WEBVTT 00:00:00.140 --> 00:00:06.899 frames win games we proved it NVIDIA 00:00:04.319 --> 00:00:13.799 advertises it and now we get to double how many we get for free NVIDIA says yes 00:00:11.099 --> 00:00:20.939 thanks to a brand new feature exclusive to 40 series gpus called dlss 3.0 like 00:00:18.300 --> 00:00:25.019 previous generations of dlss it improves your frame rate by trading off some 00:00:22.439 --> 00:00:30.539 image quality unlike old school dlss however dlss 3.0 works not by lowering 00:00:28.320 --> 00:00:36.780 the resolution but instead by pretending it's running faster than it really is 00:00:32.940 --> 00:00:38.940 sounds like BS but I promise you it's 00:00:36.780 --> 00:00:43.200 not what does it actually do should you turn it on and can we even tell the 00:00:41.160 --> 00:00:47.399 difference I'll tell you right after I tell you about our sponsor Boost Mobile 00:00:45.239 --> 00:00:53.520 Boost Mobile provides you with unlimited data talk and text for just 25 a month 00:00:51.120 --> 00:00:58.399 all on one of America's largest 5G networks click the link below for a 00:00:55.199 --> 00:00:58.399 limited time offer 00:01:05.659 --> 00:01:10.560 dlss 3.0 or more accurately deep 00:01:08.760 --> 00:01:15.060 learning frame generation which I'll call it from now on is NVIDIA's answer 00:01:12.720 --> 00:01:20.220 to low frame rates only officially available for NVIDIA's RTX 40 series 00:01:17.640 --> 00:01:24.960 gpus for now it mandates the use of NVIDIA's reflex low latency Tech and 00:01:22.560 --> 00:01:29.520 using an optical flow model on the GPU's tensor cores inserts fake frames in 00:01:27.780 --> 00:01:33.420 between the real ones that your GPU is actually rendering literally doubling 00:01:31.380 --> 00:01:37.259 your frame rate even without upscaling that's right it can be enabled without 00:01:34.799 --> 00:01:42.420 dlss which is why I won't be calling it that this obviously makes gameplay look 00:01:40.140 --> 00:01:47.939 a lot smoother but you might be thinking back to the battle days of the 2010s 00:01:44.939 --> 00:01:51.000 when TVs would advertise 120 hertz and 00:01:47.939 --> 00:01:54.060 240 hertz when all they really did was 00:01:51.000 --> 00:01:56.220 hold back multiple frames to blend them 00:01:54.060 --> 00:02:00.540 more or less by brute force that resulted in an often broken looking 00:01:58.259 --> 00:02:04.140 effects that added substantial input lag and was only really good for keeping 00:02:02.579 --> 00:02:08.399 track of where the puck is so you don't miss a sports ball hole in one what 00:02:06.060 --> 00:02:13.680 makes dlfg different is that it's making an educated guess at what the next frame 00:02:11.160 --> 00:02:17.220 should look like in near real time based on the previous frame and some motion 00:02:15.599 --> 00:02:24.000 vectors provided by the game engine which should look a lot better and be 00:02:20.760 --> 00:02:26.160 more responsive unfortunately dlfg isn't 00:02:24.000 --> 00:02:30.060 available in very many games right now so I'm going to pick two to closely 00:02:28.440 --> 00:02:34.980 examine starting with Marvel's Spider-Man remastered this game looks 00:02:32.760 --> 00:02:38.819 shockingly good with dlfg and able to give us double the frame rate though 00:02:36.900 --> 00:02:43.140 there are some things to take note of first off there's a sort of forward 00:02:41.280 --> 00:02:47.819 ghosting that happens from time to time as dlfg does its thing sort of the same 00:02:45.360 --> 00:02:51.540 way dlss 2.0 is ghosting happened you don't see it often but you can see it 00:02:49.560 --> 00:02:55.019 clearly here when Spider-Man jumps out of his apartment Window as a kind of 00:02:53.160 --> 00:02:58.739 Halo thanks to the game engine's insistence that he's moving but the way 00:02:56.940 --> 00:03:02.640 the camera moves counteracts the motion Vector data from the game engine in 00:03:00.720 --> 00:03:06.180 gameplay this manifests is a kind of Shimmer on the walk and wall crawling 00:03:04.500 --> 00:03:10.260 animations looking almost like a lighting error and when scenes change 00:03:08.340 --> 00:03:14.700 dramatically from shot to shot you'll find yourself staring at a frame made 00:03:12.180 --> 00:03:18.480 entirely of visual artifacts something common to all games compatible with 00:03:16.560 --> 00:03:21.959 frame generation right now and NVIDIA says it's a bug somewhere in the chain 00:03:20.159 --> 00:03:26.040 and the flag should be getting set that disables frame generation during such 00:03:23.519 --> 00:03:30.480 transitions it'll be fixed soon the more severe but paradoxically even less 00:03:28.379 --> 00:03:34.739 noticeable artifact is the breakup of sharp shapes when in front of or behind 00:03:32.879 --> 00:03:39.000 character models as digital Foundry pointed out the rooftop running part of 00:03:36.780 --> 00:03:42.540 the intro shows this off pretty well and same deal with the Running Animation 00:03:40.500 --> 00:03:46.440 when running up certain buildings my feelings are more or less the same as 00:03:44.280 --> 00:03:50.099 theirs unless you really know what you're looking for you might not even 00:03:48.000 --> 00:03:54.120 notice these artifacts especially since they look a bit like the kind of 00:03:51.900 --> 00:03:57.420 artifacts you'd get from motion blur or streaming video that you've 00:03:55.620 --> 00:04:00.720 subconsciously learned to ignore by now but anyway we'll see if our own Gamers 00:03:59.280 --> 00:04:06.959 can notice them later on in the video and honestly I would be surprised if they could this 00:04:04.980 --> 00:04:10.920 game has far more distracting visual anomalies that are unrelated to dlfg 00:04:09.239 --> 00:04:15.239 like the way its ambient occlusion crawls along certain scenes and how it 00:04:13.260 --> 00:04:19.320 renders even nearby background objects at a lower blurred resolution regardless 00:04:17.459 --> 00:04:23.280 of your depth of field settings all of which is to say that despite the pixel 00:04:21.060 --> 00:04:27.180 peeping we're doing here dlfg doesn't seriously harm image quality and those 00:04:25.620 --> 00:04:32.940 artifacts are gone in the next frame anyway they're only on screen for 00:04:28.979 --> 00:04:34.740 anywhere from 16.7 milliseconds to 8.3 00:04:32.940 --> 00:04:38.940 and even smaller as the frame rate climbs of course in order for frame 00:04:36.780 --> 00:04:43.919 generation to do its thing it really needs a certain frame rate to be usable 00:04:40.919 --> 00:04:45.479 NVIDIA recommended 40 to 60 is the 00:04:43.919 --> 00:04:49.020 Baseline when I ask them about it in terms of graphical artifacts while 00:04:47.340 --> 00:04:53.400 they're still minor they definitely got a lot more pronounced when I locked the 00:04:50.639 --> 00:04:57.540 frame rate to 60 instead of 120 and when I dropped it down even further to 30 you 00:04:55.440 --> 00:05:01.080 can start to see how much guesswork dlfg has to do as there's more changes 00:04:59.580 --> 00:05:05.940 between the frames like check out Spider-Man's mask in the this scene I 00:05:03.780 --> 00:05:11.040 had to see how deep that rabbit hole goes so I dropped it down once again to 00:05:08.280 --> 00:05:15.600 15 frames per second and that exposed even more weirdness check out how it 00:05:13.320 --> 00:05:20.460 handles the intro video here it looks like the GPU is failing like it's really 00:05:18.180 --> 00:05:24.300 weird the game clearly was never meant to run at this frame rate though there 00:05:22.560 --> 00:05:28.500 are some fascinating things that happen to the laptop display here and to Peter 00:05:26.639 --> 00:05:32.520 himself when he's flailing around and when he finally jumps out of the window 00:05:29.940 --> 00:05:37.139 it looks almost like 2 000 Sarah YouTube but higher resolution he even mostly 00:05:35.039 --> 00:05:40.440 disappears behind the railing and the railing doesn't fare much better the 00:05:38.880 --> 00:05:45.479 other game I tested for image quality is f122 and when it looks good it looks 00:05:43.320 --> 00:05:49.440 really good but when it looks bad it looks really bad right off the bat the 00:05:47.820 --> 00:05:54.300 thing you'll notice most is the thing that's front and center the driver name 00:05:51.240 --> 00:05:56.340 tags it's not such a huge deal at 120 00:05:54.300 --> 00:06:01.259 hertz but they Shimmer and shake at 60 FPS and things get worse from there I'm 00:05:59.280 --> 00:06:06.060 not sure why such small motions relatively speaking are causing eui to 00:06:03.660 --> 00:06:10.380 corrupt like this when it wasn't nearly as noticeable in Spider-Man even with 00:06:07.860 --> 00:06:15.060 fast motion but there it is another curious thing you'll notice is at the 00:06:12.419 --> 00:06:19.500 bottom of the screen even at 120 hertz you'll see Phantom pieces of the minimap 00:06:17.280 --> 00:06:23.460 flash on screen from time to time and if you look really closely you can make out 00:06:21.360 --> 00:06:27.720 what looks like an F1 car's tires with Optical camouflage applied to them when 00:06:25.319 --> 00:06:31.380 we crank the frame rate down you can see this way more readily it looks like 00:06:29.460 --> 00:06:35.220 NVIDIA trained the data set exclusively using the hood camera and possibly even 00:06:33.360 --> 00:06:39.360 on a single track while we're down in low FPS territory you can also see some 00:06:37.440 --> 00:06:43.139 more examples of ghosting and overshoot on the car's tires and sometimes even 00:06:41.160 --> 00:06:48.360 the front Wings start to overshoot ahead of the car it's really too bad because 00:06:45.840 --> 00:06:52.919 at high frame rates most of these issues are easily ignored and f122 itself would 00:06:51.120 --> 00:06:56.819 be an ideal candidate for gamers looking to crank the visuals on weaker Hardware 00:06:54.660 --> 00:07:00.539 because it relies more on anticipation than twitch aiming and it is often 00:06:58.860 --> 00:07:04.560 played with a controller after all it's far more tolerant of input lag than say 00:07:02.580 --> 00:07:08.880 a competitive shooter the same could be said of most simulators even PC Building 00:07:07.259 --> 00:07:13.080 simulator check out our Floatplane Exclusive by the way for the uncut Jake 00:07:10.440 --> 00:07:16.800 Vs Linus Shenanigans and what about input lag 00:07:14.759 --> 00:07:22.139 we were never going to get 120 hertz level input latency with 60 real frames 00:07:19.620 --> 00:07:25.500 and 60 fake frames at least in theory we should see roughly the same input 00:07:23.520 --> 00:07:29.340 latency as the original frame rate right due to the shortage of games available 00:07:26.940 --> 00:07:33.060 to test with and with Labs tied up with pre-testing for the upcoming rdna 3 00:07:31.319 --> 00:07:37.020 release we've only got input latency results for cyberpunk but we can already 00:07:35.160 --> 00:07:42.479 see an interesting pattern here the complete dlss 3.0 package provided input 00:07:39.900 --> 00:07:47.340 latency somewhere between dlss 2.0 and Native 4k on both our high-end and low 00:07:44.759 --> 00:07:51.000 end CPU benches which is expected due to the extra processing frame generation 00:07:48.960 --> 00:07:55.080 requires NVIDIA cited about an extra three milliseconds though in our case we 00:07:53.039 --> 00:07:59.819 measured more like four and a half or even as high as 13 on the core I3 that's 00:07:57.539 --> 00:08:04.740 almost a frame at 60 FPS for that one perhaps this is a driver optimization 00:08:01.919 --> 00:08:08.460 thing but it seems like even though it can boost frame rates and CPU bound 00:08:06.660 --> 00:08:13.199 scenarios it can only help responsiveness so much still it is lower 00:08:11.039 --> 00:08:17.400 than running at 4K native and if you're not quite a CPU bound you aren't likely 00:08:15.300 --> 00:08:21.360 to perceive the difference at all that brings up a good question we've looked 00:08:19.259 --> 00:08:25.500 at all of this under a microscope so far but how does it actually feel to play 00:08:23.759 --> 00:08:31.199 with does it make things significantly worse do people even notice when it's on 00:08:27.840 --> 00:08:33.060 and which do they prefer to find out we 00:08:31.199 --> 00:08:36.300 set up three identical test benches connected to three identical monitors 00:08:34.860 --> 00:08:41.279 each displaying a different configuration one rendering natively one 00:08:38.640 --> 00:08:47.100 using dlss set to Quality and one using dlss 3.0 with frame Generation all other 00:08:44.219 --> 00:08:50.820 settings were identical for guinea pigs we selected five subjects of varying 00:08:48.959 --> 00:08:54.720 familiarity with Graphics technology and asked them to play the games and rank 00:08:52.620 --> 00:08:59.279 each setup in order of visual Fidelity and responsiveness first up Marvel 00:08:56.820 --> 00:09:03.779 Spider-Man Captain 120 frames per second note that this means that dlss 3.0 was 00:09:01.860 --> 00:09:08.339 running at 60fps and doubling the frames using AI let's see what the normies 00:09:05.580 --> 00:09:12.480 think see I notice in this one here in the reflections there's like a bit more 00:09:10.680 --> 00:09:18.300 flickering okay this this one seems a little slower 00:09:15.180 --> 00:09:19.860 I think I do like this one this is like 00:09:18.300 --> 00:09:24.240 fast it's like someone hit the turbo button on the computer I'm gonna say 00:09:21.480 --> 00:09:27.180 that this middle one looks the best so the normies seemed to not be able to 00:09:25.740 --> 00:09:32.940 tell the difference but what about cultured and Discerning gamers 00:09:30.420 --> 00:09:36.200 I don't know they seem so similar now I'm not sure what I think 00:09:36.240 --> 00:09:42.959 it's funny how they have the same guy like in every single scene it's like the 00:09:41.220 --> 00:09:46.080 same face I'm looking at yeah I feel like that's the smoothest you can kind 00:09:44.459 --> 00:09:51.120 of see a little bit of Shimmer but I think that's just a setting things not necessarily 00:09:49.019 --> 00:09:54.480 um it's fun I would choose to play on this one and I don't know why and 00:09:53.100 --> 00:10:00.300 naturally we have to bring in someone who is relatively knowledgeable about displays and Graphics yeah I'm kind of 00:09:58.260 --> 00:10:04.399 the display guy I guess I do all the reviews for displays I own a display I 00:10:02.820 --> 00:10:07.500 got it crazy right 00:10:06.420 --> 00:10:13.320 no they're all really similar the UI 00:10:09.540 --> 00:10:15.720 elements they get garbled with DLS S3 I 00:10:13.320 --> 00:10:20.760 don't know man it kind of the same thing one two three and here are our results 00:10:18.480 --> 00:10:25.860 for visual Fidelity native rendering appears nearly indistinguishable from 00:10:22.680 --> 00:10:27.480 dlss's Ai upscaling and only when frame 00:10:25.860 --> 00:10:31.500 generation gets involved does quality seem to suffer though most subjects 00:10:29.640 --> 00:10:35.700 found it difficult to detect any difference as for responsiveness while 00:10:33.540 --> 00:10:39.300 we know that frame generation measurably hurts latency many of our subjects 00:10:37.620 --> 00:10:44.700 actually felt it was more responsive than dlss 2.0 which it literally is not 00:10:42.360 --> 00:10:49.380 so what you might be asking I'm playing a game at 120 FPS natively why would I 00:10:47.160 --> 00:10:54.240 bother using dlss that's a fair question how does the technology fare at 60 FPS 00:10:52.380 --> 00:10:58.079 we took our same subjects and ran the test again swapping the configurations 00:10:56.220 --> 00:11:01.700 for each monitor to try to account for any unknown bias preferring a specific 00:11:00.420 --> 00:11:06.720 monitor I don't know 00:11:05.040 --> 00:11:13.680 I don't know dude I feel like the middle one is the worst 00:11:11.760 --> 00:11:18.060 whatever I said yesterday I don't know if I was just kind of making it up in my 00:11:16.140 --> 00:11:22.320 head okay at this point I cannot tell the difference I noticed a weird 00:11:20.279 --> 00:11:24.540 artifact and like it kind of like it's hesitating it looks like it's hesitating 00:11:23.519 --> 00:11:31.079 right now um when I move the graphic scope potato 00:11:29.100 --> 00:11:35.820 the text isn't so bad but the marker itself gets garbled when I flick it 00:11:33.660 --> 00:11:42.000 around a little bit I'm I'm still liking this left one a lot at 60 hertz we see 00:11:39.300 --> 00:11:46.500 dlss 3.0 really start to suffer in both visual quality and latency most subjects 00:11:44.700 --> 00:11:51.420 quickly selected the frame generated rig is the worst at everything we did the 00:11:48.959 --> 00:11:55.680 same 60 FPS test in a plugtail Requiem and found similar results with nobody 00:11:53.579 --> 00:11:58.740 preferring the generated frames but what if we just uncap the frame rate and 00:11:57.300 --> 00:12:02.519 really let those systems fly into the triple digits well we did just that and 00:12:00.480 --> 00:12:06.839 interestingly we see frame generation take the lead in latency our guess is 00:12:05.220 --> 00:12:10.440 that many subjects interpreted the added smoothness of the generated frames as 00:12:08.459 --> 00:12:14.339 feeling more responsive at high frame rates dlss 3.0 does a great job of 00:12:12.839 --> 00:12:18.480 adding frames without introducing unplayable latency or immersion breaking 00:12:16.440 --> 00:12:22.260 artifacts at the lower frame rates well things are more of a compromise but 00:12:20.700 --> 00:12:27.420 here's something to note the only card available with dlss 3.0 is a 4090 and if 00:12:25.860 --> 00:12:31.800 there's any card that really doesn't need frame gen is that thing the real 00:12:30.180 --> 00:12:35.100 benefit of frame generation will be providing budget Gamers the ability to 00:12:33.540 --> 00:12:38.700 crank the settings and get frame rates that they typically would not be able to 00:12:36.600 --> 00:12:41.880 achieve so it was one last test we played a game of would you rather would 00:12:40.019 --> 00:12:46.260 you rather play your game without frame generation at 30 frames per second or at 00:12:44.459 --> 00:12:51.959 60 frames per second with frame generation probably want whatever you 00:12:49.019 --> 00:12:56.700 dlss thing you just put on okay this one automatically oh definitely the first 00:12:54.240 --> 00:13:03.420 option easily previous setting I think this is better yeah I really can't tell 00:12:59.240 --> 00:13:06.660 a huge difference I think I prefer this 00:13:03.420 --> 00:13:08.880 one I want smoother I prefer the first 00:13:06.660 --> 00:13:13.260 one oh this one hands down I'll take smoothness and artifacts every day if 00:13:11.040 --> 00:13:17.040 you went into this thinking dlfg was a dud or if you still do think that I 00:13:15.600 --> 00:13:21.600 wouldn't write it off so quickly think about what dlfg is doing say you're 00:13:19.200 --> 00:13:24.899 limited to 38 PS for whatever reason you can either have an experience that's 00:13:23.040 --> 00:13:28.380 full of what looks like stuttering or you can reduce visual Fidelity in a far 00:13:26.820 --> 00:13:33.779 more noticeable Way by reducing settings of resolution or you can turn on dlfg 00:13:31.500 --> 00:13:37.500 and look at a smooth 60fps with some very minor visual artifacts that might 00:13:35.579 --> 00:13:41.220 be masked by motion blur anyway your input latency is going to be similar to 00:13:39.000 --> 00:13:45.180 30fps but because you have smoother looking gameplay you'll be able to 00:13:43.260 --> 00:13:48.899 adjust your aim and track objects far more easily it won't save you in a 00:13:47.160 --> 00:13:52.560 competitive shooter of course but it'll turn single player games into a more 00:13:50.459 --> 00:13:56.880 enjoyable experience as two clicks Philip put it in a recent video we 00:13:54.420 --> 00:14:01.800 didn't really Embrace resolution scaling with dlss early on either but it's now 00:13:59.399 --> 00:14:05.639 ubiquitous and well accepted among Gamers I mean it's either that or you 00:14:04.019 --> 00:14:09.839 run at a lower resolution and let your monitor scale it the stupid way right 00:14:07.100 --> 00:14:13.440 dlfg is still in its infancy with bugs that are actively being hunted down and 00:14:11.579 --> 00:14:16.860 squashed and before you conclude that you'd never use it because you don't 00:14:14.820 --> 00:14:21.180 want any input lag at all two clicks Phillips video had another interesting 00:14:18.600 --> 00:14:25.260 angle that has me very excited for the future use 00:14:22.680 --> 00:14:30.060 Ella for VR to Dean put from the real frame rate making 00:14:27.420 --> 00:14:34.560 games playable even sub-60 it's a separate concept from dlfg but the demo 00:14:32.579 --> 00:14:39.060 built by comrade Stinger showcases the intended effect very well here we have 00:14:36.660 --> 00:14:45.000 30fps rendering but by enabling reprojection or async Time Warp power 00:14:42.240 --> 00:14:48.959 movements especially Mouse movements are smooth and responsive you can see in 00:14:47.160 --> 00:14:53.100 this mode that it works by basically projecting the scene in a kind of a 00:14:50.820 --> 00:14:56.399 virtual display that moves back to the center of the screen as each frame is 00:14:54.600 --> 00:15:00.240 updated we'll have this link below so you can try it for yourself since the 00:14:57.779 --> 00:15:04.320 video really won't do it justice on its own two clicks Phillips idea of 00:15:01.980 --> 00:15:08.639 combining this with full gated rendering to render a low resolution extension of 00:15:06.420 --> 00:15:12.660 the screen means that you'd practically never notice it and combined with frame 00:15:10.980 --> 00:15:17.040 generation to smooth everything out you've got a recipe for comparatively 00:15:14.639 --> 00:15:22.740 very low Hardware requirements along with very high responsiveness and frame 00:15:19.620 --> 00:15:24.660 rates this would be a holy grail of 00:15:22.740 --> 00:15:28.019 sorts there are many games that could benefit from this setup but none that I 00:15:26.399 --> 00:15:33.180 can think of would be a better fit than digital combat simulator DCS requires a 00:15:30.959 --> 00:15:36.600 beefy system to run and while you can get away with a not so beefy control 00:15:34.740 --> 00:15:40.440 setup one thing that you really need while flying is to keep an eye on your 00:15:38.699 --> 00:15:45.420 target whatever that may be so you've got to be able to whip your view around 00:15:41.940 --> 00:15:48.300 a lot and low frame rates make that job 00:15:45.420 --> 00:15:53.279 so much more difficult to do while just tracking small fast moving objects and 00:15:51.420 --> 00:15:57.300 if you lose track of your Target in DCS you can go from Hunter to hunted in an 00:15:55.860 --> 00:16:03.180 instant no matter what you're flying in a hypothetical future where Eagle Dynamics and developers like them 00:16:00.600 --> 00:16:07.079 Embrace Technologies like dlfg and async Time Warp the price of admission becomes 00:16:05.100 --> 00:16:11.279 substantially lower and they'll require less power to run too it may be a 00:16:09.300 --> 00:16:15.480 situational feature for today but if it means that more Gamers can have better 00:16:13.380 --> 00:16:19.560 experiences and even experiences that they wouldn't be able to have at all 00:16:17.399 --> 00:16:23.399 otherwise I'm keeping my hopes up for a tomorrow where our AI overlords 00:16:21.300 --> 00:16:28.920 regularly serve us those delicious imitation frames and we're I talk about 00:16:25.800 --> 00:16:30.300 our delicious sponsor mine mine is the 00:16:28.920 --> 00:16:34.560 Smart data assistant that helps you discover where your data is and then 00:16:32.519 --> 00:16:38.820 helps you keep it where you want it to be with mine you can exercise your data 00:16:37.019 --> 00:16:42.839 rights and reclaim your right to be forgotten by asking services that you no 00:16:40.980 --> 00:16:47.220 longer use to delete your information with how many data breaches there are 00:16:44.940 --> 00:16:52.019 these days I don't know about you but I wouldn't want any companies that I don't 00:16:48.899 --> 00:16:53.820 use holding my data so sign up and mine 00:16:52.019 --> 00:16:59.519 will let you know how many companies are holding your information we ran our 00:16:55.980 --> 00:17:01.380 email address and over 360 companies are 00:16:59.519 --> 00:17:06.419 holding our data including many services holding financial data that our team 00:17:03.240 --> 00:17:08.220 hasn't used in years mine will even 00:17:06.419 --> 00:17:12.059 automate the process of sending an official data deletion request through 00:17:10.079 --> 00:17:15.839 your inbox so the company can then delete the personal data they have 00:17:13.500 --> 00:17:19.319 stored mine also helps companies with the ultimate goal of simplifying and 00:17:17.579 --> 00:17:25.079 improving consumers online privacy experience so sign up at saymine.com 00:17:21.959 --> 00:17:26.819 using the link below and own your data 00:17:25.079 --> 00:17:30.419 thanks for watching guys maybe go check out the video where we tested to see if 00:17:28.319 --> 00:17:34.200 we could tell when dlss 2.0 is on or not these kinds of analysis videos they tend 00:17:32.760 --> 00:17:38.039 to miss out on the experience of actually using the feature and a lot of 00:17:36.179 --> 00:17:42.320 new ones can be lost just looking at the flaws instead of looking at the hole you 00:17:40.200 --> 00:17:42.320 know