• 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.

Silent hill 2 remake upscaling methods, which is better between Xess and Tsr ?

luca_29_bg

Member
I mean between Tsr low (50 % of render resolution) and Xess performance ( or balanced) (unknown render resolution) . It seems to me that xess apply a sharpen filter to the image but i don't know the preset settings used for both perormance and balanced, balanced has lower frame rate compared to the other two settings, and it slow down more in specific problematic place of this game, so the resolution must be higher than 1440p ? ( 50% of 4k for tsr low). Tsr is blurred in comparison, but much less aliased! Take a look!

Tsr low 4k (as far as i now it's 50% render scale)

Xess balanced 4k

Xess performance 4k ( i don't know the resolution render scale, if someone knows, please tell me, it's sharpen to my eyes, same for balanced, but much more aliased with vegetation!)

 

Gonzito

Gold Member
TSR will give you better anti aliasing and image quality but probably worse performance

Xess will give you better performance but worse image quality and stability
 

adamsapple

Or is it just one of Phil's balls in my throat?
TSR will give you better anti aliasing and image quality but probably worse performance

Xess will give you better performance but worse image quality and stability

Isn't XeSS machine learning thing while TSR is purely software based?

Based on my limited understanding, shouldn't XeSS give better image uality (outside of artifacts) ?
 

Gonzito

Gold Member
Isn't XeSS machine learning thing while TSR is purely software based?

Based on my limited understanding, shouldn't XeSS give better image uality (outside of artifacts) ?

TSR is basically TAA on a much higher quality and XeSS is an upscaler like FSR or DLSS

Meaning, unless we bring DLSS to the table, your best option will always be TSR, in fact in some games I dare to say that TSR looks better than DLSS because there is less artefacting. In Silent Hill 2 in particular, DLSS produces ghosting and artefacting in shadows and character's faces. TSR doesn't

So yeah in the end it depends on the game but for this game in particular, I recommend you TSR, even more than DLSS
 

luca_29_bg

Member
TSR will give you better anti aliasing and image quality but probably worse performance

Xess will give you better performance but worse image quality and stability

Very weird! In my playthrough it's a little better with tsr low compared to xess balanced, i don't know about xess performance, i played the entire game with balanced settings, i tried tsr low in some specific problematic frame rate situations and the fps were a little higher, not so much, but higher!
 

Gonzito

Gold Member
Very weird! In my playthrough it's a little better with tsr low compared to xess balanced, i don't know about xess performance, i played the entire game with balanced settings, i tried tsr low in some specific problematic frame rate situations and the fps were a little higher, not so much, but higher!

You might see better performance because in a way TSR acts similarly to others upscaler as well but its more like an algorythm that uses upsampling from a lower resolution, so depending on the game you will see sometimes some improvements in fps compared to TAA, but generally speaking TSR is quite heavy in my experience, so other upscalers will give you usually more fps
 

luca_29_bg

Member
You might see better performance because in a way TSR acts similarly to others upscaler as well but its more like an algorythm that uses upsampling from a lower resolution, so depending on the game you will see sometimes some improvements in fps compared to TAA, but generally speaking TSR is quite heavy in my experience, so other upscalers will give you usually more fps

Thanks for the explanation!
 

kevboard

Member
Lol what? Show me where I am wrong

XeSS is AI driven upsampling. which needs additional hardware power if you play on anything that isn't an Intel GPU, as the machine learning part of the upsampler will need to run through the GPU shaders

TSR is unreal engine's build in upsampling. it is designed for good performance but is basically only a tiny bit better than FSR, and ultimately uses similar hardware power as FSR, ergo less than XeSS.

quality can of course vary. but XeSS will probably have less artifacts than TSR. although I gotta say I haven't tried both, as I just use DLSS obviously.
 
Last edited:

Gonzito

Gold Member
XeSS is AI driven upsampling. which needs additional hardware power if you play on anything that isn't an Intel GPU, as the machine learning part of the upsampler will need to run through the GPU shaders

TSR is unreal engine's build in upsampling. it is designed for good performance but is basically only a tiny bit better than FSR, and ultimately uses similar hardware power as FSR, ergo less than XeSS.

quality can of course vary. but XeSS will probably have less artifacts than TSR. although I gotta say I haven't tried both, as I just use DLSS obviously.

Okay genius, tell me something. Have you ever tried seeing a comparison between TSR and XeSS in games? Image quality and aliasing quality. TSR is always superior with much less aliasing and artefacting, every single time but you said yourself, you haven't even tried it because you use DLSS

Oh and another one, if I am wrong, why do developers dont use TSR instead of FSR on consoles for unreal engine games according to your logic. If TSR is performance friendly then why do developers don't use it?

Anyway your argument is so wrong because you can try any new modern game on unreal engine 5 game and see in 2 seconds how much better TSR is over XeSS always
 
Last edited:

kevboard

Member
Okay genius, tell me something. Have you ever tried seeing a comparison between TSR and XeSS in games? Image quality and aliasing quality. TSR is always superior with much less aliasing and artefacting, every single time but you said yourself, you haven't even tried it because you use DLSS

Oh and another one, if I am wrong, why do developers dont use TSR instead of FSR on consoles for unreal engine games according to your logic. If TSR is performance friendly then why do developers don't use it?

Anyway your argument is so wrong because you can try any new modern game on unreal engine 5 game and see in 2 seconds how much better TSR is over XeSS always

we don't know why they don't use it. probably because they suck. which is also why Silent Hill 2 uses Lumen GI... because modern game devs suck.

digital foundry tested WuKong on PC with TSR and FSR, they ran identically, so they too wondered why the fuck they used FSR when TSR usually looks a bit better (not much but a bit)

XeSS however runs significantly worse compared to both FSR and TSR because, like I said, it has to run a machine learning algorithm on the shaders, which the other 2 don't do.


edit: also to clear things up. I did use FSR and XeSS before of course and compared the 2 to eachother before. just not in this specific game.
and in literally every game I did compare them XeSS looks significantly better. so I doubt that TSR, which is basically a minutely better upsampler than FSR, looks better than XeSS when using the same internal resolution. XeSS has different labels than TSR, so you can't compare TSR quality to XeSS Quality, because XeSS quality mode runs at a lower resolution for example.
 
Last edited:
Top Bottom