NVIDIA Says AMD Reduced Image Quality Settings HD 6800 Series For Better Performance

[BababooeyHTJ]

I think an article showing the performance and quality difference would be a nice start, and quite necessary to form an opinion in this case.

omg why does this matter so much? this thread has gone on for six pages...about quality settings?

You can't make a judgment based on benchmarks alone.

I can honestly say that I'm noticing less texture shimmering on a GTX 580 with texture quality set to high quality compared to my previous HD 5870 with Catalyst AI set to high quality (10.10e hotfix ) I'd still like to see how the 5800 compare to the 6800s when they both have acess to these newer Catalyst AI options.

Did they check the texture sharpness in the comparison? When I follow the test of rage3d regarding 6800 series and AF, I notice that shimmering disappear (on my 5870) when I increase the lod bias to a positive value.

If you bump LoD to +0.65 the shimmer disappears. On NV cards nudge LoD to -0.65 and it appears.

I haven't tried this myself but here is the link to this discussion.

I also saw this posted today.

Nice news:

http://www.forum-3dcenter.org/vbulle...&postcount=930


http://www.forum-3dcenter.org/vbulle...&postcount=940

[Solus Corvus]

But the question remains: should reviewers use the HQ setting in their articles?

I won't venture my opinion just yet since I want to hear what you guys have to say.

I have always enabled the best quality options on either brand. Any reviewers that test high quality performance, even just for a section of the review, and do a quality comparison would surely receive more traffic from me.

[cegras]

But the question remains: should reviewers use the HQ setting in their articles?

I won't venture my opinion just yet since I want to hear what you guys have to say.

I say default, but with disclaimer.

[MpG]

omg why does this matter so much? this thread has gone on for six pages...about quality settings?

And what else is the point of looking for a better video card? If we didn't care about image quality, we'd still be on ancient hardware. But in the end, the whole reason we even care about benchmarks is because we want to know "how smooth, and how good-looking can I make my games?)

[Olivon]

But the question remains: should reviewers use the HQ setting in their articles?

I won't venture my opinion just yet since I want to hear what you guys have to say.

Yes,

Here in France some are doing reviews with HQ mode and I think it's a good move.

AMD lowering default quality settings is really a bad thing

[japamd]

But the question remains: should reviewers use the HQ setting in their articles?

I won't venture my opinion just yet since I want to hear what you guys have to say.

Yes, reviewers should use HQ settings, but with transparent anti-aliasing off.

[Shadov]

Yes,

Here in France some are doing reviews with HQ mode and I think it's a good move.

AMD lowering default quality settings is really a bad thing

Have you seen any visual difference in games that you play between ATI and Nvidia?

Can you control AF quality on Nvidia cards?

Are you sure the AF quality is the same on Nvidia and ATI?

Do you know that AF HQ quality on AMD disables ANY driver optimization made (even the ones without visual difference)?

That way reviewers should use comperable settings on both vendors and not crank up the settings on one brand (Radeon) that allows control of AF filters, while the other can get away with optimizations enabled (GeForce).

[zalbard]

Are you sure the AF quality is the same on Nvidia and ATI?

Do you know that AF HQ quality on AMD disables ANY driver optimization made (even the ones without visual difference)?

That way reviewers should use comperable settings on both vendors and not crank up the settings on one brand (Radeon) that allows control of AF filters, while the other can get away with optimizations enabled (GeForce).

This is why we need some testing done. With performance numbers and *.bmp screenshots.

[Olivon]

Have you seen any visual difference in games that you play between ATI and Nvidia?

Can you control AF quality on Nvidia cards?

Are you sure the AF quality is the same on Nvidia and ATI?

Do you know that AF HQ quality on AMD disables ANY driver optimization made (even the ones without visual difference)?

That way reviewers should use comperable settings on both vendors and not crank up the settings on one brand (Radeon) that allows control of AF filters, while the other can get away with optimizations enabled (GeForce).

Just read germans articles Shadov ... AMD made a mistake, nothing more to say

[Shadov]

Just read germans articles Shadov ... AMD made a mistake, nothing more to say

I wouldnt base my opinion on one article where one editor has skipped other recent games which would prove otherwise.

Believe me both AMD and Nvidia have optimizations in their drivers and it mostly depends on what games and scenes you choose to show.

But now we have a situation where AMD enables users the control of the filter, while Nvidia can optimize and tweak their drivers till someone screams again bloody murder. Of course that is if they (key word coming) *notice*.



Edit: To make it more clear, reviewers should ask for a clear statement which AMD vs NV modes to use while comparing performance or in the near future AMD could disable AF quality filtering settings in their drivers to have the same level of control over applications (and performance) as Nvidia, which is currently hidden away from the end user.

[Final8ty]

http://www.guru3d.com/article/explor...optimizations/
By: Hilbert Hagedoorn | Edited by Editor | Published: December 3, 2010

[cold2010]

Exploring ATI Image Quality Optimizations

wt this ? Exploring , show one game image Quality sorry Hilbert But

[zalbard]

http://www.guru3d.com/article/explor...optimizations/
By: Hilbert Hagedoorn | Edited by Editor | Published: December 3, 2010

OK, interesting article.
I've been staring at these ME2 screenshots to no avail.
My verdict is:
This optimisation is not a big deal... 5% performance traded for an almost-impossible-to-see difference in picture quality (at least in actual gaming scenarios). So I may as well leave this option on for a tiny bit of extra FPS, since I will never notice the difference anyway (as long as I do not stay still and stare at the screen for ages instead of actually playing the game).
On a side note, Nvidia's implementation still looks better... The gradient transaction is very smooth, while for AMD it goes from sharp to blurry with quite a noticeable edge.

[Lanek]

No no the only exemple he shown, is in ME there's no difference between quality ..

Nvidia on left, and AMD + Catalyst with optimisation on the right .... (set on quality not HQ ) ( AA is not applied the same can be disturbing but we speak about AF optimisation, not AA )

Global scene ..


Then i have zoom to the back and on specific zone.. ( for see if the AF optimisation was made far of the first scene ) ..




So what the proof in his article ? Trackmania screen from 3D center ?

[Vardant]

It's more like 10% across the whole board and the question isn't if people should leave it on, but what should the reviewers do.

[zalbard]

Stick to defaults? Which is Quality setting. A lot of people will use the default setting, plus it is nearly impossible to spot the difference in image quality...

[-Boris-]

So, the difference in quality isn't noticable, for a performance gain that isn't noticable. They get a few reviews showing better numbers while they get some fuzz about lower quality from others.

In the end it all seems so unnecessary. There is really no gain for AMD, it's all just bad publicity in turn for better numbers which people don't trust.

[Oztopher]

It would also be unfair to have AMD use the high quality in reviews as this disables optimisations all together, and as Shadov said, Nvidia will still be able to use their optimisations.

I lose quite a bit of fps in some games by disabling catalyst AI on my 4890, but who would ever disable catalyst AI? Even reviewers wouldn't. This is the equivalent of what is happening here. But i guess the only thing is, that AMD have added extra optimisations to the 68xx compared to the 4xxx and 5xxx series. Which is another unfair part.

There should be an open standard of compulsory (undetectable but fps gaining) optimisations

edit: Wait, how would i go about adjusting lod bias for my card? I want standard+ quality, i don't care for the pathetic % of fps loss.

[Lanek]

So, the difference in quality isn't noticable, for a performance gain that isn't noticable. They get a few reviews showing better numbers while they get some fuzz about lower quality from others.

In the end it all seems so unnecessary. There is really no gain for AMD, it's all just bad publicity in turn for better numbers which people don't trust.

It is in some game, but at same time the article of 3Dcenter mislead a lot, using old games as Trackmania ( a game they use for showe they was allready a problem with the algorythm of ATI till 3 years ), mixing the FP16 demotion who is only able in some game ( DOW, Oblivion,) .. and put all of this together for use it as proof AMD have decrease greatly the image quality on last driver, is a bit strange.. why never use BC2, Dirt2.. well the games who are tested now in review..

Same goes for the fps test, you need use an old driver 10.9 set standard, and then use the last driver set it to quality, and see how much is the difference, if you just test moving the slide on the last driver, you can't claim AMD have change it ...

If you disable all optimisation in the Nvidia driver and you bench a game, you will get the same performance lost..

[Lanek]

Forum lag double post.

[SKYMTL]

http://www.guru3d.com/article/explor...optimizations/
By: Hilbert Hagedoorn | Edited by Editor | Published: December 3, 2010

I don't understand.

They showed images from Mass Effect 2 / 3DCenter test and then ran benchmarks in Dirt and Far Cry 2? What about all the other games they use?

I mean, of COURSE going from the default setting to HQ will impact performance...

[AMDDeathstar]

But the question remains: should reviewers use the HQ setting in their articles?

I won't venture my opinion just yet since I want to hear what you guys have to say.

This has been going on for about a year now and for the longest time I couldn't see much difference in quality.

That was until I got the idea instead of use of using my main monitor an Acer 22" I tried my secondary monitor a Samsung 22" then I could start seeing some of the differences.

For me gaming on my Acer I probably wouldn't notice, unless something was totally FUBAR.

What I'm saying is the quality differences have to show up across different monitors, what you see on a 30" Dell I might not see on my old 22" Acer.

[Sn0wm@n]

Just read germans articles Shadov ... AMD made a mistake, nothing more to say

and what is this mistake ????

[STEvil]

This has been going on for about a year now and for the longest time I couldn't see much difference in quality.

That was until I got the idea instead of use of using my main monitor an Acer 22" I tried my secondary monitor a Samsung 22" then I could start seeing some of the differences.

For me gaming on my Acer I probably wouldn't notice, unless something was totally FUBAR.

What I'm saying is the quality differences have to show up across different monitors, what you see on a 30" Dell I might not see on my old 22" Acer.

Is one S-IPS and the other TN?

[generics_user]

@SKYMTL:

once this settings becomes available on the 5xxx series you have to do a comparison between the old driver with catalys ai on and th enew ones with defaults + HQ

i bet that HQ is going to perform worse than the old AI on and default is going to perform better than AI on