselves in both feet when they, for literally no reason, decided to not use AI for upscaling purposes, and now I'm just assuming they're flopping around until the 8000 series comes out, which is just supposed to be 7000 series, but like... actually good. Most likely they'll have AI cores that actually do shit, FSR will finally actually be worth more than my spare toenail clippings, and then the races can FINALLY begin for real.
The truth is that they all work pretty well. When in use nobody zooms in to peak at pixels. It's not very useful.
Love the clickbait titles but DLSS still destroys FSR in many games. Look at Hardware Unboxed video.
8:05 - talking smack about FSR meanwhile DLSS pouch is having complete meltdown 💀
Definitely a game issue, as FSR had the same issue for a second there
Upscaler is only good from 1440p onwards, anything below it is trash as there's not enough native details for the upscaling to be accurate.
I honestly can't stand DLSS or FSR at 1440p on quality setting, let alone any of the performance options...
@@Micromationjust turn in framegen only.
Yep. How will an upscaler work if there is no detail.
Wow, what a huge difference ( watching in 144p )
NVIDIA have official won the gpu wars again...👑👑👑
they always should considering they are way more expensive more often than not.
I may get console wars, but GPU war is really low
The difference are NULL AND VOID while actually gaming... You had to PAUSE and ZOOM IN to find the issues and nobody thats gaming is noticing any of that....
@dorion9111 this isn't true. You can tell them apart when playing easily. The reason for zoom in and slow down is due to compression and bitrate of RUclips. Unless you have poor vision you can tell straight away in game
@@lilpain1997 I mean kinda? At 1440P FSR 2.1 Quality mode in many games you can't always tell when it's even on outside of getting higher FPS. 3.1 is just that much better now. I think nvidia's last stronghold in this area is aggressive upscaling on lower end cards at this point.
@charlesearl4321 disagree, I use 3440x1440 and standard 2560x1440 for anything that doesn't support my 21:9. I can tell fsr2.1 is on instantly even at quality. It's poor overall. 3.1 is a step in the right direction though and better.
agreed, fsrQ still is pretty bad, just not as bad as before, but its still pretty jarring to see while in game, it was the reason i sold my 7900XT.
Eventually we will get to a point where there is no difference in image quality, but for that amd needs to invest in ML to close that gap, and that wont happen until their gpu's have the hardware for it.
@@ishiddddd4783The 7000 series actually has hardware AI acceleration, AMD just can't really build their upscaler based on hardware that maybe a third of their active cards have.
I'm still excited for it. Mainly for the vulkan support, and decoupled frame gen. This means that one mod that would let you convert dlss framegen to fsr framegen should now be possible for many more games such as Portal RTX, allowing my 3090ti to get framegen in it.
its pretty great but indeed night raytracing abit hard to comperhend, compared to day which is easy/pleasing to see straight away
Your explanation of what these technologies do is very amateur. FSR and DLSS are super sampling methods. Taking a low res. image and scaling it to fit whatever res. screen you have is just that, scaling. It happens when I connect an Xbox 360 (or any non native res.) to my 4k TV by the display. Not very impressive although there are better and worse scaling techniques. FSR and DLSS upsample the image, adding visual information to the original render and outputting to a higher resolution. DLSS stands for deep learning (machine learning) super sampling. Not trying to be rude but educating your viewers on how these things work in greater detail would add value to your channel. Cheers.
Bro AMD is caching up fast I cant wait for the day that AMD will be better than Nvidia
That will probably not happen
NV is not as stupid as Intel, NV will be always on step ahead of AMD
your kidding right?
Not really, DLSS will always be ahead. AMD is doing what they alway do, see nvidia do something new then follow it. Issue is they are always behind doing this.
@@lilpain1997 One day in the future, and that day will come, trust me, Nvidia will make you pay mothly subscribtion for DLSS.
The thing too is they need FSR to be viable because all the modern consoles (PS5 and XSX) run on AMD chips and they wanna push the framerates while the games look good :b
I wish you had compared the higher settings too. The performance mode looks like crap on all upscalers. It's when you get to about 75% (i.e. 1080p->1440p) that it starts looking acceptable. For 720p->1440p I would probably prefer integer scaling over any of these upscalers and just pretend that I have a 720p display. It's only the most insane setup that would not be CPU bottlenecked at 720p, which would defeat the purpose of upscaling, since lowering resolution doesn't help your CPU.
ghosting on monitor will be way worse
Regardless of the output quality by all upscaling tgechnologies whether it be DLSS, FSR, XeSS or Lossless Scaling etc ... they all have their benefits or drawbacks! Having a choice of upscaler is a great thing for the consumer regardless as choice gives each user their desired outcome which imo is a great thing. I generally use DLSS .... but FSR 3.1 has got me contemplating Radeons upscaling solution as an alternative as had Lossless Scaling :D
Great video Vex and thank you for covering these technologies.
but dlss 3 being better isn't benefits nor a drawback... It is to show how inferior the others are. I won't even talk how bad lossless scaling was.
Generally they only have drawbacks... the only difference is how big they are. The only benefit DLSS has is that it can be combined with DLDSR which has no equivalent on AMD or Intel cards but it's basically a workaround to a problem that doesn't need to exist (really bad console oriented anti-aliasing in modern games with no alternatives)
As an AMD user, I always go for Xess. It just looks better than FSR in most games. Sometimes FSR looks worse than even TSR.
It's really dissapointing. AMD seriously need to up their game in terms of upscaling. Maybe, like Intel, go the AI route, and make 2 versions, one open-source and one strictly for AMD GPUs?
Maybe fsr4.1 amd might use ai upscaling for once
@@JahonCrossdo you have an AMD card?
@waltuhputurdaway currently have a nividia card
I miss the days when Rasterization was King.
As long as you have to compare "performance" modes of the upscalers to spot the difference, the quality of upscaling is good enough for me. Because "performance" mode is there only as a last option for really desperate usecases. Most of the time, balance or even quality presets are being used by players.
Great video, I like how you literally zoom and point out stuff. No one else does this, make sit easier to understand and see stuff that I usually miss. Keep it up! Subscribed
I feel like what is happening with FSR is that whomever is working on it needs to learn to rest their eyes when working on their upscaler
Its the same thing when you are mixing and mastering a song. Sometimes you need to let your ears rest so you can get the best mix possible
I say this because i feel like its the same type of scenario with the devs working on FSR. If they take time in between and rest their eyes, they could better see if a change is an improvement or not. Im sure looking at the same ghosting images and editing code alongside that can tire your eyes out very quickly from comparing before and after constantly. To the point that it all starts looking the same to the person working on it. It would help to explain why there is such little differences between the two, and how in some scenarios FSR 2 looks better.
I know its a weird comparison, but im sure someone is tracking what im trying to say
AMD should start working on AI based upscaling since, lot of handhelds use AMD APUs. Also switch 2 comes with DLSS and would crush all AMD handhelds in visual quality. With all the AI stuff being stuffed onto everyone, may be it is time for AMD to put good use of AI and use it in upscaling the most. RT improvements are also required.
The amount of detail that FSR 3.1 brings is way better than fsr 2.2 not sure why vex is double guessing himself here cause the difference is obvious to me however some of my friends have a hard time telling 1440p from 4k so I guess I understand were some people come from because they cannot tell
DLSS 3 is currently better , idk for how long tho , FSR 3 input lag is very annoying
This is why I run native res even at the sacrifice of frame rates
If u cant run it Native , Dont run it at all.
Weirdo
I’am so fracking proud of you kid!! You are doing an amazing job keep up the great work! I really appreciate and enjoy the videos.
In practice the difference between upscalers is pretty marginal, no one plays games in super slow motion at high magnifications. This video is nothing more than making a mountain out of an ant hill.
People did not care that but people do care the artifact caused by these upscaler especially in fast moving camera or environment.
@@arenzricodexd4409 Upscaling will always have artifacts, you are after all inventing pixels, you shouldn't expect miracles from it and it's not gonna get much better than this because the cost/quality just gets too high, simply rendering the actual pixels becomes cheaper.
@@utahnl that's why this war for "which upscaling is better" is stupid in the first place. There are other ways of improving the performance without introducing those artifacts. Funny thing is people are now obsessed with FPS number again with things like FG. i thought VRR was supposed to make those high FPS to be irrelevent for most gamer other than competitive games where they want that high FPS for latency reason.
If dlss fsr xess are all turned on performance mode, the most important function should be to increase fps. The purpose of players turning on this function is because the gpu is not powerful enough. Why not compare fps? And there is no comparison between balance and quality mode. Just stare at the performance mode to see the picture quality. Does anyone slow down and zoom in to see the picture? ? ?
Performance mode with FSR is extremely noticeable at 1440p and 4k, meanwhile DLSS manages to make a surprisingly coherent image, image quality does matter and so does fps.
in my opinion DLSS FSR etc they only make sense when there is a lot of information
so tv/monitor 2160p
and only quality option (1440p>2160p)
This can look even better than TAA
DLSS FSR on 1080p native has completly zero sense, and 1440p... ok maybe quality option...
I've always said this. The only reasonable use case of upscaling to 1080p is really old hardware and handhelds like the steam deck or rog ally. Otherwise it should be 1440p quality or in 4k quality/balanced.
People act like DLSS is magic but using performance mode at 1080p destroys the image too so.. What's the point?
Yea, but you will have zealots telling you upscalling is the best thing since sliced bread since it can give your old GPU a second life. Like, homie, these features look like garbage on budget hardware. If you REALLY want to play Helldivers 2 and you can only afford RX6600, FSR performance will allow you to have stable 60fps at 1080p. It will also give you headache, nausea and after few hours you will want to gouge your eyes out. Anything less than 4k and upscalling is falling off the cliff and introduces weird artifacts. Same with frame generation - literally unplayable if you're not getting 60-70fps MINIMUM native... but it's time and time again, people saying shit on the internet. Just like recommending 3060 over 6600XT because MUH RAY TRACING... no homie, you ain't using RT in anything but Minecraft (and maybe Portal IF you combine it with upscalling at 1080p LMAO), calm down. I honestly hate this timeline... it feels like console wars in PS360 era and bunch of morons cheering for their corporate overlords o.0
TL;DR these are NOT the features for poor people and developers already leaned on it way too hard (Remnant 2 runs like GARBAGE on top end hardware...) to compensate their ineptitude.
@@thomaswhite9160 when PPI is good then yes scalling can help like in Steamdeck, but imho steamdeck is not a hardware to play AAA games, its shines in emulation retro games, metroidvania like hollow knight etc etc, and maybe older AAA games from ps360 era, where those tech like FSR is no needed.
people swallow marketing and that's sad
@@V1lkas I mean, you can play AAA games, not amazingly, but you can.. I wouldn't do a full playthrough of a game there, but to play a bit a way from home is quite ok. I have an ROG Ally and during a trip I played a bit of Elden ring, sure not the whole playthrough, but a bit on a weekend when I was away from my desktop PC, and just recently I also played a bit of Ghost of Tsushima on the ally, and it's serviciable... Again, not to have as your ONLY means of playing games, but to play bits and pieces, it isn't bad.
All of the talk around DLSS/FRS, framegen, RTX and stuff confuse the hell out of me. I can kind of see it for 4k if you need it to hit some fps "bracket" right? but what's the point of using DLSS if say you're already at like 120 fps at 1440p? Doesn't native look better? does going from 120 fps native to 150 fps DLSS/FSR make a difference in the fluidity of single player games? do we even notice it when actually playing instead of looking at screenshots or videos at 25% speed? is the performance gain so big that it justifies losing graphical fidelity when you're already over 60/120/240 fps?
I feel like we're running in circles now just measuring fps and pixels for the hell of it.
Maybe. These upscalers can be used for anything to your liking. So if your running the game at 50fps you could turn on upscaling to get to 60fps with hopefully with little to no difference. Or if you already have a good frame rate, you could get extra performance to then increase settings like raytracing. The point is more performance with good image quality. So for testing this you just look closer at the image to find out the differences. Which is why people zoom in and slow down the game to better see the differences. Also because of youtube compression.
Ofc one of the problem is that you need a higher hz monitor to benefit from upscaling. most older monitors are 60 hz and newest tech gamer monitors goes up to 540 hz.
That would, if you had a gpu that could drive such a beast, be 540 different frames rendered every second.
The most important thing is to utilize as much of the monitors hz as possible, if your monitor has a cap of 120 hz...you will see it used to its maximum @ 120 fps. moving the mouse cursor very fast on the screen is a perfect example of how fps looks. the trails of the mouse becomes alot more fluid on higher frames per second. and if on a lower hz monitor, it will be choppy and the mouse cursor will lag and pop up randomly everywhere.
you wont notice a big difference going from 120 to 150,. but going from 60 to 120 is a easy to spot. it becomes very fluid.
I consider it to be the least important and something that should be given the soviet treatment.
Can I just get raw performance and companies releasing properly ready and complete games instead of beta tests with a years patching?
Most of the people won't care about these details in games.
A number of channels have found it to be a mild improvement - but I think they were mostly using it at 4k, which obviously limits its usefulness.
In my situation I can hardly even tell. Like even when slowed down and zoomed in I have to focus so hard that my eyes were hurting throughout this vid. Sure there were some moments where it was noticeable but it's not as big of a difference as Vex makes it out to be imo.
This is most likely a hot take, but I can’t tell the difference between fsr and dlss
I can tell the difference about 50% of the time. And I don’t know about you, but only managing 50% distinction of the time for a test that was frankly, very unfair and one-sided considering the level of up scaling he was doing (which no one will be doing) and the graphics card he was using for both tests (definitely not an AMD card), I would say that you may pretty much conclude it was the same thing.
I much prefer XeSS over FSR these days
I don't like upscaling, it makes some stuff inside the game looks softer
It's kinda sad that you didn't try other presets as well and just went with performance. I mean I get it you wanted to try the worst case scenario to make differentiating easier but I for one never use anything but FSR quality since it's kinda pointless going lower than that with considering fsr capabilities.
I keep thinking that there is some agreement that amd and Nvidia made that amd would keep their upscaler from competing with Nvidia's. Nvidia is now the most valuable company in the world.. agreements are made outside of the public eyes.. you better believe that is the truth. Evem intel comes out from the start with a better upscaler and amd has been working on it as long as nvidia. It just makes no ducking sense guys.
AMD uses a "dumb" upscaling approach, there's nothing more to it than static math deciding how it upscales compared to a more sophisticated (but computationally costlier) approach where an AI algorithm trained on desirable outcomes analyses frames & decides which approach will likely result in the desirable outcome.
You can't really compete with that unless the costlier approach makes a gigantic blunder.
@@MLWJ1993 would you say that Intel uses something similar to Nvidia but the instruction uses less die space and less energy to run the algorithm? I suppose you're right with reguards to the dumb approach... AMD could potentially get better image quality but it would be more GPU intensive making it less frame and thus defeating the purpose. If AMD did not use an instruction to accelerate upscaling next gen.. 8000 will be a fail.
even regular comment think this way
@christophermullins7163 Intel uses 2 different approaches within XeSS. One only works on Intel Arc GPU's & is the XMX version, which is similar to Nvidia's approach.
Then there's the DP4A mode that's a thinned out version of XMX that will run on any GPU capable of using DP4A (which is all fairly recent GPU's). However DP4A instructions cost you quite a bit more performance to get a result inbetween DLSS/XeSS XMX & FSR.
Did AMD just kill DLSS? No.
Bye.
Honestly, dlss just have a strong case, especially for people who likes to play on higher resolution, or like to plug their pc to a 4k TV and use upscaling. However, the visual difference between fsr and dlss in "quality" mode is really really minor, so using fsr in 4k quality mode or even 1440p quality mode is something that is very reasonable, if you don't have the budget to get a 4070ti or a 4080, of if you just prefer amd and go with the mid to high end gpus beginning with rx 6700xt up to 7900 xtx.
Haha I’m watching this on my phone at 360p and can’t see the difference in anything
Really I don't see any difference because I am seeing it 360p 😅
DLSS looks kinda choopy in slow motion footage.
oh noes! All the people who bought the Am5, Zen 4 Ryzen 7 7800 X3D get to FEEL GOOD about their purchase now!!
Like who cares about slow motion just to see the difference? as long as it performs like DLSS with a bit of flaws it does not matter. AMD is getting better and better in making FSR competitive to DLSS I have to complaints the fact that low to medium budget graphics card can now enjoy the game with higher fps.
your brain isn't braining is it? he is slowing down the motion so you can see the difference he is seeing in a normal gaming session because of youtube is making the video quality worse so he has to slow motion the difference for you to be able to see it.
I think it's great I'll be able to buy a lower-end PC and compensate with FSR 3
1:26 The one on the left is FSR.
I would love to see a review from a different pov. Instead of both rendered at the same low res and upscaled at 1440p, try to make their fps on par. For example if fsr3.1 have a higher fps, then afjust dlss to match that one. And it goes the same with fsr3.1 if it has a lower fps. Then compare their qualities at that fps.
Since XeSS has an AI upscaler open for every GPU (DP4a version), the next step for FSR *should* be integrating machine learning, because i think there isn't too much improvement window left from 3.1. And let's not forget that RDNA3 have it's own AI accelerating cores that AMD still don't use for anything, there is room for an "XMX" version of FSR to help them sell Radeon (even iGPUs).
Would this not make a difference using an AMD card it was designed to work with? It looks like you are just using a 4070 ti for both, which was designed to work with DLSS.
FSR is the same on every GPU, and is designed to work on all gpus,so your point is meaningless
If I want better grafics I just have to put my glasses on.
Maybe everyone should start doing good optimization again just like old times because it looks like everyone just forget about such a words and now simple aiming for some"technologies" to improve overall gameplay but to be honest more and more of latest games have a bad optimization and a tons of bugs with graphics especially with lights
While I accept that DLSS probably still has the edge, I really don't think there is as much difference as is being made out here. Other RUclips channels have done side by side for FSR3.1 and DLSS across the same games. Straight screen captures at 60fps 1440p/4k and I can't tell the difference. I don't pixel peep and slow my game to 15 fps when I play. So long as the game looks good and I get a stable +60fps I couldn't care less about the minor details you are pointing out here. TBH I couldn't even notice some of the issues that were being discussed in this video.
I've seen comparisons using 4k performance and FSR 3.1 holds up literally very good to DLSS. Also with 1440p Quality it seems to get very close to DLSS on quality too. I wish he tested that.
RUclips compression is butchering these videos hence why slowing down makes more accurate representation for the viewer. For best effect you want to watch video in 4k, even if you have 1080p monitor because 4k videos on YT have higher bitrate (and thus less blockiness and compression related artifacts). So it isn't exactly slowing down and pixel peeping because that's the limitation of watching these comparisons on video. In person it's much more obvious. Personally I can't stand any upscalling... the thing she points out, like ghosting and artifacting are really jarring.
@@Micromation Just for clarification I have both AMD and NVIDIA systems so I'm not a fanboy of either. I buy my systems for best price to performance at any given time. Having used both scalers on their retrospective GPUs I can honestly say I struggle to see much difference in the way DLSS and FSR perform when running both on quality at 1440p or 4k when regular gaming. I really think we're at a point now where the upscalers are so close in performance that it shouldn't be the deciding factor as to whether you go team green or red. Heck, go team blue if you think that offers the best bang for buck to you.
@@madcowgimbo I tend to avoid upscaling, unless absolutely necessary. It's actually one of main reasons why I've opted out of 4k and went into ultrawide OLED 1440p instead. Can't stand DLSS, XeSS or FSR and the artifacts and shimmering they produce so I'd rather have more FPS and native resolution instead. I also do have both AMD and Nvidia cards.
Another scam fsr from amd
the longer AMD work on FSR without ML support the deeper the hole they are digging to themselves
what if entire time that high end users never have any ai cores in 7900xt, 7800xt, 7900xtx as reason amd don't want to talked about it when they already fired radeon group marketing team for just following orders from lisa su?? there should be a lawsuit...people spend over $900 bucks to have filler tiles and amd falsely call it ai cores is real reason AMD have not mention it anymore and not use it due to odd shape die that tsmc use filler tiles to make complete shape for cooler support. AMD fans should sue amd for theft again...since amd fx
I think AMD thought that they could get away with not using AI/Deep learning with their upscale and are now just grabbing staws. While I love AMD CPUs I don't understand their GPU strategy. It's always like this Nvidia releases a new feature like CUDA, RT, or DLSS and then AMD tries to play catchup as cheaply as possible. CUDA is now 17 years old yet for creative types it's still the only worthwhile option. RT and DLSS are 6 years old yet AMD still doesn't put RT cores in their GPUs and their upscaler still doesn't use AI. I can understand them not being able to do as much research as Nvidia and therefore not creating new features but even their catchup game is shit. They got beaten by Intel in RT and upscaled on their first try. Sure Intel GPUs are currently only budget cards but that won't stay the case. If AMD doesn't step up their game and Intel beats them on raster performance for a lower price AMD can pretty much pack their bags.
I think they call it fsr3 instead of fsr3.1 because you will be able to replace the dlls manually later with a updated version
Yes, it is true that fsr 3.1 received minimal improvements. However, it is undeniable that fsr 3.1 is the worse looking upscaler at the moment.
I mean if you need to slow down the game by half and zoom-in to spot differences and issues there is no issue because in motion while gaming normally you won't see them. Pizel peeping at its finest. "Huge freakin' difference", no it is not, not while actually gaming.
True, and I think that's the main flaw a lot of these test are doing, they are assuming we are playing games with binoculars and in slow motion, clearly if you go out of your way to look for flaws, you will find them in any tech, regardless of how good the tech is.
What really matters is the perceived difference in quality under normal play conditions, basically under the conditions that 99.9% of gamers are likely to play their games with, and I suspect most gamers don't care how an image looks when zoomed in or under slow motion, they are more interested in how the game looks under normal play conditions, which a lot of these tech sites and RUclipsrs seems to be getting wrong.
DLSS is still overall better, but FSR 3.1 has closed the gap that I suspect for most gamers, it doesn't matter which upscaling tech is used, but with that said, I wouldn't use any upscaling tech unless it's 1440p or above resolution and quality mode, going below that and all these upscaling tech show their flaws, same with frame gen, 60fps or above.
With that said, if you are playing on the small screen like with a Steam Deck, you can get away with more in lowering the resolution and frame rate, but more testing is needed on that.
What you really want to see with these test is normal zoom level, without any slow motion and a good high quality video stream, that would be far more useful to 99.9% of gamers than all this zooming in and slow motion, which can make any tech look better or worse than another, but what really matters is the perceived difference between the different tech to most gamers, if the gap becomes so small under normal play to most gamers, then the difference between them all doesn't matter much.
Honestly, if you really want to test these tech out, you really are better off testing them yourselves, because a lot of tech sites and RUclipsrs are doing it all wrong which doesn't represent normal play conditions, and don't get me wrong, I've no issues with zooming in and slow motion, but more coverage really needs to be shown where it matters under normal play conditions, as I suspect that matters far more to gamers then zooming in and slow motion, to put it another way, these videos are kinda pointless.
So what? These kind of reviews make the tech better by basically complaining to them till they make it perfect. Being a nitpicky MF is a good thing.😂😂
I do see it instantly while gaming. I see FSR fizzle all the time and DLSS is always a clear improvement for me.
As i put before in another comment. You don't need to do eitherof those in person. RUclips isnt great as it compresses the video... the vid will get compressed before as well. Bitrate aint great on RUclips. Slowing and zooming helps with either of those things.
Man, believe me, u really see it, use FSR quality in the First Descendant a new game free in Steam, ghosting shimmering everywhere.
Upscaling will be a crutch for badly optimised games should not encourage this 😢
I'm grateful for these recent technological advancements. Many of us grew up playing with 8 blobs of pixels/object then moved on to 480p then 720p before it got okay starting from FHD and now 4K looks great.
I don't expect images to be flawless when upscaled from old-time resolutions, I mainly expect better FPS and pay the price with some minor artifacting. It's a fair exchange and much better than a native low res.
How I see it for the same price we either buy a stronger AMD card and run chosen resolution natively or buy a weaker Nvidia and run that res with dlss quality. it's roughly the same image for the same money. Better native FPS at AMD, better RT at nVidia.