• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

AMD's FSR3 possibly next month ?

Nexus18 said:
Lets wait and see how FSR 3 looks outside of the sponsored titles first before starting with the "dlss killer" hype again.....
Click to expand...

From what I saw FSR3 lost a lot of detail from native. The the Forspoken 'cliff' view, rock to the right lost a lot of definition, cliff in the background lost definition and shadows, grass looked like it droped a texture setting.
What I saw didn't impress me, I'd rather have a good 30% than a mediocre 300%.

Hopefully the quality tier is still decent and what they presented was just to get the high multiplier for marketing
 
Aegis said:
My prediction:

  • FSR3 frame-gen with Anti-Lag+ on RDNA3* GPUs will be broadly comparable with DLSS3 (I'd expect slightly worse latency).
  • FSR3 frame-gen on Intel and Nvidia (especially low/mid-range cards) will look ok, but have terrible latency.
  • HYPR-RX (driver-level frame gen & upscaling) will look horrible and only be used as an absolute last resort.

"AMD was a little guarded when they showed us the TLOUp1 demo, because it was early and an optical flow-only solution will not match the quality of FSR 3 or DLSS 3 in most content. This is a tech with obvious limitations."

*not sure if RDNA2 will support Anti-Lag+..?
Click to expand...


Biggest problems with FSR:

* just like every other version of FSR it will have significantl image quality reduction compared to dlss 3.5

* fsr3 is computational heavy; as such only recent GPUs support it, amd specifically says enabling fsr3 on old GPUs will result in performance loss not performance gain
 
Last edited:
decto said:
From what I saw FSR3 lost a lot of detail from native. The the Forspoken 'cliff' view, rock to the right lost a lot of definition, cliff in the background lost definition and shadows, grass looked like it droped a texture setting.
What I saw didn't impress me, I'd rather have a good 30% than a mediocre 300%.

Hopefully the quality tier is still decent and what they presented was just to get the high multiplier for marketing
Click to expand...


tAxn9ZW.png


The shot on the right is using FSR performance mode so obviously going to look a bit blurrier than native. I personally never use performance mode anyway so am quite impressed that you can barely tell the difference. FSR Quality or Balanced will be much better.
 
fs123 said:
tAxn9ZW.png


The shot on the right is using FSR performance mode so obviously going to look a bit blurrier than native. I personally never use performance mode anyway so am quite impressed that you can barely tell the difference. FSR Quality or Balanced will be much better.
Click to expand...

These images tell us nothing, you have to do comparisons with video in motion not still images. FSR has always being good at looking stable in still images, but start turning the camera and it breaks down
 
Last edited:
humbug said:
Wait... how can the quality of a generated image be worse, its not like that aspect is scaling it.
Click to expand...
Doing this kind of interpolation at the speed required for there not to be massive input latency is really hard - Nvidia does it with dedicated hardware (the optical flow accelerator), motion vectors from the game engine and 'AI' - and even then depending on the content and base frame-rate, artifacts can be introduced as DLSS3 is making its 'best guess' at what the interpolated frame should be.

Imagine the second hand of a clock - on frame 1 it's at 12:00:00, at frame 2 it's at 12:00:02 - if you did a simple linear interpolation to generate a frame in-between those at 12:00:01 it wouldn't look correct - instead of rotating, the second hand would smear between the two positions. TVs that do this sort of thing have the luxury of delaying the video by a half second or more whilst they do their processing (and can consequently work across a great deal more frames) - for gaming that kind of post-processing simply doesn't work - it's much too slow.
 
Grim5 said:
These images tell us nothing, you have to do comparisons with video in motion not still images. FSR has always being good at looking stable in still images, but start turning the camera and it breaks down
Click to expand...

Those images tell us that the comparison is in FSR performance mode which is the worst for image quality whether still or moving. The latest version of FSR upscaling is by no means worse than DLSS in motion if you ever bothered to try it. Of course that may not suit your narrative :rolleyes:.
 
JediFragger said:
FSR3 is suspected to still be based on FSR2 but with the additional frame generation features. It's not good enough imo.
Click to expand...

Maybe not but I was just speaking hypothetically. If it does turn out to be as good then it doesn't really matter how it's achieved IMO.

Nexus18 said:
Lets wait and see how FSR 3 looks outside of the sponsored titles first before starting with the "dlss killer" hype again..... FSR 2 is still not a patch on dlss outside of a couple of hand picked games i.e. remember the claims of zero ghosting with deathloop? Only to end up with ghosting in most games where FSR is used (and more than dlss, which got lambasted for it's initial ghosting issues)

From the initial reports, it does sound to be of to a more promising start than FSR 1 and 2 though.

There is also this from reddit:



Hoping it is good as will be nice to have a "free" upgrade ;) :p Won't be surprised if I end up turning it off and never using it due to it lagging behind.
Click to expand...

Again no hype train, just speaking hypothetically with the little information we have at the moment.
 
SpudMaster said:
Maybe not but I was just speaking hypothetically. If it does turn out to be as good then it doesn't really matter how it's achieved IMO.
Click to expand...
Spending time tuning FSR2 and it looks good (ala Deathloop), problem is it seems most devs just want to implement for a tickbox and forget!!
 
Last edited:
  • Like
Reactions: TNA
Cyber-Mav said:
Looking forwards to see what sort of train wreck fsr3 turns out to be. Without dedicated hardware frame gen gonna be iffy.
Click to expand...
So far Digital Foundry, upon being shown FSR3 FG by AMD said it's as good as DLSS 3 FG to their eyes - no issues with image at all. Or, let me quote: "In terms of fluidity and clarity, FSR 3 looked a match for DLSS 3 - a view shared by Alex, Rich and John, who were all present to see the demos in person. A great start for FSR 3.".
 
Tinek said:
So far Digital Foundry, upon being shown FSR3 FG by AMD said it's as good as DLSS 3 FG to their eyes - no issues with image at all. Or, let me quote: "In terms of fluidity and clarity, FSR 3 looked a match for DLSS 3 - a view shared by Alex, Rich and John, who were all present to see the demos in person. A great start for FSR 3.".
Click to expand...

My comment to DF is show us the footage then, show us the comparisons
 
Last edited:
Frame Gen is more important than upscaling because you're just doubling frame rates at native resolution.

Most people wont give a hoot about the input latency unless they are playing fast pace shooters, it diminishes the advantage Nvidia have in part with DLSS, given also that if it works with games that don't even have it, if it just works on your entire game catalogue that puts a huge great torpedo hole in DLSS.
 
You must log in or register to reply here.
Back
Top Bottom