Battlefield V DLSS Tested, The Biggest RTX Fail Of Them All

acroig said:
Actually in Metro it would serve a purpose if it wasn't so damn blurry, the FPS go up significantly, 15+ @ 1440p.
Click to expand...

Yeah but that's sort of the point. If it's blurry it doesn't actually serve a purpose. The fact the screen is blurred indicates it's rendering less data so it's no surprise it provides higher FPS.

Unfortunately, what the BFV testing shows is even if it "worked", you might be better off just upscaling from 1200P or 1080P.
 
The 3dmark Dlss offering is using similar resolutions yet doesn't suffer softness and seems to add detail. Why?
 
SIrPauly said:
The 3dmark Dlss offering is using similar resolutions yet doesn't suffer softness and seems to add detail. Why?
Click to expand...

Synthetic vs Real World.

DLSS probably won't work too well with a dynamic changing enviroment over a static Synthetic that is always the same.
 
So, it's not resolution but more improvements and maturity on Ai training and over-all Dlss technologies, one may imagine in real world.
 
SIrPauly said:
So, it's not resolution but more improvements and maturity on Ai training and over-all Dlss technologies, one may imagine in real world.
Click to expand...
Not its not improvement and maturity that causes the port royale bench to look better its because it is a canned never changing never deviating from the path benchmark that makes it allegedly look so good.

It's easy mode and shouldn't be used at all in any form to tout the IQ capabilities of DLSS. Unless they deviate from the path which would never happen because then it wouldn't be reliable as a synthetic benchmarking tool.
 
How did you just do that?

Imho,

Let me be more clear: Dlss needs improvements and maturity with Ai training and over-all Dlss technologies for real world examples. If the problem was just resolution based, then 3dMark would be soft as well.
 
SIrPauly said:
How did you just do that?

Imho,

Let me be more clear: Dlss needs improvements and maturity with Ai training and over-all Dlss technologies for real world examples. If the problem was just resolution based, then 3dMark would be soft as well.
Click to expand...
Language barrier or you're unintentionally gatekeeping info with your stilted and poorly worded replies.

The jist of it is port royale shouldn't even be in this conversation. It is not a real world depiction of what DLSS is or isnt capable of because Nvidia trained it on a static never changing easily optimized and never going off rails canned benchmark. It doesn't matter because with synthetic benchmarks the house always wins.
 
It does for me and desire real world to be as detailed and clear. Nvidia needs to go to work and make it happen.
 
SIrPauly said:
How did you just do that?

Imho,

Let me be more clear: Dlss needs improvements and maturity with Ai training and over-all Dlss technologies for real world examples. If the problem was just resolution based, then 3dMark would be soft as well.
Click to expand...

I agree and disagree. Again, Canned Benchmark vs an ever changing enviroment. The AI was just trained better and probably had an easier time learing the algorithim in port royal as opposed to an ever changing scene depending on weather, players etc etc in BFV.

As for the resolution its an easy guess as to what went wrong. Nvidia never ran the AI through Ultra Wide in that title.

Overtime DLSS will probably improve in BFV and this will probably be the same across the board for all upcoming titles that support the feature.
 
Seyiji said:
Language barrier or you're unintentionally gatekeeping info with your stilted and poorly worded replies.

The jist of it is port royale shouldn't even be in this conversation. It is not a real world depiction of what DLSS is or isnt capable of because Nvidia trained it on a static never changing easily optimized and never going off rails canned benchmark. It doesn't matter because with synthetic benchmarks the house always wins.
Click to expand...

This is what happens when Sey drops the internet lingo and types clearly.

WHO ARE YOU I DONT EVEN KNOW WHO YOU ARE ANYMORE?
 
logical said:
This is what happens when Sey drops the internet lingo and types clearly.

WHO ARE YOU I DONT EVEN KNOW WHO YOU ARE ANYMORE?
Click to expand...

CarelessPointlessGalapagosalbatross-size_restricted.gif

I's bertmens!
 
Hapatingjaky said:
I agree and disagree. Again, Canned Benchmark vs an ever changing enviroment. The AI was just trained better and probably had an easier time learing the algorithim in port royal as opposed to an ever changing scene depending on weather, players etc etc in BFV.

As for the resolution its an easy guess as to what went wrong. Nvidia never ran the AI through Ultra Wide in that title.

Overtime DLSS will probably improve in BFV and this will probably be the same across the board for all upcoming titles that support the feature.
Click to expand...

That would be my guess on ultra wide but Blaire offered me some close-ups with Battlefield V with traditional 4k and the softness really hurts the experience. Blaire contacted Nvidia to share the negative experience which is a good thing; awareness helps.

I'm still waiting for Blaire to share findings with Dlss and FreeStyle; to see if this helps.
 
DLSS may have very little to do with the clarity in 3DMark, or any game, but mostly more to do with Ray tracing. The reason i say this is Nvidia is planning on using adaptive ray tracing to remove bluring and artifacts in TAA (see link below) if true, it explains why DLSS looks clearer in 3DMark as the comparisons between TAA and DLSS, it is obviouse there is no ray tracing in the TAA examples per the video i linked a page or two ago.



https://www.dsogaming.com/news/nvid...ts-in-taa-with-adaptive-ray-tracing-in-games/
 
Last edited:
You must log in or register to reply here.
Back
Top