• Hey, guest user. Hope you're enjoying NeoGAF! Have you considered registering for an account? Come join us and add your take to the daily discourse.

[DF] Jedi Survivor update for PS5PRO

Sharpening the picture is not the point of ML solutions, you could have that even with FSR1. CLEANER picture is, and PSSR fails to deliver that, picture is much better than in P9 version but still full of image instability (film grain like noise).

Game also has no sharpening options on PC, it probably has zero DLSS sharpness while PSSR clearly has some sharpening pass applied.

What is the most funny is that game is on patch 10 and still full of bugs...



jedi8.gif


A sharpening filter does NOT magically create details.






jedi9.gif



No matter how you crank up your sharpness slider, you want get that rock details of PSSR :messenger_winking:







2-0.gif



And yes, the whole point of the AI upscale solution, is to preserve the image details even at motion, and it's not just about the sharpness.

In the process, it also removes artifacts like ghostings.










LOU-1.gif




LOU-2.gif




LOU-8.gif



It's funny to see how DLSS blurs at motion and shows ghosting artifacts like TAA does, yet people think it's amazing because it's temporarily more stable..

just like the TAA LOL
 

SlimySnake

Flashless at the Golden Globes
I'll ask again

Who puts the DLSS slider to performance?

On top of not changing the DLSS .dll file to more recent versions?
A lot of people. I personally never liked it because it looked more 1440p than 4k but at times I’ve had to bring it down to performance because some levels hit the gpu more than others.
 
I'll ask again

Who puts the DLSS slider to performance?

On top of not changing the DLSS .dll file to more recent versions?
The point is to compare only the AI upscaling solutions so both are using the same native resolution. Their tests were done properly.

The conclusions are very subjective as they are using the stills comparisons to make DLSS win. In motion PSSR is the clear winner.
 

Fafalada

Fafracer forever
Sharpness is not a feature that makes PSSR better, but a filter it uses to hide its limitations
It's not a sharpening filter - in motion (in this specific video) DLSS footage looks significantly blurier / less detailed.
Whether stability in still-shots is preferable or not is another debate - but historically DF has indexed on motion quality far more when it favored DLSS compared to other upscalers.

Why does DF care so much about stability at rest when you aren’t moving?
I mean... ☝️
 

Bojji

Member
jedi8.gif


A sharpening filter does NOT magically create details.






jedi9.gif



No matter how you crank up your sharpness slider, you want get that rock details of PSSR :messenger_winking:







2-0.gif



And yes, the whole point of the AI upscale solution, is to preserve the image details even at motion, and it's not just about the sharpness.

In the process, it also removes artifacts like ghostings.










LOU-1.gif




LOU-2.gif




LOU-8.gif



It's funny to see how DLSS blurs at motion and shows ghosting artifacts like TAA does, yet people think it's amazing because it's temporarily more stable..

just like the TAA LOL


Ghosting is a matter of DLSS presets:

YiRaYtQ.jpeg
rlD9OSP.jpeg




They tried to match settings between PS5 and PC, didn't they? On some pictures you took PS5 appear to have better LOD than PC, more foliage for example. Consoles sometimes use settings that are not matched with medium, high, ultra etc. But contain elements of some of them. You also have negative LOD bias that may be incorrect in PC version.

JS is a mess of a game from technical perspective, AMD forced FSR2 for many months in this title, they implemented DLSS finally but for example frame gen is still broken and they don't give a fuck.
 
Yep.

DF has some terrible takes. It's almost like they're afraid to give PSSR too much credit and that DLSS, being first, must therefore be better. But it's not in practice in this case. I'm not going to sit still on a screen and care that PSSR is slightly more unstable but yet in motion has far better clarity and apparent resolution.

DF ALWAYS have funny ideas when it comes to PC vs console. Like how they make great deal out of half dozen of trees that are size of 3~4 pixels at far distance on PC, yet they dismisses some night and day difference in image clarity on console as 'net positive' A clear bias and double standards. Kinda understandable as they make their living out of the advertisement of PC related products in their videos, but as a journalism stand point, they're absolutely pathetic like the rest of gaming journalists.
 
Ghosting is a matter of DLSS presets:

YiRaYtQ.jpeg
rlD9OSP.jpeg




They tried to match settings between PS5 and PC, didn't they? On some pictures you took PS5 appear to have better LOD than PC, more foliage for example. Consoles sometimes use settings that are not matched with medium, high, ultra etc. But contain elements of some of them. You also have negative LOD bias that may be incorrect in PC version.

JS is a mess of a game from technical perspective, AMD forced FSR2 for many months in this title, they implemented DLSS finally but for example frame gen is still broken and they don't give a fuck.




jediA.gif



So you feel this is due to LOD bias too?







jediB.jpg



they look close enough to have max LOD to me :messenger_tears_of_joy:
 

Tqaulity

Member
I love DF (just spoke to them at CES) but their conclusions here are just wrong on so many levels. They are killing me with the constant comparisons of PS5 Pro/PSSR to the PC where it is clearly not the point. Alex says that PSSR did not come into an "empty market" in terms of reconstruction options....but it actually did! When PS5 PRO launched, there were ~3 ML upscalers available ON PC ...how many were on PlayStation? ZERO! That is the point...that is the market for it. Their conclusions seem to indicate that PSSR is too little too late since the PC space has so many options and the state of the art has evolved. First, any comparisons to DLSS is fundamentally flawed and unfair since A) DLSS is 6 years old and PSSR is brand new and B) even in the same title, the implementation and execution will still vary between a PC and a PlayStation console. This is why it is ridiculous to compare FSR4 against PSSR at this point. We have litterally seen a controlled demo that is a few seconds long of a SINGLE title with FSR4...again with a very different code path than that on the PS5. To say that one looks better than the other is borderline irresponsible (and I saw the FSR4 demo in person at CES myself. IMO, it looked very comparable to PSSR but again was extremely limited content).

Alex says it makes you wonder about the point of the PS5 PRO and PSSR. The point is to deliver a superior experience to the base PS5 ...not to some arbitrary PC config or the "state of the art" for gaming as a whole. PSSR is intended to deliver a superior experience to TAA and FSR which is DOES in 9/10 PRO enabled games. Jedi Survivor (along with Avatar and Silent Hill 2) was one of the few cases where playing on the PRO was noticeably worse than the base PS5 which should absolutely never happen. This patch fixes that and while not perfect, most of the issues lie in the game/engine itself that is evident on Xbox, PC and PS. Oliver says in summary this patch is "better" but it's still not where we'd like it to be...ummm WHERE is that exactly? Compared to what? Native 4K? DLSS 3.7 on a high end PC? Or compared to FSR on the base PS5? Let's not forget that on the PRO, the game is running at nearly 2x the input resolution, higher quality graphics settings, and added RT in performance mode...now with a quality PSSR implementation to boot. Is there anyone here who would really say the PRO is falling short of the base PS5 experience now? All I can do is pray for you if that is the case.
 
Last edited:

Buggy Loop

Gold Member
I do - when I am trying to play modern games (AW2, Indiana Jones) at 60fps on a 4080.

surprise wtf GIF


Even on a 3080 Ti ultrawide 1440p I keep DLSS quality on modern games, even if it means I'll find optimized settings because most of the max/ultra/extreme/uber/will it run settings are downright waste of ressources for a lot of them. Are you trying to do 4k path tracing max settings or something?


And technically AW2 needs ultra performance to hold stable 60 - but that's a whole other debate.

Haven't played AW2 because epic store but what's going on here?

dlss-pt-3840.png


I mean I know 4090 is a beast but 4080 must not be that far behind.
 
Last edited:

Bojji

Member
that looks more to do with the foliage option set to higher than high settings of PC on PS5 PRO than LOD.

more foliage aside, you can still see the clear difference in image clarity, and the genuinely higher detail on PRO side.

Could be, AC Valhalla was like this. This game is a mess.

Sharpness can make one image appear to have "more detail"

oVLzPxT.jpeg


You oversharpened DLSS image before but it's not the same as in game sharpness or DLSS sharpness slider in game.
 

yogaflame

Member
I have respect for DF and I trust there technical know how, however there are times that they cannot hide being a PC warrior. But going back. Its still a big improvement and growth for pssr in just a very short time. It continues to learn at a fast pace.
 

kevboard

Member
I know what a sharpening filter is and this ain’t it. DLSS blurs a lot of details that are preserved with PSSR. This isn’t a sharpening issue. It’s the temporal accumulation shitting the bed on with DLSS.

which is why people should replace that old ass .dll they are using in this game with 3.8 and use preset E. that should improve it quite a lot.
 

Venom Snake

Gold Member
Assuming that in order to progress in the game you need to move, without a second thought i would choose a less stable but clear and sharp image over something that looks like a perfectly stable shitstain on a fat man's post-burrito underwear.

It looks like it's finally time to buy this game.
 

SKYF@ll

Member
Jedi Survivor (PC version) has been updated many times, but the DLSS version remains outdated?
If the LOD and leaf detail of the PS5 Pro version are high, why is there no improvement in the PC version?
PSSR is less stable than DLSS overall, but it seems to be better in terms of clarity and motion.
We need to acknowledge the pros and cons of PSSR and DLSS.
There is no need to compete, and we can expect that any upscaler will be much better in a year.
https://imgbox.com/VMZ6g0yo.png
 
Could be, AC Valhalla was like this. This game is a mess.

Sharpness can make one image appear to have "more detail"

oVLzPxT.jpeg


You oversharpened DLSS image before but it's not the same as in game sharpness or DLSS sharpness slider in game.





DLSS.gif



looks about the same to me :messenger_winking:










jediC.gif



And this one.. not even close LOL
 
Last edited:

Fabieter

Member
Digital Foundry:
  • Overall, DLSS is the clear winner
  • This game uses DLSS 3.1 by default, which is 2 years old at this point
  • DLSS4 is less than 2 weeks away, which will increase the gap even further
  • PSSR is still not a a clear upgrade over FSR2 in this game
  • FSR4 is also right around the corner

Platform warriors:

"YEAH! SUCK IT PC GAMERS!!!11"

Iam not a fan of the pro and wasnt from the get go but some of those DF guys are as unprofessional as it gets.
 

Fafalada

Fafracer forever
Are you trying to do 4k path tracing max settings or something?
I mean yes 4k is the output resolution (else those DLSS settings wouldn't be meaningful to discuss relative to the console especially).
But AW2 PT craters the framerate at 'any' setting that enables it - in the forests(and the mind-place, that's really heavy for some reason), it's impossible to stay at 60 without either disabling PT alltogether, or dropping it to Ultra performance. Other areas will vary - there are places that hold well with Performance or even Balanced using low PT presets, but the game doesn't offer an adaptive option so that's all moot.
I actually contemplated using Ultra Perf but that just gets way too blurry to tolerate on a big screen TV. But its just really difficult to swallow running this game without PT (pretty much the only reason I got it lol🤦‍♂️) as then I might as well play the console version. I did also try various frame-gen options but the frame pacing is just really bad when running sub 60/without VRR, and the input-lag starts to get in the way of the already yanky gameplay (also why I don't want to play it at 30). Maybe I need to wait until I get a new TV in the end (not sure I'm in the market for GPU anytime soon)...

Sort of reminds me of the problems I had with Control (back when I was still on a 2080ti - which just 'couldn't' keep 60 at balanced, I had to drop to Performance even after trying to overclock the GPU) - it's like Remedy always knows to release games that I don't have a good enough GPU for yet :messenger_grinning_sweat:

Indiana Jones I haven't gotten as far in yet - Jungle was sub 60 with any of the RT options enabled - but the college level after ran very well, even with full options and higher DLSS settings, so it seems to be even more content dependant.

I mean I know 4090 is a beast but 4080 must not be that far behind.
🤷‍♀️
 
Last edited:

Topher

Identifies as young
Iam not a fan of the pro and wasnt from the get go but some of those DF guys are as unprofessional as it gets.

Don't take that as representative of what was said. I posted the accompanying article if you want a non-BS summary.


Bottom line: genuine improvements.
 

Bojji

Member
DLSS.gif



looks about the same to me :messenger_winking:










jediC.gif



And this one.. not even close LOL

+10 if for sure less aliased.

Different games have different levels of PSSR and DLSS implementations (JS is AMD sponsored PoS tech level game). You also had the same res test in AW2 at launch and it looked like this:

j2XDKVI.jpeg
f98UmEH.jpeg
2P0a7KD.jpeg
 
Last edited:

yamaci17

Member
+10 if for sure less aliased.

Different games have different levels of PSSR and DLSS implementations (JS is AMD sponsored PoS tech level game). You also had the same res test in AW2 at launch and it looked like this:
what do they mean by "high" settings? yeah, high preset is all right but I really don't believe PS5 pro would use "high" texture quality option. that wouldn't even make sense, it has a lot of VRAM, I'm sure it allocates as much as possible memory budget as it can

on PC, if you set the preset to "high", the textures will also be set to high, and in that game, high option is tuned towards 6-8 GB GPUs and often uses a low texture budget. it is entirely possible that engine is limited to a low texture budget size due to that

no idea why they didn't max out textures at least on the PC to ensure that it is not a texture budget size issue. I'm sure the PS5 pro runs texture pool size at a setting that would be equal or higher than the ultra option on PC. actually it is entirely possible that it is completely dynamic on the console. and I'm sure it would still end up with more budget than the ultra option on PC actually

to me it looks like a texture quality difference

(im talking about the jedi survivor comparisons)
 
Last edited:

Bojji

Member
what do they mean by "high" settings? yeah, high preset is all right but I really don't believe PS5 pro would use "high" texture quality option. that wouldn't even make sense, it has a lot of VRAM, I'm sure it allocates as much as possible memory budget as it can

on PC, if you set the preset to "high", the textures will also be set to high, and in that game, high option is tuned towards 6-8 GB GPUs and often uses a low texture budget. it is entirely possible that engine is limited to a low texture budget size due to that

no idea why they didn't max out textures at least on the PC to ensure that it is not a texture budget size issue. I'm sure the PS5 pro runs texture pool size at a setting that would be equal or higher than the ultra option on PC. actually it is entirely possible that it is completely dynamic on the console. and I'm sure it would still end up with more budget than the ultra option on PC actually

to me it looks like a texture quality difference

(im talking about the jedi survivor comparisons)

I said earlier that DF could have used high settings in comparison when in reality PS5 Pro version could use some higher than high presets (or custom ones).

It could be higher res textures or negative lod bias not set up correctly by devs. Draw distance is for sure better on Pro in their comparison.
 
Last edited:
+10 if for sure less aliased.

Different games have different levels of PSSR and DLSS implementations (JS is AMD sponsored PoS tech level game). You also had the same res test in AW2 at launch and it looked like this:

j2XDKVI.jpeg
f98UmEH.jpeg
2P0a7KD.jpeg




1.gif





2.gif





3.gif





4.gif




Even in games like AW2, PSSR provides significantly better image clarity at motion, I don't get what's the point of comparing static scenes.









DLSSvs-FSR.gif



In a still scene even FSR looks pretty much identical to DLSS, what's the point or having all these AI hardware to do the fancy ML upscaler when all you do is to compare static scenes?
 
surprise wtf GIF


Even on a 3080 Ti ultrawide 1440p I keep DLSS quality on modern games, even if it means I'll find optimized settings because most of the max/ultra/extreme/uber/will it run settings are downright waste of ressources for a lot of them. Are you trying to do 4k path tracing max settings or something?




Haven't played AW2 because epic store but what's going on here?

dlss-pt-3840.png


I mean I know 4090 is a beast but 4080 must not be that far behind.
According to techpowerup the RTX4090 is around 24% faster compared to RTX 4080S (95fps vs 118fps average fps at 4K native based on 25 games tested).

As for Alan Wake 2, performance in this game fluctuates a lot depending on the location

Less demanding location on my RTX4080S

4K DLSS Performance, max settings, PT

AW2-DLSSP.jpg


FG

DLSSP-FG.jpg


1440p DLSS Quality, max settings, PT

1440p-2-DLSSQ.jpg


FG

1440p-2-DLSSQ-FG.jpg


More demanding location

2160p DLSS performance, max settings, PT

DLSSP-2.jpg


FG

DLSSP-FG-2.jpg


1440p DLSSQ, max settings, PT

1440p-DLSSQ.jpg


FG

1440p-DLSSQ-FG.jpg
 
Last edited:

Buggy Loop

Gold Member
According to techpowerup the RTX4090 is around 24% faster compared to RTX 4080S (95fps vs 118fps average fps at 4K native based on 25 games tested).

As for Alan Wake 2, performance in this game fluctuates a lot depending on the location

Yea I don't think I'll play the game but it will be interesting to see what mega geometry does to this game's performances. Supposed to help a lot for forests.
 

kevboard

Member
I'll ask again

Who puts the DLSS slider to performance?

On top of not changing the DLSS .dll file to more recent versions?

to be fair, for that last part it's the devs fault for not updating it since launch.

especially performance mode is highly variable from version to version, and even more so from preset to preset. with the best preset for performance mode not being present in the version they are using 😬

shouldn't be too surprising given how shoddily made the game in general is
 

omegasc

Member
Weird that they mention it but didn't update the DLSS DLL to add to the comparison. It would have been good info.
They don't mention the texture loading issues (perf mode) as well.
I encountered lots of problematic areas where the fps tanks for no apparent reason, like a small cave with nothing but a chest. but worst offender is random force close bugs when using fast travel. I thought my save was busted because when using the "Continue" option it was always closing the game, but using "Load game" solved that issue and nothing was lost.
Also got 'stuck' on the cafeteria entrance once today, but going to the PS Menu by holding the PS button and coming back fixed that.
In the end, patch 10 is a LOT better to play, and the current glitches are not distracting to me.
 

Bojji

Member
1.gif





2.gif





3.gif





4.gif




Even in games like AW2, PSSR provides significantly better image clarity at motion, I don't get what's the point of comparing static scenes.









DLSSvs-FSR.gif



In a still scene even FSR looks pretty much identical to DLSS, what's the point or having all these AI hardware to do the fancy ML upscaler when all you do is to compare static scenes?


You can see AW2 in motion here:



Now returning to JW. I downloaded the game, turns out DLSS is much less sharp than even FSR 2! They clearly implemented it in half ass way, 0 sharpness with no in game control of it when PSSR and FSR have sharpening pass:

https://imgsli.com/MzM5MzU1

0fBvNTL.jpeg
U3Eqcw5.jpeg
 
Last edited:

yamaci17

Member
You can see AW2 in motion here:



Now returning to JW. I downloaded the game, turns out DLSS is much less sharp than even FSR 2! They clearly implemented it in half ass way, 0 sharpness with no in game control of it when PSSR and FSR have sharpening pass:

https://imgsli.com/MzM5MzU1

0fBvNTL.jpeg
U3Eqcw5.jpeg

thanks. please test high vs ultra texture quality option as well. some of these unreal engine 4 games are shipped with presets where high option uses 3 GB of texture memory. that is nowhere near enough to load high quality 4K textures. epic option usually is 5 GB or higher

regardless, it is clear that sharpening is also doing a heavy lifting there for FSR (and probably for PSSR too)

while you're at it, if you're familar try enabling dlss overlay and see if it uses some weird rarely used preset such as B that is incredibly blurry. I believe C and E are decent options but certain presets can get super blurry real quick. I think those presets were made for games that lacked lots of motion vectors so DLSS acts as if it misses a lot of vectors so blurs lots of stuff even when there's no need to. or just upgrade the DLL to the latest one since that only has the E preset lol
 
Last edited:

Bojji

Member
thanks. please test high vs ultra texture quality option as well. some of these unreal engine 4 games are shipped with presets where high option uses 3 GB of texture memory. that is nowhere near enough to load high quality 4K textures. epic option usually is 5 GB or higher

regardless, it is clear that sharpening is also doing a heavy lifting there for FSR (and probably for PSSR too)

while you're at it, if you're familar try enabling dlss overlay and see if it uses some weird rarely used preset such as B that is incredibly blurry. I believe C and E are decent options but certain presets can get super blurry real quick. I think those presets were made for games that lacked lots of motion vectors so DLSS acts as if it misses a lot of vectors so blurs lots of stuff even when there's no need to. or just upgrade the DLL to the latest one since that only has the E preset lol

Yep, I will update whole DLSS now but here is one more comparison of FSR 2 (labeled "far" lol) and DLSS:


6LTXpVY.jpeg
Au4AKR9.jpeg
 

SKYF@ll

Member
You can see AW2 in motion here:



Now returning to JW. I downloaded the game, turns out DLSS is much less sharp than even FSR 2! They clearly implemented it in half ass way, 0 sharpness with no in game control of it when PSSR and FSR have sharpening pass:

https://imgsli.com/MzM5MzU1

0fBvNTL.jpeg
U3Eqcw5.jpeg

So is it currently possible to make Jedi Survivor's DLSS clearer than PSSR and eliminate ghosting? Or is it impossible?
 

Dorfdad

Gold Member
Going by the video, PSSR does look sharper than DLSS in motion in this game but like Alex said this game is using an older version of DLSS which is DLSS 3.1, so this isn't comparing the latest version of DLSS with PSSR. If Jedi Survivor used the newest version of DLSS then the results might be different.
It against you but it isn’t don’t does t matter. If switch could handle 4k it would be a better console than ps5
 

yamaci17

Member
So is it currently possible to make Jedi Survivor's DLSS clearer than PSSR and eliminate ghosting? Or is it impossible?

you can somewhat eliminate ghosting if the game ships with a preset that causes ghosting by default (just using the latest DLL would be enough)

(this video is outdated however, as preset E was unused at that time, nowadays preset E is like the preset C without ghosting).
as for the clear part, no idea. this video gives a good look on how different presets can actually affect ghosting but they look similar.

but I know for a fact that preset A can be blurrier than other presets


so what does jedi survivor use by default? I have no clue. i don't remember. someone has to check. I wouldn't actually be surprised if it ends up using the preset A lol. could be preset D too.
 
Last edited:
Top Bottom