WEBVTT 00:00:00.000 --> 00:00:04.840 Whether you want to use AI upscaling or not, many modern games basically forced you. 00:00:04.840 --> 00:00:08.880 That is, if you want to turn up all the details and still get playable framerates. 00:00:08.880 --> 00:00:13.400 But I don't want to shell out for a new graphics card just to unlock a feature that I never really wanted 00:00:13.400 --> 00:00:17.800 in the first place. There's got to be a way to squeeze more 00:00:17.800 --> 00:00:22.680 out of the GPU I already have. And there is. It's one of the best kept open secrets 00:00:22.680 --> 00:00:29.240 to get more performance for your dollar and cost just $6.99, nice, or 420 cents on sale. 00:00:29.320 --> 00:00:33.160 Extra nice. It's made by one developer and the best part 00:00:33.160 --> 00:00:38.160 is it works on pretty much any graphics card, new or old, in every game. 00:00:38.160 --> 00:00:41.920 Meet Lostless Scaling, an app that can both upscale 00:00:41.920 --> 00:00:47.000 your games from lower resolutions and even use frame generation to multiply your FPS 00:00:47.000 --> 00:00:50.320 by as much as you want, albeit with mixed results 00:00:50.320 --> 00:00:53.680 once you get past a certain point. And that's not all. 00:00:53.680 --> 00:01:06.360 Man, I need to see this for myself. To start things off, both David and I 00:01:06.360 --> 00:01:12.800 are going to be experiencing doom-the-jar cages on a 4070 Ti at 4K high without any form of scaling 00:01:12.800 --> 00:01:16.800 or frame generation. This is really a great example of a modern game 00:01:16.800 --> 00:01:22.040 that really doesn't run perfect if you're not using any kind of cheat. 00:01:22.040 --> 00:01:26.240 So why don't we cheat? Control Alt S will toggle Lostless Scaling, 00:01:26.240 --> 00:01:32.440 which we've configured with two times frame gen, which is a lot smoother. 00:01:32.440 --> 00:01:39.880 But like any frame generation technology, I'm still feeling 40 FPS worth of input latency. 00:01:39.880 --> 00:01:45.360 With that said, I was expecting the image quality of those interpolated frames to be worse. 00:01:45.360 --> 00:01:50.080 Oh, it's really not bad. It's really not bad. I think if I didn't know frame gen was on, 00:01:50.080 --> 00:01:54.960 I would have a hard time immediately recognizing that it was. Which is both a good thing and a bad one. 00:01:54.960 --> 00:01:58.560 Yes. OK, now you've tried the two-time multiplier. Do you want to go higher? 00:01:58.560 --> 00:02:02.800 I do want to go higher. This is your first look at this interface, too. 00:02:02.800 --> 00:02:06.560 Super straightforward. So clean. Multiplier. Oh, god. 00:02:06.560 --> 00:02:10.320 OK. Just try three to start. OK. 00:02:10.320 --> 00:02:16.000 It's definitely smoother. But now I'm really getting that uncanny. 00:02:16.000 --> 00:02:20.600 It almost feels like motion smoothing on my TV. 00:02:20.600 --> 00:02:25.040 You know, like it's not right. Yeah. But I mean, when you look back over at this native 4K, 00:02:25.040 --> 00:02:29.160 which experience do you think you'd rather have? 00:02:29.160 --> 00:02:33.680 I'm looking for motion anomalies, and they're not. They're there, but they're not bad. 00:02:33.680 --> 00:02:37.680 Well, I had reason to believe that this would be substantially worse frame gen than the stuff 00:02:37.680 --> 00:02:42.360 that NVIDIA and AMB has. But it's pretty darn good. 00:02:42.360 --> 00:02:47.000 But can I multiply it even more? Oh, yeah. I'm going to 5X. 00:02:47.000 --> 00:02:50.880 That's more than NVIDIA can do. So I have about 45 FPS native. 00:02:50.920 --> 00:02:54.680 Oh, wow. Oh, that's interesting. 00:02:54.680 --> 00:02:58.840 It's not even pretending that I'm able to do much more than about, oh, no, that 00:02:58.840 --> 00:03:01.840 is about 5X right now. It's just your base frame rate dropped so much 00:03:01.840 --> 00:03:06.120 because so much computation is happening on your graphics card. Right. 00:03:06.120 --> 00:03:09.480 So there's a reason that NVIDIA capped out their solution 00:03:09.480 --> 00:03:14.040 at 4X. It's just so costly on the compute side 00:03:14.040 --> 00:03:17.720 that you're not getting enough base FPS to even get a net benefit from it. 00:03:17.720 --> 00:03:22.240 Interesting. The higher your multiplier goes, the more 00:03:22.240 --> 00:03:25.560 I think native is the better choice. 00:03:25.560 --> 00:03:28.920 Yeah, this is not only not a better experience 00:03:28.920 --> 00:03:32.680 than what I was trying before, but this is not a good experience. 00:03:32.680 --> 00:03:36.240 What we want to explore next is fixed versus adaptive frame 00:03:36.240 --> 00:03:39.960 gen. So fixed is just the multiplier like NVIDIA has. 00:03:39.960 --> 00:03:46.920 But adaptive will add in between frames to hit a set frame rate, sort of like a computational variable 00:03:46.920 --> 00:03:51.600 refresh rate to oversimplify it a lot. So instead of having my multiplier, 00:03:51.600 --> 00:03:54.680 I'm going to go adaptive and I'm going to say I want a game 00:03:54.680 --> 00:03:58.640 at, I think 120 is asking a lot at these settings. 00:03:58.640 --> 00:04:04.520 But maybe I want a game at about 85 FPS. I think that's a really nice sweet spot. 00:04:04.520 --> 00:04:11.040 It's much, much better feeling than 60, but not so intense as like 120 or more. 00:04:11.040 --> 00:04:14.600 My base frame rate really is still a little bit too low. 00:04:14.600 --> 00:04:19.960 Like 40 feels input laggy. I would probably want to turn this game down a little bit, 00:04:19.960 --> 00:04:25.200 but in terms of the smoothness, it looks pretty darn good. 00:04:25.200 --> 00:04:31.240 Well, I'm not playing so I can't ascertain the input latency, but in terms of visuals, that looks high refresh rate. 00:04:31.240 --> 00:04:35.760 It looks great. It looks really good. You can do 120 though. 00:04:35.760 --> 00:04:40.120 I wouldn't. It feels too uncanny, like the disconnect 00:04:40.120 --> 00:04:43.440 between how the latency feels and how the smoothness looks 00:04:43.440 --> 00:04:48.160 is just too great. There's actually one more slider we can mess around with to change the performance. 00:04:48.160 --> 00:04:52.800 So why don't you switch to loss of scale and I'll show you. This is the flow scale. 00:04:52.800 --> 00:04:55.880 Basically, it's the size of the images that 00:04:55.880 --> 00:04:59.400 are going into loss of scale. So it takes the full screen and then it lowers the resolution 00:04:59.400 --> 00:05:05.800 to process that the lower your flow scale is. And so right now, the sweet spots between 50 and 60%, 00:05:05.800 --> 00:05:10.600 which is where we are, but you could theoretically get a better image out of the higher flow scale 00:05:10.600 --> 00:05:14.280 because there's more information for it to process and more performance at the low end 00:05:14.280 --> 00:05:18.160 because those images are being shrunk and processed fast. Interesting. So let's try 40. 00:05:18.160 --> 00:05:22.000 Sure. OK, so turning it down is supposed to give me 00:05:22.000 --> 00:05:27.000 like maybe less computational overhead. A little bit. It seems like you're like a few frames higher, 00:05:27.000 --> 00:05:30.400 but then you might notice more anomalies in your generation. 00:05:30.400 --> 00:05:34.960 What a great utility. Can you imagine a world where like GPU vendors just built 00:05:34.960 --> 00:05:39.680 stuff like this that's completely platform agnostic and just no nonsense. 00:05:39.680 --> 00:05:44.720 Here's the sliders. Here's what everything does. There's no obfuscating what the actual performance is. 00:05:44.720 --> 00:05:47.960 Like their FPS counter tells me how many real frames I have 00:05:47.960 --> 00:05:51.400 and then how many of them are generated. So good. Like just don't mark it to me. 00:05:51.400 --> 00:05:55.960 Just tell me what it is. So good. Obviously, there are limitations and we're seeing them. 00:05:55.960 --> 00:06:00.800 But overall, if I didn't see how well this worked with my own eyes, I wouldn't believe 00:06:00.800 --> 00:06:04.000 that this could be made by just one person. How? 00:06:04.000 --> 00:06:10.840 OK, well, like many programs that benefit from open source projects, it's not actually one person, 00:06:10.840 --> 00:06:16.360 but one of the most impressive pieces, lossless scaling frame gen is. 00:06:16.360 --> 00:06:22.920 So we reached out to the creator of lossless scaling who goes by the pseudonym THS to explain how it wakes. 00:06:22.920 --> 00:06:26.160 Lossless scaling frame generation is a convolutional neural network 00:06:26.160 --> 00:06:30.400 developed from scratch. It takes only the final frame as input, 00:06:30.400 --> 00:06:36.120 and specifically in the LSFG3 version, is capable of generating any number of intermediate frames 00:06:36.120 --> 00:06:40.800 for arbitrary time stamps. This enables both adaptive and any multiplier fixed modes. 00:06:40.800 --> 00:06:44.640 Compared to DLSS or FSR, which receive extensive information 00:06:44.640 --> 00:06:47.960 from the game engine, LSFG operates with significantly 00:06:47.960 --> 00:06:52.040 less input data. While this limits its potential output quality. 00:06:52.040 --> 00:06:56.200 Yeah, about that. Lacking motion vectors to help process the image 00:06:56.200 --> 00:07:02.880 and not being able to differentiate HUD elements from game objects means that visual anomalies can be pretty 00:07:02.880 --> 00:07:09.760 distracting in some situations. But it also gives LSFG a major advantage in terms 00:07:09.800 --> 00:07:14.320 of the wide range of use cases. To ensure broad compatibility across GPUs, 00:07:14.320 --> 00:07:17.440 the trained model was manually ported to DirectX 11 00:07:17.440 --> 00:07:21.200 compute shaders. LS captures via game frames and processes them 00:07:21.200 --> 00:07:24.440 via LSFG shaders, functioning effectively 00:07:24.440 --> 00:07:27.840 as a DirectX 11 application that generates and displays 00:07:27.840 --> 00:07:30.960 frames at the appropriate times based on the selected mode. 00:07:30.960 --> 00:07:37.080 That is probably the coolest part. There is no requirement for specialized AI hardware, 00:07:37.080 --> 00:07:40.640 and DX11 GPUs have been mainstreamed for well over 00:07:40.640 --> 00:07:45.320 a decade now. So yeah, the image quality can only go so far, 00:07:45.320 --> 00:07:48.800 but just about anyone can take advantage of it. 00:07:48.800 --> 00:07:53.600 There are some other caveats. A big one is that you're going to want to keep your GPU usage 00:07:53.600 --> 00:07:57.840 to around 85% to avoid spikes that really hurt lossless 00:07:57.840 --> 00:08:03.160 scaling's quality, which, wait, 85%, I bought the whole GPU. 00:08:03.160 --> 00:08:06.120 I want to use the whole GPU. 00:08:06.160 --> 00:08:11.160 Well, don't worry, THS thought of that. Another super cool thing about lossless scaling 00:08:11.160 --> 00:08:15.080 is that it can be run on a secondary GPU 00:08:15.080 --> 00:08:19.800 or even your integrated graphics, which reduces the load on your primary gaming card 00:08:19.800 --> 00:08:27.040 and frees up a little bit of extra performance. Now, our results with an IGPU were pretty underwhelming, 00:08:27.880 --> 00:08:31.640 but the idea of an upscaling co-processor 00:08:31.640 --> 00:08:36.080 is one that really intrigues me. A lot of people have old GPUs kicking around 00:08:36.080 --> 00:08:39.200 and hey, maybe this $7 piece of software 00:08:39.200 --> 00:08:42.800 can give it a fresh raison d'etre. Let's give it a shot. 00:08:42.800 --> 00:08:46.720 Now let's get this started. There are a couple things that we changed from the B-roll though. 00:08:46.720 --> 00:08:53.140 I actually ended up using the slot that was right next to my primary card because it runs at 8x rather than 4x 00:08:53.140 --> 00:08:56.800 and the data rate to our secondary GPU actually does matter. 00:08:56.800 --> 00:09:01.240 The other thing that we need to do is we have to unplug from our primary GPU 00:09:01.240 --> 00:09:05.040 and actually run our display off of the secondary GPU. 00:09:05.040 --> 00:09:09.000 So this one drives the display, but this one is providing the main gaming grunt. 00:09:09.000 --> 00:09:13.440 It's just routed through the secondary card. What is this to the Velcro? 00:09:13.440 --> 00:09:16.900 Oh, the Velcro is to push these cards apart a little bit 00:09:16.900 --> 00:09:20.880 so this one doesn't overheat. LTTstore.com. 00:09:20.880 --> 00:09:25.960 Now we just need to configure lossless scaling. So we'll change our preferred GPU to RTX 2060. 00:09:25.960 --> 00:09:29.160 That's our preferred GPU for scaling to use. 00:09:29.160 --> 00:09:32.240 Then we're gonna go into our graphics settings in Windows 00:09:32.240 --> 00:09:37.080 and we're gonna change our default high-performance GPU to our faster one that we don't actually 00:09:37.080 --> 00:09:40.760 have our display plugged into. So for us, that's a 4070 Ti. 00:09:40.760 --> 00:09:46.000 You may also find that you need to add an exception for the specific game, but not always. 00:09:46.000 --> 00:09:51.480 Let's lose the adaptive scaling for now then and let's go back to trying like a 3x multiplier. 00:09:53.640 --> 00:09:56.480 Am I getting better performance than before? I feel like I'm over. 00:09:56.880 --> 00:10:00.160 It's as bad, if not worse than it was before. Okay. 00:10:01.000 --> 00:10:04.800 And this sucks. This is really bad. 00:10:04.800 --> 00:10:09.560 Almost at 30 FPS. Yeah, which is not high enough 00:10:09.560 --> 00:10:12.880 for a good gaming experience. You know what though? 00:10:14.240 --> 00:10:17.360 The 2060 is probably not powerful enough 00:10:17.360 --> 00:10:20.960 to do 4K frame gen. So you are at the mercy somewhat 00:10:20.960 --> 00:10:26.080 of the performance of that secondary card. It can't just be like any ancient card 00:10:26.080 --> 00:10:29.520 if you're hoping to run at the highest resolution in detail settings. 00:10:29.520 --> 00:10:32.680 Do we have like a 3060? Yeah, we do. Oh, perfect. 00:10:32.680 --> 00:10:37.120 Okay, let's do that. I feel good. It's gonna work good this time. Oh, it's gonna be so good this time. 00:10:37.120 --> 00:10:41.880 If we'd done like 2x frame gen or if we were running at a lower resolution with that 2060, 00:10:41.880 --> 00:10:45.040 probably would have been fine. But we want 4K gaming. 00:10:45.040 --> 00:10:48.560 At least for now. 4K. Well, yeah. Here we go, here we go. 00:10:48.560 --> 00:10:52.960 So my native performance is exactly what I'd expect. Control Alt S and 00:10:52.960 --> 00:10:58.200 Woo! And my native performance stays exactly the same. 00:10:58.200 --> 00:11:03.840 And I get frame gen. This is the best experience I've had with it so far. 00:11:03.840 --> 00:11:08.760 Okay, this is pretty good. Doesn't do anything about the latency. 00:11:08.760 --> 00:11:12.040 But wow, what a cool idea. 00:11:12.040 --> 00:11:15.960 You wanna see how high the number we can go? I mean, sure. 00:11:17.040 --> 00:11:20.480 Yeah, I guess with the dedicated card we could multiply higher. 00:11:20.480 --> 00:11:23.160 Or we can do it because of the base resolution. Should I go to five this time? 00:11:23.800 --> 00:11:27.800 Okay, I'm going to five. I'm going to five, boys. Dude! 00:11:27.800 --> 00:11:31.520 Look how smooth that is. That's crazy smooth. It looks high refresh rate. 00:11:31.520 --> 00:11:37.000 It is, but it looks it. Dude! I am getting a 5x frame gen multiplier. 00:11:37.000 --> 00:11:41.400 Wow. It still feels in terms of input lag like 40 FPS, but. 00:11:44.760 --> 00:11:49.960 Artifacting is not immediately perceptible either. I mean, this is a pretty like, you know, 00:11:49.960 --> 00:11:53.040 early 2000s brown filter over everything game. 00:11:53.040 --> 00:11:56.240 You lost your scaling. Oh, I did? Yeah. 00:11:56.240 --> 00:12:00.280 What just happened? Oh my God, I think I lost the scale. Oh, no, wait, no, it's back. 00:12:00.280 --> 00:12:05.840 200 frames. No! Now, it's worth noting that the resolution 00:12:05.840 --> 00:12:09.600 that I perceive in motion is not as good 00:12:09.600 --> 00:12:14.640 as when I'm holding still. And that's a significant drawback of these kinds of scaling technologies 00:12:14.640 --> 00:12:20.200 is that if they don't know which direction an object is going because they don't have motion vector data 00:12:20.200 --> 00:12:23.640 from the game engine, you're going to lose some of the clarity 00:12:23.640 --> 00:12:28.200 when you are in motion. I wouldn't use this for a competitive game by any stretch of the imagination. 00:12:28.200 --> 00:12:32.080 Look at your HUD. Look at that crosshair. Sometimes we move really fast. 00:12:32.080 --> 00:12:36.000 Whoa! Whoa! You know what's funny is when I was playing, 00:12:36.000 --> 00:12:41.440 I didn't even notice it though. No, but it's, whoa! As soon as you're looking for it, it's crazy. 00:12:41.440 --> 00:12:46.440 Yeah, that's pretty bad. We are doing 5x frame gen though. 00:12:46.440 --> 00:12:52.240 And that's another thing that is a disadvantage when you don't have any kind of input from the game. 00:12:52.240 --> 00:12:56.560 Like, hey, maybe you don't alter these HUD elements. 00:12:56.560 --> 00:13:00.800 It's something he could add in too. In reshade, you can separate HUD elements manually. 00:13:00.800 --> 00:13:06.280 And so I could see on a game per game basis allowing people to have the HUD be processed separately 00:13:06.280 --> 00:13:09.320 from the rest of the scaling, maybe. Maybe it'd be pretty tough 00:13:09.320 --> 00:13:14.480 because then you're losing some of the advantages of this software is that it just kind of works on everything. 00:13:14.480 --> 00:13:17.680 100%. And then you're building profiles again. And then you're basically just doing 00:13:17.680 --> 00:13:21.360 what AMD and NVIDIA are doing. Fair enough. Overall, I don't think there's a question 00:13:21.360 --> 00:13:24.760 that I would rather play with this on, hmm. 00:13:24.760 --> 00:13:28.600 Maybe turn off the 5x so it doesn't look as stupid. Yeah, here, let's... 00:13:28.600 --> 00:13:32.360 Go back to two. Let's go to a more reasonable scaling multiplier. 00:13:32.360 --> 00:13:37.040 Maybe two. Yeah. At this point, I don't think there's much of a question. 00:13:37.040 --> 00:13:42.000 I would, ah. Now that you're seeing the thing, 00:13:42.000 --> 00:13:45.680 it's like it's, it wasn't as bad before. I think it's even worse. 00:13:45.680 --> 00:13:48.760 That's interesting. Okay, look, let me try yours again. 00:13:48.760 --> 00:13:51.280 Oh, okay. 00:13:52.920 --> 00:13:57.240 No, no, no, it's fine. It's fine, it doesn't matter. It's just like, I was questioning 00:13:57.240 --> 00:14:03.280 whether I would turn this on and I'm like, oh, maybe neither of these are what I would configure 00:14:03.280 --> 00:14:07.800 because I would want a higher base frame rate regardless, even if it means turning down some details. 00:14:07.800 --> 00:14:11.360 But there's no question that the software is pretty impressive 00:14:11.360 --> 00:14:14.600 and we haven't even touched on the actual scaling 00:14:14.600 --> 00:14:18.920 that it was originally named for. THS said that the reason they originally made 00:14:18.920 --> 00:14:24.960 lossless scaling was that they bought a 4K monitor for coding and their computer couldn't handle it for gaming. 00:14:24.960 --> 00:14:28.480 Now, obviously you can just run your monitor 00:14:28.480 --> 00:14:33.200 at a non-native resolution, but at the time, graphics drivers could only do 00:14:33.200 --> 00:14:38.200 nasty hobbits' soft, bilinear scaling, yuck. 00:14:38.480 --> 00:14:41.520 So, THS developed the first version 00:14:41.520 --> 00:14:44.940 to make sharp, sexy, integer scaling a breeze 00:14:44.940 --> 00:14:49.780 and then just kept adding features until we got the flexible array of options 00:14:49.780 --> 00:14:54.020 that we're left with today. It's even great for non-gaming applications. 00:14:54.020 --> 00:14:58.140 Do you love anime? Well, you can boost the sharpness with anime 4K. 00:14:58.140 --> 00:15:02.700 More into emulating retro games, THS included XBR scaling to make pixel art games 00:15:02.700 --> 00:15:06.340 really look their best. And if you wanna scale your demanding 3D games, 00:15:06.340 --> 00:15:09.740 well, hey, you can try AMD's FidelityFX Super Resolution, 00:15:09.740 --> 00:15:15.140 FSR, or NVIDIA image scaling, even in games that don't support them. 00:15:15.140 --> 00:15:20.100 That versatility and the convenient package it's wrapped in is genuinely fantastic. 00:15:20.100 --> 00:15:23.180 Let's try using FSR. Let's start with Doom the Dark Ages, 00:15:23.180 --> 00:15:28.180 which actually does support FSR. So David is gonna turn it on in-game 00:15:28.180 --> 00:15:32.700 and then I am gonna turn it on using lossless scaling. I'm also gonna disable my secondary GPU 00:15:32.700 --> 00:15:38.260 because that's not really apples to apples otherwise. So I guess I want the optimized version, 00:15:38.260 --> 00:15:42.740 a non-original optimized version of AMD FSR, which is better to do for the less powerful. 00:15:42.740 --> 00:15:47.900 RGPs are also powerful. Yeah, let's leave that off. Sharpness, let's leave that in the middle. 00:15:47.900 --> 00:15:52.180 To do that, I need to set my game to borderless windowed and then I'm gonna set my resolution 00:15:52.180 --> 00:15:56.220 to whatever I would want my input resolution to be for my scaling technology. 00:15:56.220 --> 00:16:00.860 So let's go with 1080p. I wanna really play at a high frame rate. 00:16:02.060 --> 00:16:07.380 Now we go into lossless scaling and we set our type to FSR. 00:16:07.380 --> 00:16:10.380 We're just gonna leave it on auto, I guess. 00:16:10.460 --> 00:16:13.700 Will that do a 2x factor? Well, it's gonna bring it 00:16:13.700 --> 00:16:17.860 to whatever your monitor's resolution is. Oh, brilliant. 00:16:17.860 --> 00:16:22.860 Okay, oh yeah, right. I'm definitely scaling because I was running 00:16:22.860 --> 00:16:27.180 at that lower resolution before and if it's not on, it's running like this 00:16:27.180 --> 00:16:31.660 and if it's on, it's running like that. That's a lot of daring. 00:16:31.660 --> 00:16:35.980 We are running into some v-sync issues. I've noticed there's limitations 00:16:35.980 --> 00:16:39.060 based on the monitor's refresh rate when you're using lossless scaling 00:16:39.060 --> 00:16:43.700 and because we're going to a capture that is limited to 60, we are gonna be kinda locked around 60 00:16:43.700 --> 00:16:47.140 even if you turn v-sync off and everything. Let's lose the capture then 00:16:47.140 --> 00:16:50.420 and we'll just use the over the shoulder. Wait, how are you at 93 FPS? 00:16:50.420 --> 00:16:56.100 I have FSRs on. Oh, that's right, I forgot. Okay, so now we're pixel peeping a little bit. 00:16:56.100 --> 00:16:59.140 Why don't we just play for a little bit first? Hey, I'm getting more FPS than you. 00:16:59.140 --> 00:17:03.140 No, I just wanna make sure that my settings are correct. FSR performance would be 1080p. 00:17:03.140 --> 00:17:06.420 Ooh, now I'm not feeling the latency, boys. Well, those are real frames. 00:17:06.420 --> 00:17:11.980 They're just lower resolution. You know what, man, they look pretty okay to me. 00:17:11.980 --> 00:17:15.260 Yeah, I can't immediately tell the difference in sharpness between ours. 00:17:15.260 --> 00:17:18.700 Oh, I'm up at 100 now. Performance is quite similar. 00:17:18.700 --> 00:17:23.180 And mine was easier to set up than yours was. That's fair. 00:17:23.180 --> 00:17:27.460 Got a little finicky problems. It's definitely not as sharp. 00:17:27.460 --> 00:17:31.620 Like I'm looking at these chains that there's baddies kinda hanging down from or whatever. 00:17:31.620 --> 00:17:37.020 Like... But same with mine, the native FSR is not perfect. 00:17:37.020 --> 00:17:40.940 Now, is this supposed to be as good as proper integrated FSR? 00:17:40.940 --> 00:17:44.860 I don't think so, because I think 00:17:44.860 --> 00:17:47.860 the integrated FSR gets access to a little more information 00:17:47.860 --> 00:17:50.980 than just the output frame. It really doesn't look that different to me, though. 00:17:50.980 --> 00:17:55.900 I mean, can you do me a favor and go find that chain again and then just like do the same back and forth? 00:17:55.900 --> 00:18:00.300 Cause I've got crazy ghosting on mine. Whoa, buddy. 00:18:00.300 --> 00:18:05.020 Mine's not doing it. Huh. I would have assumed that the one built into Doom 00:18:05.020 --> 00:18:10.020 would be better, but I don't know if that's necessarily true. Another big one is distant foliage. 00:18:10.020 --> 00:18:14.500 So let's just find a tree or something to look at. Here we go, against these mountains, 00:18:14.500 --> 00:18:18.500 there are like these birds and stuff behind you. Behind? Up there, yeah. 00:18:18.500 --> 00:18:21.900 I am not perceiving a major difference here. 00:18:21.900 --> 00:18:26.780 No. Oh, you know what? No, this tree is better on yours. 00:18:27.780 --> 00:18:31.780 Yeah, you're right. Definitely. So it looks like it's gonna be a bit of a mixed bag, 00:18:31.780 --> 00:18:35.740 but what's cool about loss of scaling is I have all other kinds of different options. 00:18:35.740 --> 00:18:39.540 Which one do you want? LF1, let's do it. That's the scaling that they built. 00:18:41.860 --> 00:18:45.260 Uh, yeah. I don't think it's as good as that. 00:18:45.260 --> 00:18:47.460 Yeah. Well, let's look at the chain. 00:18:49.020 --> 00:18:53.180 Yeah, the tree's not as good. I can tell you that much for sure. Hey, my chain's still better than your chain though. 00:18:53.180 --> 00:18:57.140 Got them. This is so crazy. I mean, that's the thing with all of these technologies 00:18:57.140 --> 00:19:02.420 is they do tend to be a bit of a mixed bag and you're gonna have to play around and find out what works best for you. 00:19:02.420 --> 00:19:07.300 And again, what's cool about loss of scaling is that I don't even need the game to support it at all. 00:19:07.300 --> 00:19:12.260 So why don't we fire up a game that doesn't support loss of scaling? Like, Halo Infinite? 00:19:12.260 --> 00:19:16.340 No. Why? Because... Why do you guys never let me play Halo Infinite 00:19:16.340 --> 00:19:21.020 on set anymore? Well, first off, you play too long. Secondly, because of the way Halo Infinite 00:19:21.020 --> 00:19:25.620 does its resolution, we can't change it to a smaller size to scale up. 00:19:25.620 --> 00:19:29.300 So it won't work with loss of scaling in that way. 00:19:29.300 --> 00:19:32.380 Yay, Halo Infinite! Woohoo! 00:19:32.380 --> 00:19:35.660 And I'm getting more FPS, boys! 00:19:36.660 --> 00:19:40.420 Not by that much. As for image quality, 00:19:41.500 --> 00:19:44.540 let's have a look here. Hey, meet me at the Bazaar. The Bazaar. 00:19:44.540 --> 00:19:48.020 The Bazaar! Stop it, I'm doing a video! 00:19:48.020 --> 00:19:51.620 Leave me alone. You shouldn't put so many bots in the game then. 00:19:51.620 --> 00:19:54.380 Don't leave the option to change much. Okay, let's see. 00:19:55.860 --> 00:19:59.300 I mean... Yeah, I'm upscaling from 1440, 00:19:59.300 --> 00:20:01.780 but like, this is pretty good. 00:20:02.740 --> 00:20:05.780 The text is a little fuzzy. Like, that's noticeable to me, 00:20:05.780 --> 00:20:09.580 but like, I have to go looking for it. Not bad. 00:20:09.580 --> 00:20:13.060 Of course, I'm not getting a huge performance benefit right now either. 00:20:13.060 --> 00:20:16.420 I'm getting about the same frame rate that you are. 00:20:16.420 --> 00:20:20.900 Wait, do I not have it on? I do have it on, because I have my performance counter. 00:20:20.900 --> 00:20:26.060 Okay, I'm gonna turn and frame you as well, just for loss. My image quality looks darn near the same as yours, 00:20:26.060 --> 00:20:29.980 but my FPS is also darn near the same as yours. 00:20:30.900 --> 00:20:34.980 You're doing a little bit better than me. I'd be interested to try the adaptive frame rate. 00:20:34.980 --> 00:20:39.020 I think that's one of the cool use cases for it. If you have an old eSports monitor 00:20:39.020 --> 00:20:43.100 that doesn't support variable refresh rate, you can kind of just force it computationally 00:20:43.100 --> 00:20:47.900 and hit a consistent frame rate, even if some of those frames are generated. 00:20:47.900 --> 00:20:52.140 Okay, let's turn on frame gen and have a look at how things go. Yeah. 00:20:52.140 --> 00:20:56.180 Okay, so... I love that this, the first thing I think 00:20:56.180 --> 00:20:58.580 when I see LSFG is let's f***ing go. 00:20:59.340 --> 00:21:03.220 Let's f***ing go. Okay, I'm gonna try 2x multiplier to start. 00:21:03.220 --> 00:21:06.260 With the higher input frame rate, dude, 00:21:06.260 --> 00:21:10.580 obviously it doesn't feel quite right. It looks really smooth. 00:21:10.580 --> 00:21:13.860 I can force it to artifact, like look at this crane in the back. 00:21:13.860 --> 00:21:18.340 Oh yeah. Like I can make it kind of do weird stuff, but it's not the sort of thing that I would notice 00:21:18.340 --> 00:21:23.140 if I was just playing the game. No. Okay, like hold on, let me try yours for a sec. 00:21:23.140 --> 00:21:27.580 Let me try yours for a sec. Dude, I would play on that one for sure. 00:21:27.580 --> 00:21:32.780 No question, actually. Like I thought I was gonna be like, hmm, I guess the pros and cons, you see, 00:21:32.780 --> 00:21:36.260 they kind of somewhat outweigh or they don't outweigh. 00:21:36.260 --> 00:21:40.020 Like... And it's surprising, this is kind of like an eSports situation, 00:21:40.020 --> 00:21:44.220 and I thought this would be totally something that is not in the wheelhouse of loss of scaling. 00:21:44.220 --> 00:21:49.780 Okay, switch again. I wanna see, I wanna confirm. It just goes to show that frame generation is best 00:21:49.780 --> 00:21:53.540 in situations where you least need it. My base experience is pretty good. 00:21:53.540 --> 00:21:56.740 It's just a little boosted by that frame gen. 00:21:56.740 --> 00:22:00.500 I gotta say I like it, which you might see as a reversal 00:22:00.500 --> 00:22:04.380 from my previous position, where generally I avoid upscaling 00:22:04.380 --> 00:22:08.900 or frame generation technologies whenever possible. But there's a big difference to me 00:22:08.900 --> 00:22:12.380 between what AMD and especially NVIDIA have been selling, 00:22:12.380 --> 00:22:16.580 which is fake frames for real dollars and what THS is selling, 00:22:16.580 --> 00:22:19.860 which is a shot of adrenaline for your older card 00:22:19.860 --> 00:22:22.900 for the very reasonable price of $7. 00:22:22.900 --> 00:22:27.900 And I know that some folks are still gonna be adamantly against using scaling technologies 00:22:27.900 --> 00:22:33.860 or frame generation technologies of any sort, but the controls in here are granular enough 00:22:33.860 --> 00:22:39.660 that I think even if all you want is an extra five or 10% headroom on an older card 00:22:39.660 --> 00:22:44.500 or to hit a target frame rate, you might appreciate having this tool in your toolkit 00:22:44.500 --> 00:22:48.500 and deciding for yourself if it's worth the compromises. 00:22:48.500 --> 00:22:53.060 I guarantee that you have $7 games in your Steam library right now that you've played less 00:22:53.060 --> 00:22:56.360 than you'll tinker around with this. Just like you should check out the last time 00:22:56.360 --> 00:23:00.360 we did something fun with a GPU. Okay, that was fun and very dumb, 00:23:00.360 --> 00:23:04.560 but boy, did it ever go brrrrrrr, and like this kind, brrrrrrr. 00:23:04.560 --> 00:23:05.400 It was cold.