When enough is enough (AF quality on g70)

[HaLDoL]

The fact that you don't see shimmering is good, but it's also subjective on your part.

Seeing shimminering is also subjective, some see alot, some have never seen it. Some even see it on ATI cards (oh no, that must be subjective). Point is: to him it's objective, to you (of course) not.
But all I know is that an ATi employee cannot be an objective judge on this matter.

 

[Graphics_Krazy]

I am curious too why the big name sites that PLAY / FRAPS their benchmarks didn't notice that the shimmering was WORSE than even a 6800. Its really annoying in HL2 and BF2 - you cant miss it.

They have some funky LOD switching going on quite early.

 

[CJ]

But all I know is that an ATi employee cannot be an objective judge on this matter.

Maybe it's hard for you to understand, but some people can actually be objective even when it comes to competitors. It's called professionalism.

 

[Sxotty]

Maybe it's hard for you to understand, but some people can actually be objective even when it comes to competitors. It's called professionalism.

Maybe it is hard for you to understand, but trusting someone you do not personally know to be objective about their competitor is not wise.
It's called "naive"

 

[CJ]

Maybe it's too hard for you to understand, but... who was talking about trust? I was just commenting on the generalisation made by HaLDoL and just pointing out that there are actually some people who can be objective about their competitor. That doesn't say anything about you - or anyone else - having to trust them which is a completely different matter.

 

[andypski]

Seeing shimminering is also subjective, some see alot, some have never seen it. Some even see it on ATI cards (oh no, that must be subjective). Point is: to him it's objective, to you (of course) not.
But all I know is that an ATi employee cannot be an objective judge on this matter.

I have never made any secret of the fact that I am an ATI employee - that way people can make objective unbiased judgements about my subjective biased opinions.

Nevertheless it is perfectly possible to be objective about the contents of these videos no matter who you work for. We can all look at them and say "A shimmers more than B", or "B and C shimmer about the same amount" without resorting to overly subjective evaluation or bias.

Even if you get to that point there are still many questions to answer before the data can be evaluated - were the videos captured appropriately? Were the appropriate settings used in the driver and application? Were any mistakes made in the collection of the data?

This is the main reason that lossless compression is vital in these cases - anything else would introduce uncertainty.

Making a statement like "I don't see shimmering in my games", however, is purely subjective. Nobody else can see through your eyes at the time, there is no evidence that they can look at the same way that _you_ look at it, and therefore no underlying basis for anyone else to evaluate your data or experience. As such it is purely subjective, and can't really contribute anything to the question at hand because although it may be a true statement no-one else can verify it.

 

[andypski]

Maybe it is hard for you to understand, but trusting someone you do not personally know to be objective about their competitor is not wise.
It's called "naive"

I would certainly agree that to take my statements at face value without question when I'm obviously an interested party would be a mistake.

You should always evaluate and say "Yes, that's a good point" or "No - he's talking rubbish and spreading FUD"

But why is that really any different from how you should treat what anyone else posts?

 

[BenSkywalker]

Do people here grasp that the 'shimmering' textures, however horrible they appear to us, are regularly held up as an example of superior graphics by a great deal of laymen and enthusiasts in general? Anand utilized shimmering textures as an example of ATi's filtering supremacy not all that long ago(pointing this out as expecting reviewers to catch nuances they are clueless about is setting yourself up for a let down).

Personally I find it a sizeable let down that all the viable parts as of now have enormously degraded filtering quality compared to a last generation console(hardware wise at least). Hopefully a new player will enter the market with a bit of an eye towards image quality and get us out of the rut that ATi and nVidia have us in.

 

[sonix666]

What consoles match PC specs currently? Except for the new generation of consoles there is no way you can compare a console to a PC. Much lower resolutions and lower resolution textures because of memory constraints.

 

[Ailuros]

I would certainly agree that to take my statements at face value without question when I'm obviously an interested party would be a mistake.

You should always evaluate and say "Yes, that's a good point" or "No - he's talking rubbish and spreading FUD"

But why is that really any different from how you should treat what anyone else posts?

I was actually going to say that; I take comments at face value and try to think first if there's any merit to them or not.

I couldn't care less what the messenger's credentials state; to a certain degree there's always a chance that I'll end reading rubbish from an "expert" in a specific field, as there is a chance that I might see something from a layman that actually makes sense.

 

[Ailuros]

Do people here grasp that the 'shimmering' textures, however horrible they appear to us, are regularly held up as an example of superior graphics by a great deal of laymen and enthusiasts in general? Anand utilized shimmering textures as an example of ATi's filtering supremacy not all that long ago(pointing this out as expecting reviewers to catch nuances they are clueless about is setting yourself up for a let down).

Personally I find it a sizeable let down that all the viable parts as of now have enormously degraded filtering quality compared to a last generation console(hardware wise at least). Hopefully a new player will enter the market with a bit of an eye towards image quality and get us out of the rut that ATi and nVidia have us in.

Haven't seen you in a long while Ben here; shall I say welcome back?

Agreed.

 

[Razor1]

I have never made any secret of the fact that I am an ATI employee - that way people can make objective unbiased judgements about my subjective biased opinions.

Nevertheless it is perfectly possible to be objective about the contents of these videos no matter who you work for. We can all look at them and say "A shimmers more than B", or "B and C shimmer about the same amount" without resorting to overly subjective evaluation or bias.

Even if you get to that point there are still many questions to answer before the data can be evaluated - were the videos captured appropriately? Were the appropriate settings used in the driver and application? Were any mistakes made in the collection of the data?

This is the main reason that lossless compression is vital in these cases - anything else would introduce uncertainty.


Making a statement like "I don't see shimmering in my games", however, is purely subjective. Nobody else can see through your eyes at the time, there is no evidence that they can look at the same way that _you_ look at it, and therefore no underlying basis for anyone else to evaluate your data or experience. As such it is purely subjective, and can't really contribute anything to the question at hand because although it may be a true statement no-one else can verify it.

Well also so far people who have been having shimmering I can see them in thier videos. But I can't reproduce that type of shimmering in any game levels. So it might be that the shimmering increased depending on diffferent types of shaders or lack their of, or different types of textures as used in the 3dcenter movies. Saying its subjective is not really right when I do believe the videos aren't faked and the shimmering is truely there. But replicating the problem in different situations will prove the fact that its a general problem with AF algos, or only in certain circumstances.

 

[wireframe]

Tridam (or anyone else involved in the production of the videos),

If I understood correctly, these videos were produced in UT2003, presumably with some custom built map. Would it be possible for you to share the actual map so those of us with UT2003 can run it natively on our computers?

 

[andypski]

Saying its subjective is not really right when I do believe the videos aren't faked and the shimmering is truely there. But replicating the problem in different situations will prove the fact that its a general problem with AF algos, or only in certain circumstances.

Please understand - I wasn't trying to suggest that what you were saying was necessarily incorrect or untrue in any way - I just regard it as subjective simply because at the moment it's impossible for another independent observer to take your statement and verify whether they do or don't see shimmering in the same circumstances. Different people have different levels of sensitivity to shimmer, and hence the subjective aspects creep in.

If, on the other hand, you have a case such as we have demonstrated here where seemingly more or less everyone can agree that there is some shimmer then it can be analysed objectively. Why is there shimmer? How bad is it relatively speaking? Can anything be done to 'fix' or improve it?

I would certainly expect any shimmering that is present in applications to vary significantly in visibility depending on the texture content and shaders being used, so you might well only see it in specific places.

 

[Razor1]

I understand, I'm trying to take vidoes of UT 2004 in flyby of Deck 17, but the recorder I'm using, Camtasia studio 2, seems to have a high cpu usage and its lagging badly, if I can get it working then I'll do other levels too.

Tridam, what settings did you use and how did you apply them? Through the CP or in game?

 

[caboosemoose]

The shimmering texture issue is a tricky one. There is currently a clear difference between ATI and NVIDIA at present but NVIDIA are handling it pretty cannily. They have it set just below the level at which most reviewers either A. notice it or B. judge it to be bad enough to merit a ticking off. I think one issue that critics of those who publish reviews should note is that reviewers often do not have long with the boards before publishing reviews. Given that they have to get through a number of benchmwark tests (with both the new card and existing cards) I think it's harder than you might imagine to spend time carefully analysing IQ. That sort of analysis often comes later, typically when the contest has pretty much already been decided.

It's interesting that NVIDIA consider this level of IQ acceptable even when they have a decent hardware advantage.

Incidentally, I have no problem recording the shimmering effect in HL2 using FRAPS for those who are looking for an easy way to see or record it.

 

[Sxotty]

Andy I wasn't saying you were a devious bastard btw, just that taking anyone's word you don't know is silly, and moreso if they do have a vested interest in pointing out behavior that shows their interest in a better light.

On another note, I am happy about whoever linked to Humus' tweak to enable filtering on alpha textures. Ever since SSAA went away I was really annoyed that ATI could not filter the edges of transparent textures, that is one of the reasons I sold my 800PE to be honest. (plus needing money to go to Australia).

In any case I will let youguys hash all this out and worry about it the next time I have vast sums to waste on a video card.

 

[Hyp-X]

It's interesting that NVIDIA consider this level of IQ acceptable even when they have a decent hardware advantage.

Well they have a serious hardware disadvantage when doing anisotropic filtering - hence the agressive "optimizations".

 

[Jawed]

Well they have a serious hardware disadvantage when doing anisotropic filtering - hence the agressive "optimizations".

Ooh, that sounds interesting, what is it?

Jawed

 

[tEd]

The shimmering texture issue is a tricky one. There is currently a clear difference between ATI and NVIDIA at present but NVIDIA are handling it pretty cannily. They have it set just below the level at which most reviewers either A. notice it or B. judge it to be bad enough to merit a ticking off. I think one issue that critics of those who publish reviews should note is that reviewers often do not have long with the boards before publishing reviews. Given that they have to get through a number of benchmwark tests (with both the new card and existing cards) I think it's harder than you might imagine to spend time carefully analysing IQ. That sort of analysis often comes later, typically when the contest has pretty much already been decided.

It's interesting that NVIDIA consider this level of IQ acceptable even when they have a decent hardware advantage.

Incidentally, I have no problem recording the shimmering effect in HL2 using FRAPS for those who are looking for an easy way to see or record it.

Actually with quality setting which the reviewers benched there is no doubt that it shimmers everything else is simply a lie. On High Quality it is probably very depnending games and situation.