Trilinear Filtering Comparison(R420 vs R360 vs NV40 vs NV38)

[jvd]

Quote:

Coloured mipmaps naturally show full trilinear - statement

our image quality analysis reveals that there could be visible differences in the image. - issue

It should be noted that trilinear was originally invented to smooth transitions between mip-levels, and in the original definition mip-levels should be filtered versions of each other, as coloured mip-levels clearly are not. - statement

Despite this, we understand that people often make use of hardware for purposes beyond that originally envisioned, so we try to make sure that everything always functions exactly as expected. - issue

:?

whats your point . There could be visable diffrences with other video cards. Yet all other video cards have visual diffrences with the refrence rastner . IF there is a % diffrence allowed shouldn't that extend to this ?

Not only that but they say could. They also ask in thier release that people report any issues .



How is the other one an issue ? THey are making sure that everything always functions as expected. To me that is not an issue that is a good thing.

I like it when things work as expected. You know like when my car starts. When my shower works. All that is good and is working as expected.

anonymouscoward: Image quality is a relative term. The real question is, "does the claimed 'trilinear filtering' produce a byte-for-byte replica of 'true trilinear filtering'?" Whether or not the image quality is "the same" or "essentially" the same is irrelevant to this questions

Andy/Raja: Byte for byte compared to what? "True trilinear" is an approximation of what would be the correct filtering, a blending between two versions, one which is too blurry and one too sharp. An improved filter is not byte for byte identical to anything other than itself, but that doesn't mean it isn't a better approximation.

There are "apparently" visible between their own products, let alone getting another IHV involved.

You're not gonna have filtering that appeals to everyone. Some will spot the usefullness in other areas and so on.

Function as expected... according to what? Reviewers, ananlyst and enthusiast were all expecting full Trilinear filtering. ATi hadn't mentioned that they may not be producing it at all times.

You are celebrating their enthusiasm about working with people who discover the problems, yet you are not happy with hearing the problems being discovered or investigated. It is going to help them resolve the issue eventually, though how is anyone suppose to react when they see something that wasn't even discussed at launch before?

It's like asking someone to identify a problem with a Armegus. You gotta know WTF a Armegus is and what it is supposed to do before you can report a problem with it doing what it was ment to do.

I'm not complaining about anyone bringing issues forth. Its the lack of legit issues and the amount of witch hunting i'm complaining about.

So far I have yet to see any places where this is a problem.

 

[Drak]

LOL. For real though, by Aaron's reasoning if Nvidia never improves IQ neither should ATI. That simply is false by all standards.

Exactly, we're buying ATI not NVIDIA. And paying good money for it too. What does what we want have anything to do with NVIDIA?

If ATI is only going to say "We're going to give you better than NVIDIA", not "We're going to give you the best we can for your money", then we desperately need a third company.

And it's not as if giving us the option is going to be soooo difficult to implement. According to Dave's understanding, the test to enable the optimisation is done in software in the driver. The driver is already spending CPU cycles to "analyse" the texture. I'm sure it can spare a miserable "if statement" to set the optimisation flag to off for that texture. Geee, why are we even discussing this?

 

[Quitch]

ARGHHHH! FULL TRILINEAR IS NOT THE BEST QUALITY AVAILABLE!

Where are you from? Trilinear is a fudge, and using it all the time is an even bigger fudge. This adaptive algorithm so far seems to be the best of the IQ and FPS worlds, yet people keep complaining and demanding their "force tri" option. For the love of God, why??

 

[Quitch]

We don't know what option E is. We don't need to know. We may/will never know

YOU dont' need to know.
[\quote]

Exactly. Thanks for agreeing.

Come on. Don't be so arrogant. We have no idea what the performance hit is on R420 with 'real' TRI. Would you be so gung-ho if you found out that you could get real TRI with a 2% drop in performance and noticeable image quality improvement? These are all unknowns but your ridiculous 32X SSAA analogy just goes to show the lengths you will go to to evangelize this optimization.

We don't need to know. Seriously, if the currently exposed technique has equivelent or better image quality than previous trilinear methods, and better performance in general, does it matter if there is another method available, and if so, why?

And I'm not going to any length to evangelize this. Show me an issue and I'll deal with it, but the arguements so far are simply illogical. There is a method, that is used, that has higher performance, that appears to have equivelent or better image quality.

So far no one has been able to demonstrate an issue with the quality available.

I'll repeat, so far no one has been able to demonstrate an issue with the quality available.

It appears that some people have their britches in a tither because they weren't personally informed of this change.

Compare/contrast this with Brilinear where people were quickly able to produce screen shots from real game play where there were noticable image quality degredations.

Aaron Spink
speaking for myself inc.

This is dead on a post I made in another topic. I couldn't agree more.

 

[digitalwanderer]

I can't tell the difference, I just can't.

I drove meself a little nuts last night comparing playing games with AF on my 9600 to my 9700, but I couldn't tell a difference to save me ass.

I'm not saying that means there is no difference, just that I couldn't find it and I was looking hard!

If this is a "cheat", it's a damned good one! 8)

 

[lordvader1982]

Everyone's scrutinising the fine differences in image quality, but the fact is that most differences as to whether an image is better or worse is somewhat subjective.
I have, however, found a purely objective approach, and the results are very surprising.
When looking at the images with my eyes closed, I found that the results were identical!!!

 

[MrGaribaldi]

ARGHHHH! FULL TRILINEAR IS NOT THE BEST QUALITY AVAILABLE!

Where are you from? Trilinear is a fudge, and using it all the time is an even bigger fudge. This adaptive algorithm so far seems to be the best of the IQ and FPS worlds, yet people keep complaining and demanding their "force tri" option. For the love of God, why??

Isn't it obvious? "Tri" being larger than "Bi", it stands to reason that anything based on "Tri" will be the best solution until something based on "Qua"(d) comes around...


I think it's a 2 part answer to the question:
1. Most of us aren't educated enough to understand the difference between Bilinear and Trilinear filtering, and why neither of them is the best solution

2. People want as much apple-to-apple comparisons to base their judgement on. (Nevermind that such comparisons are (and will be) impossible (due to differences in approaches/algorithms), at least on the bit-wise comparisons being done now.)

Now, 1. is being solved by the more knowledgeable members telling us the difference (though some of us seems to need spoonfeeding), but 2 will be with us until we can come up with a better way to compare IQ.

 

[ShePearl]

"Tri" being larger than "Bi", it stands to reason that anything based on "Tri" will be the best solution until something based on "Qua"(d) comes around...

 

[Sxotty]

What I liked was how many people like the bilinear shot better than trilinear and said it looked better, I suppose video card manufacturers should simply stop filtering altoghether since it looks sharper

 

[Drak]

ARGHHHH! FULL TRILINEAR IS NOT THE BEST QUALITY AVAILABLE!

Where are you from? Trilinear is a fudge, and using it all the time is an even bigger fudge. This adaptive algorithm so far seems to be the best of the IQ and FPS worlds, yet people keep complaining and demanding their "force tri" option. For the love of God, why??

Why do you say trilinear filtering is a fudge?

 

[GraphixViolence]

Because it's nothing more than a way to cover up artifacts resulting from insufficient texture resolution, and there's no perfect solution. Just like anti-aliasing. The more samples you take, the better it can get, but it will never be a perfect substitute for just increasing the resolution. Trilinear is certainly not the be all and end all of texture filtering.

 

[Quitch]

Because it's an attempt to patch up mip maps, which in themselves are a fudge to stop texture crawl. Tri has sharp areas, and then blurry areas where the blending between mip map levels occur. The less filtering, the less blurring. Trilinear is popular because it's virtually invisible, but if there are situations where the mip map divides would not be very visible, then it is desirable to less filtering than trilinear offers because that leads to sharper, less blurry textures.

The reason that trilinear has been so worshipped up to now is that it was the best we had, the control panel chose a method and stuck with it. The only way to improve was for the developer to on a case by case basis. Now we get something that comes along and decides, in what appears to be a rather intelligent manner, what level of filtering is required. This is a very good thing, and we should all be thankful. Are we? No, because we've been raised on a simplified diet of "trilinear good, bilinear bad" and no longer know the reasoning behind it, of the down-sides of filtering.

 

[Bjorn]

This is a very good thing, and we should all be thankful. Are we? No, because we've been raised on a simplified diet of "trilinear good, bilinear bad" and no longer know the reasoning behind it, of the down-sides of filtering.

I think that most people are unhappy about how all this was handled, PR documents and all that. Not the actual filtering and what it offers (assuming that there isn't any "problems" with it).

 

[Quitch]

I think most people are unhappy because they really don't understand the issue, others because they weren't told (we haven't been told about a lot of the Catalyst changes, so I really don't see why this is such a shock).

The reviewers guide is a quandry, but ATI are probably coming at it from the angle that their algorithm delivers trilinear unless it decides it can lower the quality...

Frankly, if the reviewers guide is the worst of their crimes then I'm not going to lose a wink of sleep tonight.

 

[Bouncing Zabaglione Bros.]

I think that most people are unhappy about how all this was handled, PR documents and all that. Not the actual filtering and what it offers (assuming that there isn't any "problems" with it).

Yes, but for a lot of people that unhappiness has spilled into the actual discussion on the merits of this adaptive filtering method, and the results we get.

It just goes to show that there are a lot of people (including websites and reviewers) who just have not done their research before deciding to call this adaptive filtering "bad". Not becuase the filtering is bad, but because they feel that they were cheated and misled by ATI, and that anything not trilinear (as they have come to understand it) must therefore equal "worse quality".

 

[KimB]

Not becuase the filtering is bad, but because they feel that they were cheated and misled by ATI, and that anything not trilinear (as they have come to understand it) must therefore equal "worse quality".

Well, um, it is. ATI's technique, according to them, merely attempts to manage the image quality problems so that they aren't noticed. It won't have better image quality, and we don't yet know where the pathological cases lie.

 

[:PopCorn]

I think that most people are unhappy about how all this was handled, PR documents and all that. Not the actual filtering and what it offers (assuming that there isn't any "problems" with it).

Yes, but for a lot of people that unhappiness has spilled into the actual discussion on the merits of this adaptive filtering method, and the results we get.

It just goes to show that there are a lot of people (including websites and reviewers) who just have not done their research before deciding to call this adaptive filtering "bad". Not becuase the filtering is bad, but because they feel that they were cheated and misled by ATI, and that anything not trilinear (as they have come to understand it) must therefore equal "worse quality".

Bingo. Agree 110%

 

[Seiko]

I'm confused

Personally I think this is turing into a witch hunt. Are we now saying that all of the reviews that gave the IQ edge to ATI are now wrong? The fact that it still matches if not slightly surpasses Nvidias IQ and comes with less of a hit is a bad thing? The fact that no review site menioned that the filtering should/could have been improved and wasn't up to par with their expectations? Hmmm, this really comes across to me at least as a little information is a dangerous thing. Until this little piece of news came out everyone was apparently happy? The new technique has obviously proven that from an in game visual IQ perceivment it shows no sign of visual degredation.

Some may feel I'm blindly defending ATI but I feel I've been on both nVidias case and ATis but this simply doesn't register as a problem to me. It appears to me that those complaining could be categorised under the techy badge as the new technique blurs if not blatantly diregards the traditional tri methods. Personally however I think that unless the units of measurements have subsequently changed, i.e. from visually perceived IQ to check box style formula compliance I don't see the issue.

From a consumer perspective I'm glad ATI are pushing the boundary and seeking new ways to apply common sense techniques. Some have mentioned that they'd like to see yet another option to play with. I'm afraid I'd disagree again as labels have caused no end of turmoil for review sites and readers alike. The so called apples to apples is obviously not going to happen when trying to use levels of Tri, FSAA set to pre-categorised values, i.e. x2, x4 etc. We all remember how 3DFXs x2 FSAA blew away Nvidias x2 in terms of IQ and yet this was a so called apples to apples comparison? Why just because the settings both displayed x2. It was plain to see visually that the 3DFX x2 ratio was far superior to Nvidias and yet FPS where compared. I'm beginning to think and I know the techies won't like this but hey don't shoot me, that we may be better off loosing the naming conventions and individual tri/FSAA settings altogether. May be a simple IQ setting could be used so that we can see the best IQ for a given framerate target. Texture sharpness and FSAA obviously come into this but should it be up to the review sites to set them individually? Lets be honest so many simply set the numbers to match and bingo assume it's a fair comparison. Obviously this a dangerous assumption and although some now actually try and make a visual IQ test to check they look approximately the same under general gameplay alot of people appear to be saying, "Well it doesn't matter if the IQ is the same, the method of implementation should be judged?"

I'm confused, surely we all agree that the visually perceived IQ is the holy grail here and not the internal implementation?

 

[digitalwanderer]

I think that most people are unhappy about how all this was handled, PR documents and all that. Not the actual filtering and what it offers (assuming that there isn't any "problems" with it).

Yes, but for a lot of people that unhappiness has spilled into the actual discussion on the merits of this adaptive filtering method, and the results we get.

It just goes to show that there are a lot of people (including websites and reviewers) who just have not done their research before deciding to call this adaptive filtering "bad". Not becuase the filtering is bad, but because they feel that they were cheated and misled by ATI, and that anything not trilinear (as they have come to understand it) must therefore equal "worse quality".

Yup, I think that is the fuel powering this witchhunt currently!

 

[Bjorn]

Re: I'm confused

Personally I think this is turing into a witch hunt. Are we now saying that all of the reviews that gave the IQ edge to ATI are now wrong? The fact that it still matches if not slightly surpasses Nvidias IQ and comes with less of a hit is a bad thing?

I thought that most reviews says that NVidia has slightly better AF/filtering quality. Although i have to admit that i haven't read that many reviews.