Forspoken: FSR 2.1 vs. DLSS Comparison

[maxus24]

Forspoken is out now, with support for NVIDIA's DLSS Super Resolution (DLSS 2.5) and AMD's FidelityFX Super Resolution 2.1 (FSR 2.1). In this mini-review we take a look, comparing the image quality and performance gains offered by these technologies.

Show full review

 

[gffermari]

...this is the only time that FSR looks better than DLSS.
I wonder why.

 

[Vya Domus]

...this is the only time that FSR looks better than DLSS.
I wonder why.

Ah, yes, AMD must have used some magic pixel fairy dust and make the image look better. That makes the most sense to me.

 

[gffermari]

Not AMD, Square Enix.

FSR 2.x nearly matches DLSS 2.x. It's extremely unlikely to provide better image.
In this case, and it's the only one case, the DLSS is miles worse than the FSR.

So yes. Square Enix developed the most unoptimized game.
It's like targeting the greens. There is no excuse for that. It's not a tech demo (like Portal RTX).

 

[Dammeron]

...this is the only time that FSR looks better than DLSS.
I wonder why.

It's pretty amazing how bad DLSS Q looks in this game - short-sighted people might think they forgot to put their glasses on.

 

[Gucky]

FSR-Q 2.1 looks sharper then DLSS-Q. Is there a sharpen setting?
The new DLSS 2.5 doesn't make use of the sharpen feature anymore, maybe that "broke" it.
Maybe try DLSS 2.4.

 

[evernessince]

Is it possible to have the "DLSS" and "FSR" labels hidden on each image comparison unless I press a button? A blind comparison is much more effective and removes some potential bias.

 

[neatfeatguy]

All the grass with DLSS was just a blurred mess. I haven't made use of RT or DLSS with my GPUs in any games I've played. I don't see the benefit of them. Perhaps one day they may appeal to me, but for now they're overpriced gimmicks.

 

[forgotten05]

I noticed that when tpu compares fsr and dlss, they manually update dlss version. Why though? In some reviews I saw some of the editors were mentioning they were focusing on stock experience yet they replace dlss file in nearly every fsr vs dlss comparison if compatible. Most of the people who plays the game are not aware or doesn't even bother to update it. In short, very few people do it. It ruins what it has been shipped and when updated, it doesn't represent game's stock experience.
Personally I think it should be stock, but that's me, I don't know.

 

[watzupken]

I think that is the flipside of all these upscaling technology. The image quality is really as good as the optimization. Did the Tensor cores do anything smart to improve image quality? May be, but a good part seems to be contributed by human developers' effort.

I noticed that when tpu compares fsr and dlss, they manually update dlss version. Why though? In some reviews I saw some of the editors were mentioning they were focusing on stock experience yet they replace dlss file in nearly every fsr vs dlss comparison if compatible. Most of the people who plays the game are not aware or doesn't even bother to update it. In short, very few people do it. It ruins what it has been shipped and when updated, it doesn't represent game's stock experience.
Personally I think it should be stock, but that's me, I don't know.

I believe it could be like driver, where you are building/ optimizing what is currently available. In other words. running the latest DLSS version should not cause regression in image quality (which is the ideal case). In this case, it looks bad and likely not contributed by the DLSS version.

 

[forgotten05]

I think that is the flipside of all these upscaling technology. The image quality is really as good as the optimization. Did the Tensor cores do anything smart to improve image quality? May be, but a good part seems to be contributed by human developers' effort.


I believe it could be like driver, where you are building/ optimizing what is currently available. In other words. running the latest DLSS version should not cause regression in image quality (which is the ideal case). In this case, it looks bad and likely not contributed by the DLSS version.

My bad saying it's just ruining. What I meant is it ruins the aim of comparison since people who are looking to these comparison won't find the same result on their end or it will lead them in a wrong way since that is not what they have since most people will be using stock settings and won't replace it.

 

[Sithaer]

All the grass with DLSS was just a blurred mess. I haven't made use of RT or DLSS with my GPUs in any games I've played. I don't see the benefit of them. Perhaps one day they may appeal to me, but for now they're overpriced gimmicks.

Opposite for me, ever since I have my 3060 Ti and tried them I've been using DLSS in almost every game I've played that had it implemented.
Personally I have hard time noticing the loss of image quality while actually playing the game but I sure as hell notice 20-30+ FPS differences when it turns the game from a borderline playable experience to a good enough one for me.
In some cases DLSS Quality looks even better to my eyes than native TAA regardless of what I'm told or read, if thats what my eyes are telling me then I'm defo using it and get the free performance with higher settings on top.

RT is game/implementation dependant, I do like RT reflections and lights but don't care much for shadows/ambient occlusion and always check how the game looks/feels with or w/o and then decide if I want to use it or not. 'Currently I have RT reflections and lights enabled on Ultra in Cyberpunk and I prefer it this way'

 

[Vayra86]

So we're walking through fully empty environments here and we can look far into the distance: all GPU performance destroyed. Dead Space gets a remaster on Frostbite: all GPU performance destroyed in tight confined environments while your average Battlefield runs a 64 player instance on it just fine.

Funny and coincidental how new card releases coincide with terribly optimized titles and performance analyses Gosh. You'd almost think you had to go out and buy a GPU that can hardly play this at 1200 bucks. Don't you all feel motivated now?

 

[Zubasa]

Ah, yes, AMD must have used some magic pixel fairy dust and make the image look better. That makes the most sense to me.

Same reason why Polaris doesn't run natively in this game but works with DXVK.
This game is just a mess. Not surprisng given the Dev team.

 

[gridracedriver]

DLSS

 

[Vya Domus]

In this case, and it's the only one case, the DLSS is miles worse than the FSR.

Except it isn't, they're about the same with FSR having more shimmering. FSR often looks very similar in terms of sharpness and detail with more shimmering, this is nothing new, you're imagining things.

 

[Dammeron]

All the grass with DLSS was just a blurred mess. I haven't made use of RT or DLSS with my GPUs in any games I've played. I don't see the benefit of them. Perhaps one day they may appeal to me, but for now they're overpriced gimmicks.

It's not just the grass - the whole screen is lacking sharpness and detail.

As for using DLSS - it depends on the implementation - here it's just plain awful, but in some cases, Q mode can look even better than the native mode. And You can't call it a gimmick, if it gives You a lot more performance without having to resort to changing the resolution (worst case scenario), or graphic details. It matter especially to those who don't have a powerful GPU.

 

[EatingDirt]

FSR-Q 2.1 looks sharper then DLSS-Q. Is there a sharpen setting?
The new DLSS 2.5 doesn't make use of the sharpen feature anymore, maybe that "broke" it.
Maybe try DLSS 2.4.

This is mostly the answer to everyone's question on why FSR looks better. It's clear that DLSS has no/little sharpening or FSR just has more. If you zoom WAY in, you'll notice the sharpening 'blocks'.

This is always an issue with comparing DLSS & FSR. Half the time the game doesn't have a sharpening slider and they're at a set level by the developers and you can't change it, the other half of the time, the same sharpening settings in the game actually sharpen the image differently for FSR & DLSS.

 

[AsRock]

Ah, yes, AMD must have used some magic pixel fairy dust and make the image look better. That makes the most sense to me.

If that's so it's a real shame they did not use it on the rest of the game to make it worth while buying in the 1st place.

 

[Dammeron]

If that's so it's a real shame they did not use it on the rest of the game to make it worth while buying in the 1st place.

Don't You have enough dust in the game? It all looks like a giant particle showdown.

 

[AsRock]

Don't You have enough dust in the game? It all looks like a giant particle showdown.

takes more than pretty graphics to make a good game, there are so many other problems with it. The lack of thought in the story, the annoyances of being interrupted all the time and so forth.

Graphics are good enough for me, it's a good part of the rest of the game that sucks. Never mind the gameplay seems kinda Meh at best or like the story to it bland comes to mind.

 

[Gucky]

This is mostly the answer to everyone's question on why FSR looks better. It's clear that DLSS has no/little sharpening or FSR just has more. If you zoom WAY in, you'll notice the sharpening 'blocks'.

This is always an issue with comparing DLSS & FSR. Half the time the game doesn't have a sharpening slider and they're at a set level by the developers and you can't change it, the other half of the time, the same sharpening settings in the game actually sharpen the image differently for FSR & DLSS.

Digital Foundry made a Video for the PC Version and they use DLSS 2.4.12, which seems come with the game. They don't have the problem with the sharpening...but there is a Problem with Motion Blur and specular highlights.
They also show that RTAO and RT Shadows almost don't do anything and don't seem to work properly...

 

[Avro Arrow]

Ok, something's wrong here. In those slider bars, it's quite clear that different shadow settings had been enabled in both FSR and DLSS. When you pass the slider over the rock in the middle of the screen the rock looks white if the shadow is absent or dark grey if the shadow is present. The shadow extends across the grass above so this is not an apples-to-apples comparison. You can't use a comparative slider bar when the two images have different graphics settings!

The shadow also has nothing to do with either FSR or DLSS because at 2160p the extra shadows are absent for Native and DLSS Quality but present for FSR Quality. Then, at 1440p they reverse and are only present for Native and DLSS Quality while absent for FSR Quality. After that, they're all over the place. The other bad aspect of this is the fact that with different shadow settings, the FPS comparisons aren't valid either. This article is one big dumpster fire because of this and I stopped reading it.

The optics of this make it appear that someone really wants to make DLSS Quality look more like native than FSR Quality at higher resolutions and are using the shadow settings to achieve that goal. Only at 1080p was this comparison done properly with the extra shadows not being there across the board. If I had done this graphic, I'd be embarrassed as hell and would've re-done it immediately because I would rather release an article a day late than release it on time with optics as bad as this.

Check for yourself:

2160p:
Native TAA - Shadow is absent
FSR Quality - Shadow is present
DLSS Quality - Shadow is absent
FSR Balanced - Shadow is present
DLSS Balanced - Shadow is absent
FSR Performance - Shadow is absent
DLSS Performance - Shadow is absent
FSR Ultra Performance - Shadow is present
DLSS Ultra Performance - Shadow is absent

1440p:
Native TAA - Shadow is present
FSR Quality - Shadow is absent
DLSS Quality - Shadow is present
FSR Balanced - Shadow is absent
DLSS Balanced - Shadow is absent
FSR Performance - Shadow is absent
DLSS Performance - Shadow is absent

This renders the entire comparison slider useless because the images weren't the same to begin with.

 

[neatfeatguy]

Ok, something's wrong here. In those slider bars, it's quite clear that different shadow settings had been enabled in both FSR and DLSS. When you pass the slider over the rock in the middle of the screen the rock looks white if the shadow is absent or dark grey if the shadow is present. The shadow extends across the grass above so this is not an apples-to-apples comparison. You can't use a comparative slider bar when the two images have different graphics settings!

The shadow also has nothing to do with either FSR or DLSS because at 2160p the extra shadows are absent for Native and DLSS Quality but present for FSR Quality. Then, at 1440p they reverse and are only present for Native and DLSS Quality while absent for FSR Quality. After that, they're all over the place. The other bad aspect of this is the fact that with different shadow settings, the FPS comparisons aren't valid either. This article is one big dumpster fire because of this and I stopped reading it.

The optics of this make it appear that someone really wants to make DLSS Quality look more like native than FSR Quality at higher resolutions and are using the shadow settings to achieve that goal. Only at 1080p was this comparison done properly with the extra shadows not being there across the board. If I had done this graphic, I'd be embarrassed as hell and would've re-done it immediately because I would rather release an article a day late than release it on time with optics as bad as this.

Check for yourself:

2160p:
Native TAA - Shadow is absent
FSR Quality - Shadow is present
DLSS Quality - Shadow is absent
FSR Balanced - Shadow is present
DLSS Balanced - Shadow is absent
FSR Performance - Shadow is absent
DLSS Performance - Shadow is absent
FSR Ultra Performance - Shadow is present
DLSS Ultra Performance - Shadow is absent

1440p:
Native TAA - Shadow is present
FSR Quality - Shadow is absent
DLSS Quality - Shadow is present
FSR Balanced - Shadow is absent
DLSS Balanced - Shadow is absent
FSR Performance - Shadow is absent
DLSS Performance - Shadow is absent

This renders the entire comparison slider useless because the images weren't the same to begin with.

Are you talking about what appears to be the cloud shadows?

 

[Gahl1k]

Ok, something's wrong here. In those slider bars, it's quite clear that different shadow settings had been enabled in both FSR and DLSS. When you pass the slider over the rock in the middle of the screen the rock looks white if the shadow is absent or dark grey if the shadow is present. The shadow extends across the grass above so this is not an apples-to-apples comparison. You can't use a comparative slider bar when the two images have different graphics settings!

The shadow also has nothing to do with either FSR or DLSS because at 2160p the extra shadows are absent for Native and DLSS Quality but present for FSR Quality. Then, at 1440p they reverse and are only present for Native and DLSS Quality while absent for FSR Quality. After that, they're all over the place. The other bad aspect of this is the fact that with different shadow settings, the FPS comparisons aren't valid either. This article is one big dumpster fire because of this and I stopped reading it.

The optics of this make it appear that someone really wants to make DLSS Quality look more like native than FSR Quality at higher resolutions and are using the shadow settings to achieve that goal. Only at 1080p was this comparison done properly with the extra shadows not being there across the board. If I had done this graphic, I'd be embarrassed as hell and would've re-done it immediately because I would rather release an article a day late than release it on time with optics as bad as this.

Check for yourself:

2160p:
Native TAA - Shadow is absent
FSR Quality - Shadow is present
DLSS Quality - Shadow is absent
FSR Balanced - Shadow is present
DLSS Balanced - Shadow is absent
FSR Performance - Shadow is absent
DLSS Performance - Shadow is absent
FSR Ultra Performance - Shadow is present
DLSS Ultra Performance - Shadow is absent

1440p:
Native TAA - Shadow is present
FSR Quality - Shadow is absent
DLSS Quality - Shadow is present
FSR Balanced - Shadow is absent
DLSS Balanced - Shadow is absent
FSR Performance - Shadow is absent
DLSS Performance - Shadow is absent

This renders the entire comparison slider useless because the images weren't the same to begin with.

I'm not sure what your end goal is with this mental gymnastics, but the game uses volumetric cloud system that casts shadows. It's not within the author's ability to hold the shadows in place while recording—that's why they're 'absent' in some images (you can see the shadows moving upwards on the higher rocks and distant building). I also fail to see how the shadows relate to your argument of DLSS being comparable to Native. FSR is the closest to Native—you can see it in the foliage, the buildings, the rocks, the protagonist, and the stone road—while DLSS comes bundled with Vaseline®.