Testing Nvidia vs. AMD Image Quality

[Xarick]

Okay.. seriously .. as an owner of a 5850 I am a tad disappointed in that response by AMD. I would rather they say.. we are investigating the shimmering caused by our filtering optimizations on HQ and will release a fix soon.

 

[railven]

AMD responds to NVIDIA’s image-quality complaints

http://www.hexus.net/content/item.php?item=27717

Small article touching on this subject matter!

Obviously the real test comes down to user experience, and in our thoroughly unscientific testing we couldn't see any particular degradation in image quality when using the default Catalyst settings. There will obviously be differences in the companies' hardware and drivers that will render textures in different ways, but we doubt that the majority will be able to notice any difference in real world scenarios. And for those that do notice, the settings can always be changed manually.

That seems to echo some of the opinions of posters here. Glad to see level headed people still exist in this world.

 

[railven]

Okay.. seriously .. as an owner of a 5850 I am a tad disappointed in that response by AMD. I would rather they say.. we are investigating the shimmering caused by our filtering optimizations on HQ and will release a fix soon.

Shimmering exists on the nVidia side as well. GTX 460 default vs 5870 Cat 10.11 Default in my house look the same, shimmers here and there, and the Radeon has higher contrast from what I was seeing.

I had to investigate my own issues with Cat 10.11 and came across with both nVidia and ATI have really good hardware this generation. Choices are great peeps! Buy your favorite color 😀

 

[Attic]

That seems to echo some of the opinions of posters here. Glad to see level headed people still exist in this world.

Yes, I agree.

The reasons behind the OP should be clearly evident (Read: nVidia focus group member) I don't see a reason to fan the flames of this issue, blown up by the blogpost of an nvida employee, otherwise.

We have plenty of trusted review sites that have chosen to not belabor the IQ aspect of the cards, but rather their performance and features. It appears disingenious to make a stink about the IQ of the cards when the insiginficance of the issue is the only sturdy ground it can stand on.

 

[Xarick]

Interesting stuff here:
http://www.forum-3dcenter.org/vbulletin/showthread.php?t=495132

 

[SirPauly]

That seems to echo some of the opinions of posters here. Glad to see level headed people still exist in this world.

But not all opinions though. Tolerances are more subjective than IQ to me. With testing and examples to showcase differences is important and for sites to truly investigate. Web-sites differ because the individuals tolerances may differ or knowledge.

IQ usually has taken a back seat to performance numbers and blanket views based on a few performance metrics to generate awareness to sell product now. For me, personally applaud sites that really break things down and investigate quality -- showcase examples for their findings.

 

[Stoneburner]

I knew something was wrong when I couldn't spot the plastic surgery scars during my pr()n viewings.

 

[Dribble]

Yes, I agree.

The reasons behind the OP should be clearly evident (Read: nVidia focus group member) I don't see a reason to fan the flames of this issue, blown up by the blogpost of an nvida employee, otherwise.

Well that's not my opinion, and no one buys my graphics cards for me. For a performance test the image quality should be the same. If the developer specifies you use shader X in their game then both companies should use that shader, it is not ok for one manufacturer to replace that shader with a lower quality one.

Sure when you play the game you can lower IQ as much as you want, but for tests to work out which card is fastest they should both be using the same IQ otherwise the test is null and void.

 

[3DVagabond]

Well that's not my opinion, and no one buys my graphics cards for me. For a performance test the image quality should be the same. If the developer specifies you use shader X in their game then both companies should use that shader, it is not ok for one manufacturer to replace that shader with a lower quality one.

Sure when you play the game you can lower IQ as much as you want, but for tests to work out which card is fastest they should both be using the same IQ otherwise the test is null and void.

Game optimizations are done by both companies. What you want is the manufacturer's card(s) that you can turn them off in. AFAIK only AMD allows that. Makes it impossible to do the comparison you want though. It's certainly not fair to only turn them off for one manufacturer.


The point of the post you responded to is that this whole thing is nVidia marketing spin being perpetuated around the web by nVidia focus group members. There's nothing of substance here for anyone to be concerned about.

http://www.guru3d.com/article/radeon-hd-6850-6870-review/9
So the difference at default driver setting in-between AMD and NVIDIA is as far as we are concerned NIL.

 

[Keysplayr]

The point of the post you responded to is that this whole thing is nVidia marketing spin being perpetuated around the web by nVidia focus group members. There's nothing of substance here for anyone to be concerned about.

So it is of your opinion that there is nothing to see here, "Move along now." ??
Duly noted. But why are you of this opinion? If benchmarks are run without producing equal image quality, then as another poster said here, those benches are null and void. Meaningless. Problem is, the first run of benches are DONE. They are out of the gate already, and now try to get the review sites to re-run all of those benches under IQ scrutiny as those 4 German sites did, and are continuing to do.

 

[MrK6]

So it is of your opinion that there is nothing to see here, "Move along now." ??
Duly noted. But why are you of this opinion? If benchmarks are run without producing equal image quality, then as another poster said here, those benches are null and void. Meaningless. Problem is, the first run of benches are DONE. They are out of the gate already, and now try to get the review sites to re-run all of those benches under IQ scrutiny as those 4 German sites did, and are continuing to do.

And so far multiple other sites have confirmed that the image quality is equal, with both companies' drivers having optimizations here and there. There is no issue here besides sour grapes from the Green Team that once again AMD has shown them up.

 

[Outrage]

If benchmarks are run without producing equal image quality, then as another poster said here, those benches are null and void. Meaningless. Problem is, the first run of benches are DONE. They are out of the gate already, and now try to get the review sites to re-run all of those benches under IQ scrutiny as those 4 German sites did, and are continuing to do.

what games where used in reviews of 68xx that shows this af problem ?

 

[Keysplayr]

And so far multiple other sites have confirmed that the image quality is equal, with both companies' drivers having optimizations here and there. There is no issue here besides sour grapes from the Green Team that once again AMD has shown them up.

Excuse me, but how can two sites "confirm" two different results?
No I'm sorry, I need more than that man. More importantly, why write off the German sites, and embrace this new data without question? Or are you questioning the new data? Sorry, I haven't followed that closely.

 

[SirPauly]

The point of the post you responded to is that this whole thing is nVidia marketing spin being perpetuated around the web by nVidia focus group members. There's nothing of substance here for anyone to be concerned about.

Ya have to be kidding me. The mip-maps transition quality was basically broken on 5XXX series on higher noise textures with complex regular patterns and this is on nVidia? Focus groups? Countless gamers talked about this and there were just as many desiring to point fingers and blame on nVidia, bias, agenda, and it was truly broken on AMD's part.

Thankfully, there were gamers with keen eyes and web-sites with keen eyes instead of words and pointing fingers.

 

[Keysplayr]

Thankfully, there were gamers with keen eyes and web-sites with keen eyes instead of words and pointing fingers.

More to come, most likely. I'll bet they aren't letting this rest AFAIK and I hope it's contagious across all review sites. Now that the can has been opened, you can probably bet there will be many more tests with much higher levels of scrutiny than before.

 

[zebrax2]

I suggest for someone to run a poll of 5xxx/6xxx users if they are able see any difference at all between quality and high quality and if there is was the performance improvement worth it. I personally don't have any cards that would qualify.

 

[Skurge]

Excuse me, but how can two sites "confirm" two different results?
No I'm sorry, I need more than that man. More importantly, why write off the German sites, and embrace this new data without question? Or are you questioning the new data? Sorry, I haven't followed that closely.

They only found issues on 5yr old games.

Would you like me to start a new thread about how the gun texture looks worse on the gun in MW2? So nV are up to the same tricks as AMD. Yet you completely ignore that fact.

 

[3DVagabond]

Ya have to be kidding me. The mip-maps transition quality was basically broken on 5XXX series on higher noise textures with complex regular patterns and this is on nVidia? Focus groups? Countless gamers talked about this and there were just as many desiring to point fingers and blame on nVidia, bias, agenda, and it was truly broken on AMD's part.

Thankfully, there were gamers with keen eyes and web-sites with keen eyes instead of words and pointing fingers.

Straw man. The 5000 filtering issues is not what is being referenced in this thread.

 

[SirPauly]

Straw man. The 5000 filtering issues is not what is being referenced in this thread.

Of course it is.

 

[SirPauly]

They only found issues on 5yr old games.

Would you like me to start a new thread about how the gun texture looks worse on the gun in MW2? So nV are up to the same tricks as AMD. Yet you completely ignore that fact.

All is fair! If there are issues with nVidia, there will be threads offered and some web-sites may offer their findings for both sides. Where is the harm? This is a good thing that web-sites may investigate deeper for both players.

 

[Stoneburner]

So, is the argument that AMD has produced optimizations that don't cause image degradation in current games but do cause such degradation in older games?

8 pages and I couldn't find a single explanation as to why/how/where.

 

[3DVagabond]

So it is of your opinion that there is nothing to see here, "Move along now." ??
Duly noted. But why are you of this opinion? If benchmarks are run without producing equal image quality, then as another poster said here, those benches are null and void. Meaningless. Problem is, the first run of benches are DONE. They are out of the gate already, and now try to get the review sites to re-run all of those benches under IQ scrutiny as those 4 German sites did, and are continuing to do.

I'm not going to bother repeating everything said here. I posted a link to a respected and independent site, Guru3D, who disagrees with those German sites. They say IQ is the same. This is a test that was run after the supposed IQ differences were announced. If you can dispute Guru3D, please do so.

 

[Keysplayr]

I'm not going to bother repeating everything said here. I posted a link to a respected and independent site, Guru3D, who disagrees with those German sites. They say IQ is the same. This is a test that was run after the supposed IQ differences were announced. If you can dispute Guru3D, please do so.

I'm not. Computerbase, PCGH, 3DCenter.org, and TweakPC are. Don't be so quick to forget what a forum favorite Computerbase was ESPECIALLY when they used 8xAA in all their benches showing ATI in a pretty light when at the time, NV took a bigger performance hit at those AA levels. Boy oh boy did this forum love them then. Not now though? But why? 😉
They were loved SPECIFICALLY for testing in ways other review sites didn't, like using 8xAA when not a lot of other sites used more than 4x. It's that Computerbase went the extra mile that they were heralded one of the best review sites on the web by ATI bretheren. I can link to threads here from that time to prove it after a bit of searching. They are still here..

 

[Stoneburner]

I'm not. Computerbase, PCGH, 3DCenter.org, and TweakPC are. Don't be so quick to forget what a forum favorite Computerbase was ESPECIALLY when they used 8xAA in all their benches showing ATI in a pretty light when at the time, NV took a bigger performance hit at those AA levels. Boy oh boy did this forum love them then. Not now though? But why? 😉

You should go to law school 🙂

 

[3DVagabond]

I'm not. Computerbase, PCGH, 3DCenter.org, and TweakPC are. Don't be so quick to forget what a forum favorite Computerbase was ESPECIALLY when they used 8xAA in all their benches showing ATI in a pretty light when at the time, NV took a bigger performance hit at those AA levels. Boy oh boy did this forum love them then. Not now though? But why? 😉
They were loved SPECIFICALLY for testing in ways other review sites didn't, like using 8xAA when not a lot of other sites used more than 4x. It's that Computerbase went the extra mile that they were heralded one of the best review sites on the web by ATI bretheren. I can link to threads here from that time to prove it after a bit of searching. They are still here..

Guru3D is disputing those other sites findings, they are not disputing Guru3D, which is what I asked for. Their past like or dislike from "ATI brethren", as you label them, is not the subject. It's a straw man. No point in linking to past threads. They would be irrelevant to the discussion here.