Can someone sum up the ATI filtering thread please?

[Geo]

I don't feel like reading all 51 pages. Just a nice, quick, easily digestible summation for those of us who don't give a damn about the technicalities and just want to know the easy-to-read conclusions.

War is Bad.
Peace is Good.
Never use plastic when you can use Wood.
Be kind to strangers.
Give good jobs to vets.
[mumble, mumble --Sorry, Steve]
Spay/neuter your pets.

 

[hoom]

As far as I have noticed, nobody has proven any real negative image quality effect.

ATI claims that they have some form of comparison between mip levels so that where the mips are similar enough (as is the generally originally intended use of mipmaps), they go brilinear because people 'won't notice' & its faster but where there is a significant difference between mip levels (generally indicative of more modern mipmap usage where odd things might be happening), they do the full trilinear because thats the expected behaviour.

 

[davepermen]

wether or not atis claim is true is now a <bleep> topic. fact is, ati implemented it over a year ago, and nobody till now realised something is wrong. this is quite a bit of a prove, showing it really works.

A. They implemented it on a low-end card. I don't think it was really a big issue to many people. But the R420 is a high-end card, and so any "cheats" are much higher-profile.
B. Some people definitely knew about it. There were questions early-on after the R420's release whether or not it was doing "brilinear" like the Radeon 9600.

A) this doesnt make any difference. you pay more, and get more performance. else, its an about equal product. so what?. and its still no cheat, as you call it. there is no visible difference proven yet, except some russian page.
B) yeah, the devteam knew it. the question about brilinear was because they knew nvidia did it, nothing else. and this optimisation is in NO way to compare with brilinear on nvidias hw.

you are a really dump person, if you don't grasp this. if you don't want to, you're just trolling. eighter way, you're rather dump. sorry to directly insult you, but this is now a long ongoing discussion, and you still haven't moved one bit right from the start.

 

[no_way]

Well, I think you can sum it up like this:
Nvidia Fans - "OMG! ATI are cheating!!!!!!!!!!!"
ATI Fans - "No they are not!!"

everyone : developer control over which algorithms get used in his application ? who gives a flying f ?

 

[Quitch]

Ok. So ATI's filtering can be toggled by the end user, unlike Nvidia Brilinear then? If so, then I don't have a problem with it. If not, then I do.

And this is why you need to read the long thread... it's being duplicated in here already with the same players and the same arguments. Maybe once we've gone through the event four times we'll figure out what we need to change at the end of the argument to prevent the causality loop!

*prepares to kill Chalnoth*

 

[KimB]

you are a really dump person, if you don't grasp this.

Just because I have a different opinion from you I'm dumb?

Anyway, this was posted once before, but here it is:
http://www.ixbt.com/video2/nv40-rx800-3.shtml

Now, it would be useful to have this in English, but it should be easy to see that, according to ixbt's tests, this is nothing more than simple "brilinear," the same brilinear that nVidia uses. It's just a slightly different implementation of the technique, and apparently it's less aggressive, and thus less visible.

So, if this is the case, I now know the obvious worst-case scenario for this technique: motion. Actually play a game and it'll be more visible than looking at a screenshot.

I'm really loving how you're having all these claims about nVidia "cheating" and ATI having this great "optimization," while it's now starting to look like ATI's doing essentially the exact same thing (just less aggressively), while at the same time disabling it for tests that would expose the technique.

 

[digitalwanderer]

 

[jvd]

you are a really dump person, if you don't grasp this.

Just because I have a different opinion from you I'm dumb?

Anyway, this was posted once before, but here it is:
http://www.ixbt.com/video2/nv40-rx800-3.shtml

Now, it would be useful to have this in English, but it should be easy to see that, according to ixbt's tests, this is nothing more than simple "brilinear," the same brilinear that nVidia uses. It's just a slightly different implementation of the technique, and apparently it's less aggressive, and thus less visible.

So, if this is the case, I now know the obvious worst-case scenario for this technique: motion. Actually play a game and it'll be more visible than looking at a screenshot.

I'm really loving how you're having all these claims about nVidia "cheating" and ATI having this great "optimization," while it's now starting to look like ATI's doing essentially the exact same thing (just less aggressively), while at the same time disabling it for tests that would expose the technique.

Right but i've told you there is no diffrence in motion over a 9700pro.

And this is in english. So I guess we are done listening to you argue .

This witch hunting is getting sad . Mabye nvidia should make some slides teaching you how to bash ati some more.... oh wait they already did that .

 

[Bjorn]

Right but i've told you there is no diffrence in motion over a 9700pro.

I'm guessing that most people wouldn't see the difference between trilinear and NVidias brilinear either.

 

[Evildeus]

Right but i've told you there is no diffrence in motion over a 9700pro.

May i ask, how do you know?

 

[vb]

Ok. So ATI's filtering can be toggled by the end user, unlike Nvidia Brilinear then? If so, then I don't have a problem with it. If not, then I do.

First you'll have to notice it.
then you'll have to decide which is better.

Or you can have problems on regular basis.

 

[cthellis42]

My heart warmed a bit at being used as a sum-up of a 50-page thread from late in the game. ^_^ Though sadly this thread seems apt to devolve just as quickly...

 

[Luminescent]

Will this sort of adaptive filtering be available in the 9700/9800 series drivers?

 

[Evildeus]

Will this sort of adaptive filtering be available in the 9700/9800 series drivers?

AFAIK, no.

 

[Dave Baumann]

Will this sort of adaptive filtering be available in the 9700/9800 series drivers?

The 150nm products do not have the hardware capability for it.

 

[Evildeus]

So know that we know the results of both poll, what are your comments?
It seems that determining which one is the R420 was difficult, but it's clearly the least prefered....

A = NV38
B = R360
C = R420

Which is R420 ?
A    19%  [ 17 ] 
B    8%  [ 8 ] 
C    25%  [ 23 ] 
Don't know / Can't tell     46%  [ 41 ]

Which is your IQ preference ?
Board A    24%  [ 19 ] 
Board B    38%  [ 30 ] 
Board C    11%  [ 9 ] 
All Look the Same / Can't Tell    24%  [ 19 ]

 

[Dave Glue]

what do you call the FX-5200 -- a piece of shit?

Oh! Oh! I know the answer to that one!

 

[Clootie]

Anyway, this was posted once before, but here it is:
http://www.ixbt.com/video2/nv40-rx800-3.shtml

Now, it would be useful to have this in English, but it should be easy to see that, according to ixbt's tests, this is nothing more than simple "brilinear," the same brilinear that nVidia uses. It's just a slightly different implementation of the technique, and apparently it's less aggressive, and thus less visible.

Yep, it's the same brilinear, switched on/off by driver per texture. Driver tries to detect if texture mip-levels are different enought and if they are - switch off optimizations. Another thing to note: current drivers (till Cat 4.5) use brilinear only on DirectX MANAGED textures.

In comparision to NVIDIA:
1) linear interpolations between MIP levels takes place in different space (before LOG(...)) => NV do it "by book", ATI saves transistors.
2) ATI filtering always uses MORE data from HIGH detail mipmaps comparing to NVIDIA (both in standart and optimized modes) - so better screenshots, but should have more shimmering in motion.

Right but i've told you there is no diffrence in motion over a 9700pro.

If you are proud owner of Radeon 9600/X800 you can download utility attached in article above: ixbt-filter.rar. And explore by yourself difference in filtering both in static screenshots and in motion. If you'll need help spotting differences - just press "D" - and program will help you in that

PS. Standart behaviour of utility is similiar to the one provided to Dave by ATI.

 

[jvd]

Right but i've told you there is no diffrence in motion over a 9700pro.

I'm guessing that most people wouldn't see the difference between trilinear and NVidias brilinear either.

I disagree as you can see it in my fx 5800 ultra in motion . Though it has gotten better. The performance has gone down also .

 

[jvd]

Right but i've told you there is no diffrence in motion over a 9700pro.

May i ask, how do you know?

As i had said before. My family owns a 9600pro. My buddy (who lives next door) owns a x800pro. I own a 9700pro and I also have a 5800ultra. We have all tested and had them running and to test again we used the same monitor and my father changed the plug so we could judge the quality.

The first test we did was we had all the cards running on trintrons all the same model but we cuoldn't see any diffrence only my sister could on the 5800ultra.

On the same monitor we can only see a diffrence with the 5800ultra and its only slightly worse than the radeons . But we all see the diffrence now.