There is a lot of controversy surrounding the image quality of Nvidia versus ATI and I would like to request Hardocp to do an analysis of this. Other sites have tried this but I have a feeling that Hardocp would do a more thorough and accurate study.
![[H]ard|Forum](/styles/hardforum/xenforo/logo_dark.png){kind=logo}
Request for Hardocp to do image quality article
- Thread starter biggles
- Start date
The controversy isn't so much in the quality. In HQ mode, both ATI and nVidia are roughly equal. Each 'side' wins some, and loses some...but they roughly wash out.
The "controversy" comes from sites that benchmark supposedly equal settings (equal resolution and equal FSAA/Aniso multipliers), use the "out of the box" optimizations, just toss up the FPS given, and call it 'fair'.
Which isn't.
As "out of the box", ATI is running in HQ mode, and "out of the box", nVidia isn't. nVidia's "Quality" mode (vs "High Quality") results in up to a 10% performance boost in some cases, giving their cards quite an unfair advantage in reviews that are not scrupulous about this point.
This isn't really a problem here at [H], though, because of how radically different the reviews are done. Instead of using 'equal settings' (that aren't) as the baseline and just posting the differences in framerate, [H] picks a specific framerate as the baseline, and tinkers with the settings to get the highest quality that can keep that number.
The result is much, MUCH harder to understand...harder to compare a large number of cards with...and harder to argue with. It is, in reality, what we all do with the cards and games. We have a specific framerate we want for it to be "playable", and we tinker with the settings until we get the highest quality we can get that keeps it. [H] just posts those findings in their reviews.
The "controversy" comes from sites that benchmark supposedly equal settings (equal resolution and equal FSAA/Aniso multipliers), use the "out of the box" optimizations, just toss up the FPS given, and call it 'fair'.
Which isn't.
As "out of the box", ATI is running in HQ mode, and "out of the box", nVidia isn't. nVidia's "Quality" mode (vs "High Quality") results in up to a 10% performance boost in some cases, giving their cards quite an unfair advantage in reviews that are not scrupulous about this point.
This isn't really a problem here at [H], though, because of how radically different the reviews are done. Instead of using 'equal settings' (that aren't) as the baseline and just posting the differences in framerate, [H] picks a specific framerate as the baseline, and tinkers with the settings to get the highest quality that can keep that number.
The result is much, MUCH harder to understand...harder to compare a large number of cards with...and harder to argue with. It is, in reality, what we all do with the cards and games. We have a specific framerate we want for it to be "playable", and we tinker with the settings until we get the highest quality we can get that keeps it. [H] just posts those findings in their reviews.
Brent_Justice
Moderator
- Joined
- Apr 17, 2000
- Messages
- 17,755
I have been thinking about doing an all out image quality comparison article. Right now though we have a ton of things on our plates, plus some things that are upcoming.
What I will do though is include more relevant comparison screenshots in evaluations. In some cases you may find two cards being playable at the same settings so you can compare them "apples to apples" in that way.
For example, in an upcoming evaluation we have an NV card that achieved 4X TR SSAA and an ATI card that achieved 4X Quality ADAA in HL2 EP 1. So in that game I'll have a screenshot comparing a fence side-by-side with those settings so you can look at it and see which one looks better to you.
What I will do though is include more relevant comparison screenshots in evaluations. In some cases you may find two cards being playable at the same settings so you can compare them "apples to apples" in that way.
For example, in an upcoming evaluation we have an NV card that achieved 4X TR SSAA and an ATI card that achieved 4X Quality ADAA in HL2 EP 1. So in that game I'll have a screenshot comparing a fence side-by-side with those settings so you can look at it and see which one looks better to you.
Sounds good. May I suggest video clips from games as well in order to address the issue of shimmering that keeps coming up in threads. Though I have never noticed shimmering, I hear this is only apparent in motion and not in static screenshots.Brent_Justice said:
JBavousett
Gawd
- Joined
- Aug 7, 2006
- Messages
- 684
NVidia is better because the logo is prettier and the box graphics on ATI cards scare me. Case closed.
I agree with the above ideas, all to often we get a set of screenshots taken of the same or similar areas in an arbitrary game, and we get no idea of the motion quality. An Intel 950 can render a really pretty picture once a second, how well does it render 100 per second? That's the real issue.
I agree with the above ideas, all to often we get a set of screenshots taken of the same or similar areas in an arbitrary game, and we get no idea of the motion quality. An Intel 950 can render a really pretty picture once a second, how well does it render 100 per second? That's the real issue.
jebo_4jc said:
It's been done in other sites.
http://www.firingsquad.com/hardware/ati_nvidia_image_quality_showdown_august06/
Though I always come to [H] first, I never just stick with [H] reviews / evaluations. I read what I can and take my own conclusions, based on what I think looks best. IQ comparisons (apart from the obvious stuff) is very subjective.
Majestic12
Weaksauce
- Joined
- Aug 28, 2006
- Messages
- 95
From a few months ago...
Link
Would be great to see an update, yeah.
Also, as much as we all dream about $800 SLI/xfire setups, I'm guessing the bulk of readers have a $200-300 graphics card. If there is a IQ article, it would be nice if they were done with single cards at reasonable (ie 16x12) resolutions.
Link
Would be great to see an update, yeah.
Also, as much as we all dream about $800 SLI/xfire setups, I'm guessing the bulk of readers have a $200-300 graphics card. If there is a IQ article, it would be nice if they were done with single cards at reasonable (ie 16x12) resolutions.
roflcopter
Limp Gawd
- Joined
- Aug 25, 2005
- Messages
- 246
Talking about open doors. Ofcourse Ati's IQ is better, but it seems to me that lately it's all about "the reds" rubbing it in on "the greens".
legendz411
2[H]4U
- Joined
- Jul 3, 2006
- Messages
- 3,534
I.E
Fan-boying
I personally think nVidia cards look FINE. I run games at 4xAA/8xAF at 10x7 and I could probly run the games I play higher or with better settings. I don't becuase I liek what I see. I don't think much higher settings would help regardless. I may try 12x10 but I think it would be too large for me to see the small things. No need for me to strain my young eyes.
I've used my dad's ATI 9600 and My FX 5200, (Arguably the lesser of the two) and my FX5200, despite its shortcomings, looked might fine at lower Resolutions and such.
IQ is subjective. There is no accurate way to record it. What looks good to the editors may not look as good to me as my set up.
That being said, I fight with my pops alot cause he is a ATI fan boy >_>
Fan-boying
I personally think nVidia cards look FINE. I run games at 4xAA/8xAF at 10x7 and I could probly run the games I play higher or with better settings. I don't becuase I liek what I see. I don't think much higher settings would help regardless. I may try 12x10 but I think it would be too large for me to see the small things. No need for me to strain my young eyes.
I've used my dad's ATI 9600 and My FX 5200, (Arguably the lesser of the two) and my FX5200, despite its shortcomings, looked might fine at lower Resolutions and such.
IQ is subjective. There is no accurate way to record it. What looks good to the editors may not look as good to me as my set up.
That being said, I fight with my pops alot cause he is a ATI fan boy >_>
First of all, Nvidia needs to remove Quality from their drivers completely. It's utter shit and it looks like utter shit. Force LOD clamp - remove the option from the drivers and just do it already like ATI does. Nvidia needs to address these items in their drivers. Too many reviews are done with the default Nvidia selection of Quality - yes, there is a difference and yes it's noticable. Once Nvidia does this I feel reviews will be much more indicitive of a true performance difference from NV to ATI. Let's even the playing field before we debate performance. Out of the box ATI does have IQ, I don't really think anyone can argue this.
Anisotrophic Filtering:
ATI's HQ AF pushes it further than Nvidias AF - so with HQ AF on there is no comparison. AF on ATI wins.
Anti-Aliasing:
AA can be more subjective especially when comparing SSAA to Adaptive. Nvidia's transparency AA seems to be better however Nvidias AA has always shown more of a performance hit than ATI's. Just seems to be the way the cards are designed. With just standard AA, ATI can run 6x AA which is obviously superior to Nvidias 4x AA, talking single cards of course here. I really miss transparency AA of the NV card, but 6x AA on the ATi seems to make up for it to me anyways. Without transparency AA ATI wins, with transparency AA Nvidia would win. ATI and Nvidia need to remove this as an option from the drives and just use transparency AA all the time - they should focus on performance with it on. Opengl doesnt support trans. aa due to the way the alpha channels are handled in the API.
Now we come to the last item which is the most talked about - shimmering. I've run Nvidia and ATI hardware over the last few generations of each and I will say that I can see shimmering, and it drives me nuts. Some people seem to be less sensitive to it, Im not one of those people. It immediately draws my attention to it and it annoys me to say the least. Certain games exhibit shimmering on ATI and Nvidia hardware - Nvidia is more prevalent. In other cases there is shimmering on NV hardware where ATI does NOT exhibit it in the least. In some cases it's game dependent in other cases it's just Nvidias' drivers. Using LOD clamp doesn't help nor does HQ, I can still see its just less prevelant. To someone sensitive to this if it's there, its very annoying and does in fact affect the gaming experience.
Graphically the AF and AA are very comparable with both companies trading blows depending on the situation.
We need to call for Nvidia to remove the Quality setting from their drivers and lock LOD clamp - we don't need these as options we need this to even up with ATI. Both companies need to enable tranparency Anti-Aliasing and focus on performance with it enabled. If we can get these changes made I think we will be on our way to actually a performance comparison worth reading.
Anisotrophic Filtering:
ATI's HQ AF pushes it further than Nvidias AF - so with HQ AF on there is no comparison. AF on ATI wins.
Anti-Aliasing:
AA can be more subjective especially when comparing SSAA to Adaptive. Nvidia's transparency AA seems to be better however Nvidias AA has always shown more of a performance hit than ATI's. Just seems to be the way the cards are designed. With just standard AA, ATI can run 6x AA which is obviously superior to Nvidias 4x AA, talking single cards of course here. I really miss transparency AA of the NV card, but 6x AA on the ATi seems to make up for it to me anyways. Without transparency AA ATI wins, with transparency AA Nvidia would win. ATI and Nvidia need to remove this as an option from the drives and just use transparency AA all the time - they should focus on performance with it on. Opengl doesnt support trans. aa due to the way the alpha channels are handled in the API.
Now we come to the last item which is the most talked about - shimmering. I've run Nvidia and ATI hardware over the last few generations of each and I will say that I can see shimmering, and it drives me nuts. Some people seem to be less sensitive to it, Im not one of those people. It immediately draws my attention to it and it annoys me to say the least. Certain games exhibit shimmering on ATI and Nvidia hardware - Nvidia is more prevalent. In other cases there is shimmering on NV hardware where ATI does NOT exhibit it in the least. In some cases it's game dependent in other cases it's just Nvidias' drivers. Using LOD clamp doesn't help nor does HQ, I can still see its just less prevelant. To someone sensitive to this if it's there, its very annoying and does in fact affect the gaming experience.
Graphically the AF and AA are very comparable with both companies trading blows depending on the situation.
We need to call for Nvidia to remove the Quality setting from their drivers and lock LOD clamp - we don't need these as options we need this to even up with ATI. Both companies need to enable tranparency Anti-Aliasing and focus on performance with it enabled. If we can get these changes made I think we will be on our way to actually a performance comparison worth reading.
chinesepiratefood
2[H]4U
- Joined
- Oct 8, 2005
- Messages
- 2,629
Brent_Justice said:
mail me a few video cards and ill write it up for ya