Support NeoGAF

[Bojji]

I still don't know why this troll hasn't been banned yet

 

[MazingerDUDE]

Yeah, "some stuff"

So, PSSR1 was not scrapped after all. Right?

 

[Bojji]

So, PSSR1 was not scrapped after all. Right?

In 2024? Maybe not. And we don't know that this "stuff" is, because old int8 version is almost identical to the full release (and that was done when Sony and AMD weren't really doing stuff together).

But right now?

 

[MazingerDUDE]

In 2024? Maybe not.

But right now?

still lives within PSSR2

 

[Bojji]

still lives within PSSR2

PSSR2 is FSR4.1, Cerny said this himself.

 

[MazingerDUDE]

And we don't know that this "stuff"

It must be something great that even AI upscaler as competent as INT8 FSR4 would incorporate.

 

[Bojji]

It must be something great that even AI upscaler as competent as INT8 FSR4 would incorporate.

But what is it? We don't know the details of FSR4, PSSR1 and amethyst outside of few PR statements and Kepler leak.

 

[MazingerDUDE]

But what is it? We don't know the details of FSR4, PSSR1 and amethyst outside of few PR statements and Kepler leak.

From DF analysis, there's ghosting found in INT8 FSR4.

And you know what PSSR is best at

 

[J Required]

From DF analysis, there's ghosting found in INT8 FSR4.

And you know what PSSR is best at

Latest DLSS has sacrificed some stability for sharpness, strange decision vs DLSS4.

PSSR is absolutely fantastic now.

 

[Bojji]

From DF analysis, there's ghosting found in INT8 FSR4.

And you know what PSSR is best at

Are you really with this gif again? Switch DLSS3 to preset C in last of us1 and you won't see it (or upgrade file to 3.8 version), it's not DLSS fault that LoU developers have chosen the worst DLSS preset ever (D).

Every upscaler has some ghosting, just like TAA that they are based on. And this can vary depending on preset alone, you don't need to change the whole upscaler to have different parameters on that (nvidia has like million different versions now).

 

[winjer]

From DF analysis, there's ghosting found in INT8 FSR4.

And you know what PSSR is best at

That is a bug in the first dll of the comunity complied version of FSR4 4.0.02 Int8.
It was fixed in the 4.0.2b dll. It also fixed performance issues with it, resulting in 10-20% reduction upscaling time.
The analysis that DF made is very out of date.

 

[MazingerDUDE]

Are you really with this gif again? Switch DLSS3 to preset C in last of us1 and you won't see it (or upgrade file to 3.8 version), it's not DLSS fault that LoU developers have chosen the worst DLSS preset ever (D).

Every upscaler has some ghosting, just like TAA that they are based on. And this can vary depending on preset alone, you don't need to change the whole upscaler to have different parameters on that (nvidia has like million different versions now).

Calm down, this is not about the DLSS

One of the best properties of PSSR is that it's very good at removing ghosting.

INT8 FSR4 had ghosting issue which had been eradicated in the final version.

Now we can guess what this 'some stuff' of PSSR that's incorporated into FSR4 right?


Nonetheless, why does it even matter in the first place?

Whichever it is, PSSR1 was used in creating FSR4 and that's a hard fact.

 

[MazingerDUDE]

That is a bug in the first dll of the comunity complied version of FSR4 4.0.02 Int8.
It was fixed in the 4.0.2b dll. It also fixed performance issues with it, resulting in 10-20% reduction upscaling time.
The analysis that DF made is very out of date.

This is what the bugged one looked like

 

[Bojji]

Nonetheless, why does it even matter in the first place?

Whichever it is, PSSR1 was used in creating FSR4 and that's a hard fact.

Is it? There was some collaboration between Sony and AMD but that was already late in development for FSR4.

I agree that that this discussion is getting nowhere because we don't have enough info. I just don't see any point in defending PSSR1 when it's now obsolete for PS5 Pro users and game developers. And I'm not talking about you, some die hard fanboys just can't cope with that massive price increase and have to filter their anger on me lol.

PSSR2 offers PC like upscaling quality on the console, and that is fucking amazing.

 

[MazingerDUDE]

I just don't see any point in defending PSSR1

Because it was a decent upscaler that provided comprehensive 4K 60 experience for the first time ever in console space?

 

[Bojji]

Because it was a decent upscaler that provided comprehensive 4K 60 experience for the first time ever in console space?

Ok, that's your opinion (and it's true for some games).

But we won't see in any games moving forward, and it was weaker than any other ML upscaler on the market in 2024. So why?

Let's forget about it and move on...

 

[MazingerDUDE]

it was weaker than any other ML upscaler on the market in 2024. So why?

Because none of those ML upscalers are available in console space

 

[Bojji]

Because none of those ML upscalers are available in console space

That is true.

 

[Feel Like I'm On 42]

I get the frustration, but I think people are oversimplifying what's actually being said.

It's not "GTA VI will be 30fps just because old Rockstar games were." That would be a dumb argument.

The point is more about how Rockstar designs their games. They consistently push simulation, density, and world detail to the point where 30fps becomes the practical target on console. GTA VI looks like it's doing that again, probably even more than before.

So when people bring up past games, it's not nostalgia logic, it's pattern recognition. Same design philosophy, same kind of technical demands, same CPU limits on current consoles.

As for PSSR2, yeah, that was unexpected, but that's mostly helping GPU-side stuff like image quality. GTA's main bottleneck is still going to be CPU and simulation, and that's not something upscaling really fixes.

So I wouldn't call it a dumb argument, just maybe a poorly explained one.

I think you're mixing up two different arguments a bit.

Nobody serious is saying "it'll be 30fps just because games used to be." That's not the point. The argument is that Rockstar tends to push simulation and CPU-heavy systems way harder than most devs, and that usually ends up at 30fps on console.

And yeah, most big games this gen have performance modes, but most of those games also aren't doing what GTA does under the hood. They'll scale things back aggressively to hit 60. Rockstar historically doesn't like doing that if it compromises the feel of the world.

You're right that we don't know for sure. It's definitely possible they offer a performance mode. But it's not crazy to be skeptical either, especially when:

• The CPU on current consoles hasn't really jumped this gen
• GTA VI looks significantly more simulation-heavy than most open-world games
• Rockstar has a track record of prioritizing world fidelity over frame rate

So I'd put it like this: it's not "impossible," but it's also not something I'd just assume is happening because other games do it. GTA isn't really playing by the same rules as most of those games.

The CPU jump this gen was quite big whadya mean it wasn't? It's just that 5 years into the gen it is starting to feel left in the dust compared to PC.

There was so much talk about the CPU and how we'd finally see more games with 60 fps this gen and that was all true. I do think it sucks the Pro didn't get a CPU upgrade though. Imagine how much better it would be if it wasn't handicapped by the same Ryze 2 cpu. We'd be seeing a lot higher quality RT if Sony hadn't skimped on the Pros cpu

 

[Senua]

Because none of those ML upscalers are available in console space

So much winking bro, I think Boji has a girlfriend though so you're barking up the wrong tree

 

[Feel Like I'm On 42]

Does anyone know if First Berzerker and Wu Chang Fallen Feather use PSSR? Both on sale right now.

new PSSR is awesome but now Im realizing how many games never got PSSR even tho they're Pro Enhanced, and that sucks.

Shame on Kojima for not having it in there esp considering how fucked up the foliage and wated look during daytime in DS2. Almost every MS game does not use it and the Outer Worlds 2 devs are Bloober levels of retarded for removing PSSR- why wouldn't they leave it as an OPTION?!

Sony and Naughty Dog: no Uncharted 4 update is really shitty for anyone that doesn't have a VRR tv.

FInal Fantasy 16- fuck off Square. Adding it to 7 Rebirth but not 16 which really needed it.

Plague Tale Requiem never got any kind of patch and looks bad at 60 fps.

MS again. All their Pro enhanced games are a joke except for Hellblade 2. Oblivion is a mess. FH5- was never given a higher quality AA solution and still has really ugly foliage as a result. This game was so over praised for its Pro update. It looks identical to Series X. Stalker 2- smeary ass TAA solution. Many issues. Avowed-how did this get Pro Enhanced tag anyway? Even DF said it isn't enhanced in any way compared to Series X! Not even higher resolution...Doom and Indiana Jones also lame enhancements.

Borderlands 4- no Pro update. If ever there's a game that should've been updated for Pro its this one.

Dying Light the Beast- developers lied about the Pro version and haven't made any efforts to improve it. All it has going for it is higher dynamic resolution when these devs were acting like it would use the Pro to its fullest potential prior to launch. I was hoping this would get RTGi on the Pro after I read the devs comments but boy was I wrong to have faith in Techland.

FromSoftware- my favorite developer but Ive started to really dislike them as a company. Elden Ring despite its performance problems on ps5 and selling 30 million copies couldn't get some patches from these guys?

Red Dead Redemption 2- nothing needs to be said here. Game has been out for 8 YEARS but we can only play it at 30 fps on the Pro

Don't get me wrong im loving PSSR2! It just highlights how many games need it now that we know its so good.

 

[Jesb]

I don't even know why pssr would be a toggle. Who would want this to be off. Nobody.

 

[Shmunter]

I don't even know why pssr would be a toggle. Who would want this to be off. Nobody.

It's like the ps4 pro overclocking ps4 games toggle. Why?

It has to do with some kind of publisher agreement where they can't fuck with their intent automatically.

 

[thtJhnn]

Now that PSSR 2.0 has been out a while, which games for you have shown the most gains with PSSR 2.0?

 

[Bojji]

I don't even know why pssr would be a toggle. Who would want this to be off. Nobody.

I think they will white list more and more games after QA testing (and they bypass the toggle) and maybe make toggle "on" by default. There are no downsides to this.

Now that PSSR 2.0 has been out a while, which games for you have shown the most gains with PSSR 2.0?

- SH2
- SHf
- Outlaws
- Jedi Survivor
- Avatar
- Hogwart's Legacy
- FFVII Rebirth
- Hell is Us
- SM2 (and old games as well)
- GT7
- Ratchet...

List goes on and on, basically every PSSR game will show some improvements and sharper image.

 

[MisterXDTV]

Does anyone know if First Berzerker and Wu Chang Fallen Feather use PSSR? Both on sale right now.

Negative, they use temporal upscalers as far as I know.

 

[DeepEnigma]

I'd say so:

Which game was he talking about here?

 

[Vick]

What game?

Which game was he talking about here?

Crimson Desert.

He played the pre-D1 version using the old PSSR, and then again with PSSR2.

 

[PicoLordens]

Took the opposite approach. Will leave it off and just hope for patches. If PSSR 1 was good enough before 2, it'll be good enough now if they are never patched.

Are u like 5 or what?

 

[Gusy]

Now that PSSR 2.0 has been out a while, which games for you have shown the most gains with PSSR 2.0?

I'll nominate Avatar and Last of Us Part 1.

Avatar is just crazy, given all the small detail and variety of foliage on display. Its nuts that we can have that geometric density at that image quality running at an almost locked 60 fps.

With LoU Part I... Its probably the best image quality Ive seen on a console game. To me it looks better than native in most scenes (The opening of the game with Tess in daylight is jaw-dropping)

 

[Lagspike_exe]

Jedi looks phenomenal now. Super sharp and detailed.

 

[phant0m]

This video Is making the Outlaws sale even more attractive. And it's ok to get it digital since the disc version is literally useless.

Game is fantastic, a real shame most of GAF shunned it because Kay was 'ugly'

Best SW atmosphere in-game since KOTOR II

 

[saintjules]

With LoU Part I... Its probably the best image quality Ive seen on a console game. To me it looks better than native in most scenes (The opening of the game with Tess in daylight is jaw-dropping)

I may need to play it one more time on the hardest difficulty. Same for TLOU2 before I officially retire these games. Been waiting to play them under the new update.

 

[RPS37]

Did they test TLOU2 or Ragnarok?

 

[dgrdsv]

PSSR2 is FSR4.1, Cerny said this himself.

Which was another one of Cerny's... exaggerations.
FSR 4.1 is an FP8 based NN upscaler which can't run on PS5Pro due to it lacking the format support in its AI h/w.
The new PSSR is likely a port of FSR4 to INT8, early version of which has probably been leaked from AMD's GitHub.
Calling FSR 4.1 and the new PSSR the same is incorrect since the NNs used for pixel weighing are different, even if the end result is close visually and the framework around them is the same.
The closest existing example would be XeSS DP4a vs XMX version on PC.

 

[Radical_3d]

Game is fantastic, a real shame most of GAF shunned it because Kay was 'ugly'

Best SW atmosphere in-game since KOTOR II

My problem is the open world stuff. This could've been a perfect Uncharted-like.

 

[Bojji]

Which was another one of Cerny's... exaggerations.
FSR 4.1 is an FP8 based NN upscaler which can't run on PS5Pro due to it lacking the format support in its AI h/w.
The new PSSR is likely a port of FSR4 to INT8, early version of which has probably been leaked from AMD's GitHub.
Calling FSR 4.1 and the new PSSR the same is incorrect since the NNs used for pixel weighing are different, even if the end result is close visually and the framework around them is the same.
The closest existing example would be XeSS DP4a vs XMX version on PC.

Full on DLSS was running in Int8 on Turing and Ampere using Tensor cores (PSSR2 is running on custom ML hardware in Pro), I don't think it's comparable to upscaler using DP4A running on shaders (and with lower quality compared to dedicated ML cores on intel GPUs).

Quality of PSSR2 and FSR4.1 is very close to being 1:1.

 

[Blizzje]

I've been playing Monster Hunter Wilds with the PSSR2 upgrade and just to get a feeling for this: does this look like the PC version using DLSS Performance at 4K? Although the textures can be bad here and there, image quality is really good I have to say. Any comparisons by someone who owns the PC version and is willing to check? DLSS Performance, DLSS Quality at 4K, how does the PS5 pro version hold up?

 

[Bojji]

I've been playing Monster Hunter Wilds with the PSSR2 upgrade and just to get a feeling for this: does this look like the PC version using DLSS Performance at 4K? Although the textures can be bad here and there, image quality is really good I have to say. Any comparisons by someone who owns the PC version and is willing to check? DLSS Performance, DLSS Quality at 4K, how does the PS5 pro version hold up?

I think monster hunter is using PSSR with 1080p input and 1080p output (so like DLAA), then it's just upscaled to 4k.

It would look much better with 1080->4k reconstruction (but I guess they don't have resources for that, it's heavier than just 1080p).

 

[Blizzje]

I think monster hunter is using PSSR with 1080p input and 1080p output (so like DLAA), then it's just upscaled to 4k.

It would look much better with 1080->4k reconstruction (but I guess they don't have resources for that, it's heavier than just 1080p).

Yeah, I read that in an analysis. However, could it also be that the PSSR 1.0 implementation was lacking and therefore was thought to be used like DLAA?
Right now, when using PSSR2, I'm wondering if it can be compared to DLSS Quality of Performance at 4K. DF mentioned that it looks 'more like a 4k output now' if I remember correctly.

 

[Skifi28]

I've been playing Monster Hunter Wilds with the PSSR2 upgrade and just to get a feeling for this: does this look like the PC version using DLSS Performance at 4K? Although the textures can be bad here and there, image quality is really good I have to say. Any comparisons by someone who owns the PC version and is willing to check? DLSS Performance, DLSS Quality at 4K, how does the PS5 pro version hold up?

I can't post any comparisons right now, but going by eye I'd say it looks very comparable to running at 1440p DLSS quality with DLSS4. It used to look much worse at launch with DLSS3.

 

[Blizzje]

I can't post any comparisons right now, but going by eye I'd say it looks very comparable to running at 1440p DLSS quality with DLSS4. It used to look much worse at launch with DLSS3.

So it isn't quite up to the level of 4K DLSS Quality, in your eyes?

 

[Jam Dingo]

I'll nominate Avatar and Last of Us Part 1.

Avatar is just crazy, given all the small detail and variety of foliage on display. Its nuts that we can have that geometric density at that image quality running at an almost locked 60 fps.

With LoU Part I... Its probably the best image quality Ive seen on a console game. To me it looks better than native in most scenes (The opening of the game with Tess in daylight is jaw-dropping)

I have to say I think TLOU 2 is the best upgrade. I've not played either that or TLOU 1 that much since the toggle, maybe an hour each, but for me 2 edges it. Which is crazy when you think it ALREADY had great IQ (comparable to the 4k mode) with PSSR 1. But it looks absolutely insane now. Literally the best image quality I've ever seen. It's like looking at those supersampled screenshots they release to the media.

 

[Skifi28]

So it isn't quite up to the level of 4K DLSS Quality, in your eyes?

I'm on a 1440p monitor on PC thus I haven't tried 4k quality so I can't make such a direct comparison. I imagine it'd be better, but it's a muddy looking game so I wouldn't expect big differences.

 

[Jam Dingo]

My problem is the open world stuff. This could've been a perfect Uncharted-like.

Tbf you can pretty much ignore all that at almost no cost to the story. I think I did a couple of side missions max, and only the prompted upgrades to the blaster and ship etc.

 

[Blizzje]

I'm on a 1440p monitor on PC thus I haven't tried 4k quality so I can't make such a direct comparison. I imagine it'd be better, but it's a muddy looking game so I wouldn't expect big differences.

Interesting, thanks. Looking forward to someone who is willing to take the PS5 Pro and PC version side by side and test 4K DLSS Quality. Interested to see if PSSR2 holds up.

 

[Physiognomonics]

Having the PSSR toggle hidden away in a submenu and set 'off' as default is crazy.

I'll wager fewer than 5% of PS5 owners have it activated.

0% of PS5 owners have it activated. ;-)

 

[omegasc]

I think monster hunter is using PSSR with 1080p input and 1080p output (so like DLAA), then it's just upscaled to 4k.

It would look much better with 1080->4k reconstruction (but I guess they don't have resources for that, it's heavier than just 1080p).

given v2.x quality, I believe they could even do 846p->4K and it would end up looking better than the current 1080p

 

[MadViking]

Any new games coming soon with PSSR2? I guess Pragmata and Saros are pretty much certain. Maybe Starfield? Not sure if they'll bother for South of Midnight..

 

[dgrdsv]

Full on DLSS was running in Int8 on Turing and Ampere using Tensor cores (PSSR2 is running on custom ML hardware in Pro), I don't think it's comparable to upscaler using DP4A running on shaders (and with lower quality compared to dedicated ML cores on intel GPUs).

DLSS versions 1-4 were using FP16, not INT8, and if it would be using INT8 it would be completely fine to compare that to DP4a XeSS version as this one is also using INT8; the difference would be in performance, with tensor cores having several times higher throughput than anything without them.

Quality of PSSR2 and FSR4.1 is very close to being 1:1.

Nope, there are definitely differences, people just weren't paying enough attention to catch them.