- Joined
- Mar 26, 2009
- Messages
- 179 (0.03/day)
That's only in the Infiltrator demo, they didn't do this trashy comparison in any real game, and most games have awful scaling algorithms with lots of problems. So no DLSS is not comparable to 1800p.
-
Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.
NVIDIA DLSS Test in Final Fantasy XV
- Thread starter W1zzard
- Start date
- Joined
- Dec 14, 2011
- Messages
- 276 (0.06/day)
| Processor | 12900K @5.1all Pcore only, 1.23v |
|---|---|
| Motherboard | MSI Edge |
| Cooling | D15 Chromax Black |
| Memory | 32GB 4000 C15 |
| Video Card(s) | 4090 Suprim X |
| Storage | Various Samsung M.2s, 860 evo other |
| Display(s) | Predator X27 / Deck (Nreal air) / LG C3 83 |
| Case | FD Torrent |
| Audio Device(s) | Hifiman Ananda / AudioEngine A5+ |
| Power Supply | Seasonic Prime TX 1000W |
| Mouse | Amazon finest (no brand) |
| Keyboard | Amazon finest (no brand) |
| VR HMD | Index |
| Benchmark Scores | I got some numbers. |
Its an interesting option. I can say easily it falls behind 4K with a real(performance impacting) AA solution, but I do prefer it over FXAA or the 1440p with AA.
It just boils down to if you NEED that performance bump. I would prefer if it leaned more towards IQ than performance, getting something closer to a real AA solution with small performance gains would be nice. Am eager to see the HQ version too.
It just boils down to if you NEED that performance bump. I would prefer if it leaned more towards IQ than performance, getting something closer to a real AA solution with small performance gains would be nice. Am eager to see the HQ version too.
- Joined
- Dec 22, 2011
- Messages
- 3,890 (0.82/day)
| Processor | AMD Ryzen 7 3700X |
|---|---|
| Motherboard | MSI MAG B550 TOMAHAWK |
| Cooling | AMD Wraith Prism |
| Memory | Team Group Dark Pro 8Pack Edition 3600Mhz CL16 |
| Video Card(s) | NVIDIA GeForce RTX 3080 FE |
| Storage | Kingston A2000 1TB + Seagate HDD workhorse |
| Display(s) | Samsung 50" QN94A Neo QLED |
| Case | Antec 1200 |
| Power Supply | Seasonic Focus GX-850 |
| Mouse | Razer Deathadder Chroma |
| Keyboard | Logitech UltraX |
| Software | Windows 11 |
Yeah, it basically all boils down to choice. And people waste their lives here day in and day out arguing that more choice is a good thing.
I'd wish they would make their minds up.
I'd wish they would make their minds up.
Some of what I noticed:
first image (zoomed out only):
The clouds are much more amorphous with DLSS in the first comparison (in comparison with TXAA). The DLSS image has reduced contrast, overall, although there are some places that have increased contrast -- like the roof of the central tower (due to a reduction in highlight intensity). The result is a more hazy/cloudy/overcast look to the scene than with TXAA. The water is much stiller in the DLSS image. In the TXAA image, as with the clouds, there is a lot more in terms of blue patching. Shadows are narrower with DLSS and less intense, although the large shadow on the far right has a harder edge with DLSS. By contrast, the left side of the image, the pole shadow, is more muted than with TXAA. Many highlights are less intense with DLSS, such as with the front of the central tower, the bricks, and the large vine. The image takes on a bluer tint with DLSS. Compare, for instance, the neutral grey in the central three buttresses (or whatever those stone things with ornate patterning are called). Perhaps the worst spot in the image, when comparing the two, is the central vine, which looks much better with TXAA, thanks to more subtle brightness gradation. The increased water contrast in the TXAA also makes the railing look better by having it blend more into the scene. The same goes for the other metal poles. They blend more harmoniously into the scene with TXAA, especially the one on the far left.
second image:
As with the first comparison, the clouds are much more amorphous. The water, though, is worse with FXAA than with TAA. Shadows look inferior with FXAA vs. TAA. The central vine looks a little more lifelike with TAA vs. FXAA. FXAA also lacks the blue tint/overcast look of DLSS. The overall effect of FXAA is in between DLSS (flattest) and TAA (most vivid/contrasty), although FXAA seems to lighten things more overall, especially the center of the roof.
third image:
NoAA looks okay to me, except for the water -- which, like FXAA, is mostly flat all-blue. The other drawback is that the most obvious shadows don't blend well. They look more artificially tacked on. The railing lacks nuance... too dark. I have just compared the top pole, which has the roof ties and I like the TAA the most. With NoAA it seems to have too much contrast. FXAA helps some with that but TAA helps more. Without analysis, though, just a look, the lack of AA doesn't seem to be a big problem. The central vine, in fact, probably looks the best with no AA. TAA also brings out detail in the large window on the bottom left. DLSS brings out the most highlighting of that window's inner border but TAA has more contrast detail (less flat inner seam/border).
fourth image:
The large shadow on the right is pretty ragged with 1440. The prominent shadows don't blend well. Water detail is minimal, better with the DLSS. DLSS, though, reduces the contrast too much of the center of the image (although it improves the railing).
sixth image:
The waterfall looks better with TAA because of more highlight detail. The clouds have a little more detail with TAA. The fine misting where the waterfall hits the water seems to be a bit more subtle around the edges with DLSS, although there seems to be less detail in that whole area. The right side of the table has jagged edges that neither AA system gets rid of, although DLSS does a little bit better with it. The flowers seem to have tiny bit more contrast with TAA. The shadows under the flower box look a bit better with TAA. The brown boxes, interestingly, have more highlight brightening with DLSS. The placid water area blurring of reflections is better with TAA, more natural because there is more blur. The top of the handrail may be a little better with DLSS.
last image:
The blur of the water at the bottom left is more natural with TAA. The potted tree at the bottom right looks more natural with TAA, less flat/painted on. The tree at the top right looks better with TAA for the same reason. The shadows on the central stairs are better with TAA. The far railing on those stairs blends better with the water with TAA. It's easier to see the railing detail in the furthest away staircase with TAA, due to better contrast. The large central staircase has better highlighting with TAA. The yellow overhang and its support looks better with TAA. With DLSS, its support looks harsh, less natural, (overly-sharpened, poorly-blended highlights). The shadow under the staircase at the bottom center looks nice with DLSS, although not bad with TAA. The water around it looks better with TAA. The pillar to the left of the potted tree looks better with TAA.
Overall, I would say that TXAA and TAA are better than the other options, based on those shots.
The low polygon models problem, that could be seen with that table in the TPU image, is even more dramatic in the bottom left of your image, too. At least, that's what it looks like to me, since the jaggies are so massive. They're so large that AA probably can't be expected to take care of those. Regardless, though, anyone who has done much with Photoshop knows what going overboard with sharpening looks like and the car in particular has that problem. The entire image, for the most part, seems excessively crisp.
first image (zoomed out only):
The clouds are much more amorphous with DLSS in the first comparison (in comparison with TXAA). The DLSS image has reduced contrast, overall, although there are some places that have increased contrast -- like the roof of the central tower (due to a reduction in highlight intensity). The result is a more hazy/cloudy/overcast look to the scene than with TXAA. The water is much stiller in the DLSS image. In the TXAA image, as with the clouds, there is a lot more in terms of blue patching. Shadows are narrower with DLSS and less intense, although the large shadow on the far right has a harder edge with DLSS. By contrast, the left side of the image, the pole shadow, is more muted than with TXAA. Many highlights are less intense with DLSS, such as with the front of the central tower, the bricks, and the large vine. The image takes on a bluer tint with DLSS. Compare, for instance, the neutral grey in the central three buttresses (or whatever those stone things with ornate patterning are called). Perhaps the worst spot in the image, when comparing the two, is the central vine, which looks much better with TXAA, thanks to more subtle brightness gradation. The increased water contrast in the TXAA also makes the railing look better by having it blend more into the scene. The same goes for the other metal poles. They blend more harmoniously into the scene with TXAA, especially the one on the far left.
second image:
As with the first comparison, the clouds are much more amorphous. The water, though, is worse with FXAA than with TAA. Shadows look inferior with FXAA vs. TAA. The central vine looks a little more lifelike with TAA vs. FXAA. FXAA also lacks the blue tint/overcast look of DLSS. The overall effect of FXAA is in between DLSS (flattest) and TAA (most vivid/contrasty), although FXAA seems to lighten things more overall, especially the center of the roof.
third image:
NoAA looks okay to me, except for the water -- which, like FXAA, is mostly flat all-blue. The other drawback is that the most obvious shadows don't blend well. They look more artificially tacked on. The railing lacks nuance... too dark. I have just compared the top pole, which has the roof ties and I like the TAA the most. With NoAA it seems to have too much contrast. FXAA helps some with that but TAA helps more. Without analysis, though, just a look, the lack of AA doesn't seem to be a big problem. The central vine, in fact, probably looks the best with no AA. TAA also brings out detail in the large window on the bottom left. DLSS brings out the most highlighting of that window's inner border but TAA has more contrast detail (less flat inner seam/border).
fourth image:
The large shadow on the right is pretty ragged with 1440. The prominent shadows don't blend well. Water detail is minimal, better with the DLSS. DLSS, though, reduces the contrast too much of the center of the image (although it improves the railing).
sixth image:
The waterfall looks better with TAA because of more highlight detail. The clouds have a little more detail with TAA. The fine misting where the waterfall hits the water seems to be a bit more subtle around the edges with DLSS, although there seems to be less detail in that whole area. The right side of the table has jagged edges that neither AA system gets rid of, although DLSS does a little bit better with it. The flowers seem to have tiny bit more contrast with TAA. The shadows under the flower box look a bit better with TAA. The brown boxes, interestingly, have more highlight brightening with DLSS. The placid water area blurring of reflections is better with TAA, more natural because there is more blur. The top of the handrail may be a little better with DLSS.
last image:
The blur of the water at the bottom left is more natural with TAA. The potted tree at the bottom right looks more natural with TAA, less flat/painted on. The tree at the top right looks better with TAA for the same reason. The shadows on the central stairs are better with TAA. The far railing on those stairs blends better with the water with TAA. It's easier to see the railing detail in the furthest away staircase with TAA, due to better contrast. The large central staircase has better highlighting with TAA. The yellow overhang and its support looks better with TAA. With DLSS, its support looks harsh, less natural, (overly-sharpened, poorly-blended highlights). The shadow under the staircase at the bottom center looks nice with DLSS, although not bad with TAA. The water around it looks better with TAA. The pillar to the left of the potted tree looks better with TAA.
Overall, I would say that TXAA and TAA are better than the other options, based on those shots.
Lots of over-sharpening. The car is the most obvious but it's all over the image. It's definitely not doing enough as an anti-aliasing mechanism in that image.
The low polygon models problem, that could be seen with that table in the TPU image, is even more dramatic in the bottom left of your image, too. At least, that's what it looks like to me, since the jaggies are so massive. They're so large that AA probably can't be expected to take care of those. Regardless, though, anyone who has done much with Photoshop knows what going overboard with sharpening looks like and the car in particular has that problem. The entire image, for the most part, seems excessively crisp.
Last edited:
- Joined
- Dec 10, 2017
- Messages
- 266 (0.10/day)
| Processor | Intel core i5 4590s |
|---|---|
| Motherboard | Asus Z97 Pro Gamer |
| Cooling | Evercool EC115A 915SP Cpu cooler,Coolermaster [200mm (front and top)+140mm rear] |
| Memory | Corsair 16GB(4x4) ddr3 CMZ16GX3M4X1600C9(Ver8.16)(XMP) |
| Video Card(s) | MSI GTX 970 GAMING 4G |
| Storage | Western Digital WDC WD2001FAS 2TB Black, Toshiba DT01ACA100 1TB |
| Display(s) | LG Flatron L177WSB |
| Case | Coolermaster CM Storm Enforcer |
| Audio Device(s) | Creative A550 Speakers 5.1 channel |
| Power Supply | SuperFlower Leadex 2 Gold 650W SF-650F14EG |
| Mouse | PLNK M-740 Optical Mouse |
| Keyboard | ibuypower GKB100 Gaming Keyboard |
| Software | Windows 7 Sp1 64 bit |
requires nvidia (middle man) acknowledgement of your game to be able to get the feature settings for it
requires rtx hardware
requires 4k display
if one of these is not satisfied...well its not the way its meant to be played. I mean ffs there is already a specific launcher, drm and anti temper, specific game optimised drivers, ridiculously expensive hardware which runs on a horrendous specific os and now this? Is 4k so mainstream?! How much money and crapware do I need just to play a pc game with a specific feature enabled? I think this trend is discriminatory and must cease
end rant
requires rtx hardware
requires 4k display
if one of these is not satisfied...well its not the way its meant to be played. I mean ffs there is already a specific launcher, drm and anti temper, specific game optimised drivers, ridiculously expensive hardware which runs on a horrendous specific os and now this? Is 4k so mainstream?! How much money and crapware do I need just to play a pc game with a specific feature enabled? I think this trend is discriminatory and must cease
end rant
- Joined
- Nov 7, 2017
- Messages
- 52 (0.02/day)
Maybe this is not the best implementation of DLSS yet, but for now I recommend to just use 4K without anti aliasing. To my eye 4K content doesn't need AA in most cases and on lower resolutions with cards that actually support DLSS there is enough power for better quality AA.
- Joined
- Sep 17, 2014
- Messages
- 22,843 (6.06/day)
- Location
- The Washing Machine
| System Name | Tiny the White Yeti |
|---|---|
| Processor | 7800X3D |
| Motherboard | MSI MAG Mortar b650m wifi |
| Cooling | CPU: Thermalright Peerless Assassin / Case: Phanteks T30-120 x3 |
| Memory | 32GB Corsair Vengeance 30CL6000 |
| Video Card(s) | ASRock RX7900XT Phantom Gaming |
| Storage | Lexar NM790 4TB + Samsung 850 EVO 1TB + Samsung 980 1TB + Crucial BX100 250GB |
| Display(s) | Gigabyte G34QWC (3440x1440) |
| Case | Lian Li A3 mATX White |
| Audio Device(s) | Harman Kardon AVR137 + 2.1 |
| Power Supply | EVGA Supernova G2 750W |
| Mouse | Steelseries Aerox 5 |
| Keyboard | Lenovo Thinkpad Trackpoint II |
| VR HMD | HD 420 - Green Edition ;) |
| Software | W11 IoT Enterprise LTSC |
| Benchmark Scores | Over 9000 |
So, DLSS is dead in the water as well.
Well played NV. Well played. You managed to make no AA look worse than no AA (because honestly that is all I am seeing here, major, visible jaggies and reduced image quality). And then there are the obvious lighting differences, where each DLSS image looks stale like it was shot with a bad camera. Contrast and color richness seem to have gotten lost somewhere along the way. So much for 'free' AA...
I'll take any other AA over this crap any day of the week. Or even no AA.
Well played NV. Well played. You managed to make no AA look worse than no AA (because honestly that is all I am seeing here, major, visible jaggies and reduced image quality). And then there are the obvious lighting differences, where each DLSS image looks stale like it was shot with a bad camera. Contrast and color richness seem to have gotten lost somewhere along the way. So much for 'free' AA...
I'll take any other AA over this crap any day of the week. Or even no AA.