Ways to solve complaining about ATI's filtering?

[Drak]

If it's the case that the hardware supports it, which I expect it does, as R420 does actually do trilinear where needed.

The hardware supports full trilinear, there's no question about it. The driver tells the hardware what to do.

I also voted for a checkbox in the CP, for the same reason - it allows ATI to keep the high ground and lets customers feel happier even if they are actually losing out on a good optimisation.

Some people (like me), need to see/feel what they're losing out on.

 

[Drak]

Well, if I shelf out $499 to buy the card and want to use *traditional* full trilinear(since the card can do it, but was disabled), will this be a good enough reason ?

If the card doesn't offer it, only "smartfiltering" (ie better, faster), then you can't have it. Just like you can't have non-angle dependent AF on R420 or NV40.

The card does offer it, but the driver doesn't!

 

[Rys]

Even a registry key toggle would do. That would allow 3rd party tool makers to support it, even if ATI don't want to.

Rys

 

[Quitch]

Which is exactly why asking for a way to disable it at this point seems a little premature. Why disable a good as/higher IQ + higher performance method, except through not really understanding it?

I don't think the cry should go out for such an option unless the method is found inferior, which so far it hasn't. Even if it is, who's to say this isn't one or two easily correctable cases? Some people seem obsessed with maintaining very old tech because they understand what it does. "I know what tri does so I wish to force it because tri is best." It's living in the past.

So, because you don't get why people would like to disable trylinear, you want to impose your point of view on others so they can't disable it?

Choice is always power. It's not the fact that we want to disable it, we want to have the choice. We can try it on/off, probably decide that we want it on and be very happy with it.

You have to remember, it's in human nature to yearn for what you can't have. People are rarely satisfied with what they have. By removing the option for full trilinear, ATI is making it more desirable than it should be.

ATI have already stated that the driver is already running an analysis in software (at texture upload time) and sets a flag or variable to enable/disable trylinear for that texture in hardware. I bet it's not even going to cost them one miserable man/hour to put an "if registry entry exists then set trylinear flag to 0". Come on. One man hour is very generous . They'd need maybe 20 minutes to implement the option and 40 minutes to test it.

No, I think the way I do because I don't believe you know what you're asking for. An adaptive method will always, assuming the method is designed well enough, be better than naive trilinear filtering. The only question left is: does ATI's method hold up to scrutiny.

Thus far, the answer appears to be yes.

The second such an option appears of course, we'll get all manner of reviews claming to be "apples-to-apples" without actually being any such thing, and not really understanding why.

 

[MrGaribaldi]

That might be great for reviewer, but for the end user it is quite irrelevant.

What's better for the reviewer should be better for the end consumer also since he'll be able to get a better picture of the cards performance in all circumstances.

Don't bet on it. Reviewers have a habit of making an arse of the options.

But DaveB has gotten a copy of a program which shows legacy-trilinear and trylinear filtering, and the blind test said that most of us couldn't see any diff...

ATI could distribute those apps to reviewers/customers so you could see for yourself that there was next-to-no difference.

That would remove the problem with unfavourable benches, and show you the differences.
Or if they don't give out such apps, create them yourself, since ATI has given you the information on what the app deems "tri-worthy" compared to "tryl-worthy". (Though finding the exact threshold will require good eyes )


And as for "apples-to-apples", it would seem that Brent is the only one who's doing this (B3D doesn't do comparisons) with a better parameter than "I'll just try to match up numbers in the CP and it'll be perfect"...

 

[Bouncing Zabaglione Bros.]

Yes but it also exposes them to reviewers with bad methods or motives saying that it is only fair to bench Tri opts off on NV40 vs Adaptive Trilinear off on X800 and they will take a huge hit for no reason since Adaptive Tri is until someone proves otherwise as good or better the NV40's Trilinear.

But that will always be the case, just like those reviewers that benched ATI and Nvidia at the same control panel AA settings - even though Nvidia was doing half the anti-aliasing at any given setting.

The answer is educating the reviewers, and hoping that the bad ones are outnumbered by the good ones.

 

[Stryyder]

Yes but it also exposes them to reviewers with bad methods or motives saying that it is only fair to bench Tri opts off on NV40 vs Adaptive Trilinear off on X800 and they will take a huge hit for no reason since Adaptive Tri is until someone proves otherwise as good or better the NV40's Trilinear.

But that will always be the case, just like those reviewers that benched ATI and Nvidia at the same control panel AA settings - even though Nvidia was doing half the anti-aliasing at any given setting.

The answer is educating the reviewers, and hoping that the bad ones are outnumbered by the good ones.

You can't educate bias and unprofessinalism out of some of these guys, they get hammered here and still defend there methods with flimsy bs explanations.

 

[Dave Baumann]

While this remains as the defaul for ATI then just compare with Tri optimisations for the both.

 

[Stryyder]

While this remains as the defaul for ATI then just compare with Tri optimisations for the both.

Noting the IQ differences of course I am assuming.

 

[Dave Baumann]

If there are any.

 

[Ylandro]

The answer is educating the reviewers, and hoping that the bad ones are outnumbered by the good ones.

Are you serious? The quality of reviews has been dropping steadily for years! And not just because the influx of bad ones, but also because good ones turn bad. (Tom & Anand come to mind....)
There is hardly any reviewer out there that I trust. Dave and Brent being the very few exceptions.

I am completely sure that as soon as ATI includes a 'full-trilinear' option 99% of the review sites with immediately use it for what they think are 'apples to apples' tests, and complete ignore this intelligent filtering. And it won't matter how much evidence ATI provides that both give the same IQ.

 

[Stryyder]

Ylandro just hit the nail on the head.

 

[Ostsol]

Well, looking at how many reviews are doing benches of the NV40 with trilinear optmizations on and off, the same might be done for the R420. In fact, perhaps reviewers will compare the IQ and performance of NVidia's and ATI's optmizations. That might be interesting to see.

 

[zyxwvu4321]

Hello all. I've followed computer stuff for a few years now, but just today registered here, as it seems to be the best place to discuss this topic. As such, I have a couple of related questions that I have been wondering about, and I'm hoping that they could be answered here:

1) Would the fact that this optimization puts pushes the R420 ahead of the NV40 when the filtering is applied imply that ATI's advantage this generation in filtering/AF performance is software, rather than hardware based, and

2) Is it possible for nVidia to impliment an optimization that produces the same effects (just as good IQ, as it appears now, and no/minimal performance loss) given that ATi is in the process of patenting it?

Thanks in advance.

 

[Bouncing Zabaglione Bros.]

Are you serious? The quality of reviews has been dropping steadily for years! And not just because the influx of bad ones, but also because good ones turn bad. (Tom & Anand come to mind....)
There is hardly any reviewer out there that I trust. Dave and Brent being the very few exceptions.

You can't educate bias and unprofessinalism out of some of these guys, they get hammered here and still defend there methods with flimsy bs explanations.

Exactly my point. Is putting in a checkbox to allow them to bench slower any better or worse than letting them claim ATI is cheating because "they didn't tell us" or using COD shots at bilinear and then claiming the adaptive method is worse quality?

Basically, no matter what you do, a certain breed of reviewer will try to crucify you or spin it out how they want to, so the last thing you should do is pander to those guys. If you play that game, that is a no-win situation for ATI.

 

[digitalwanderer]

If there are any.

If there are any, I'm starting to think that the image quality is comprable if not exact to trilinear with this new method....I'm going a bit nuts just trying to be able to tell the difference and if I DO start to notice a difference I still can't decide if it's better or worse looking, it's too damned close.

 

[nutball]

If there are any, I'm starting to think that the image quality is comprable if not exact to trilinear with this new method....I'm going a bit nuts just trying to be able to tell the difference and if I DO start to notice a difference I still can't decide if it's better or worse looking, it's too damned close.

Yeah, kinda makes you hanker for the old days, doesn't it?

 

[digitalwanderer]

Yeah, kinda makes you hanker for the old days, doesn't it?

Nah, as much as I whine about it I kind of like change...it keeps me on me toes and helps ward off stagnation.

I think in the last week or so there was a huge paradigm shift in the graphics world and I'm still kind of getting settled in to it. That doesn't mean that things are all bad, just a bit different and we're going to have to find new ways to measure some things and some new ways to compare/show features and abilities of the new hardware coming out.

I'm always hip for a challenge, and this seems more like an opportunity to try some new things more than anything else to me. 8)

 

[bGeek]

While this remains as the defaul for ATI then just compare with Tri optimisations for the both.

Noting the IQ differences of course I am assuming.

If there are any.

If there are any, I'm starting to think that the image quality is comprable if not exact to trilinear with this new method....I'm going a bit nuts just trying to be able to tell the difference and if I DO start to notice a difference I still can't decide if it's better or worse looking, it's too damned close.

I actually took Dave's comment to mean any difference between Nvidia's Brilinear and ATI's Trylinear.

 

[Stryyder]

While this remains as the defaul for ATI then just compare with Tri optimisations for the both.

Noting the IQ differences of course I am assuming.

If there are any.

If there are any, I'm starting to think that the image quality is comprable if not exact to trilinear with this new method....I'm going a bit nuts just trying to be able to tell the difference and if I DO start to notice a difference I still can't decide if it's better or worse looking, it's too damned close.

I actually took Dave's comment to mean any difference between Nvidia's Brilinear and ATI's Trylinear.

Unfortunately I took it the same way..... So is ATI adaptive trilinear on par with Brilinear or legacy Trilinear??

Why do I see reviews going from 30 pages longe to 50 in the near future.