Trilinear Test (Warning 0.5MB PNG)

[breez]

Anyway, the point of the application is to provide textures that will defeat the optimisation - so, on the left hand side you have an filtering via the optimised mode and on the right to have "optimisation defeated" (i.e. Traditional) trilinear.

So 40-45% of the voters can tell which one is the good one. I wouldn't say it's non obvious.
PS: i think you should have precise what to look for, because not everybody knows what the filtering does...

Why on earth would they need to know what it does to pick out the image? I think people are picking out the one they think looks better, which is surely the whole point??

The point was to pick up the "Traditional Trilinear", not the one that looks best (yes, there's the possibility that optimized image looks better )

 

[Evildeus]

Why on earth would they need to know what it does to pick out the image? I think people are picking out the one they think looks better, which is surely the whole point??

Because the point of trilinear is to hide the step between each mipmaps and that can decrease the sharpness of the pic?

 

[K.I.L.E.R]

Anyway, the point of the application is to provide textures that will defeat the optimisation - so, on the left hand side you have an filtering via the optimised mode and on the right to have "optimisation defeated" (i.e. Traditional) trilinear.

So 40-45% of the voters can tell which one is the good one. I wouldn't say it's non obvious.
PS: i think you should have precise what to look for, because not everybody knows what the filtering does...

I made a guess that it was on the left based on staring at the image struggling to make any decision.

I'm sure the vast majority of people couldn't tell the difference without either doing something like zooming into the image 800% or whatnot.

So the statistics are flawed.

 

[Martillo1]

Motion tests, please.

Edit:

Actually, I guessed it was the right by seeing it from more distance.

Edit:

Trylinear (ATI) v.s. Trilinear (*/NVIDIA) v.s. Brilinear (NVIDIA).

 

[tEd]

Motion tests, please.

...with nvidias brilinear too

 

[mikechai]

Motion tests, please.
Trylinear (ATI) v.s. Trilinear (*/NVIDIA) v.s. Brilinear (NVIDIA).

Excellent!


________
HONDA VF1000F

 

[Martin Eddy]

And how many votes went on the right side after bloodbob post ?

The point was using eyes not cheating the test

Actually I could see the difference better on the original pic.
Bloodbob's pic just strained my eyes more.

Oh god! I've got a killer headache after all this.

Honestly I can't see any real difference within a game ,but as someone said here when the FX farce was in full swing, "It's a slippery slope"

How long will it be before neither company gives us FULL TRILINEAR?

 

[bloodbob]

why there is not such poll when nvidia implement brilinear on nv3x

Because there would have been no difference between left and right (and because this app has only just been provided to me, by ATI, something that NVIDIA probably wouldn't have done - they have test apps in house that I've seen and asked for and never got).

Anyway, the point of the application is to provide textures that will defeat the optimisation - so, on the left hand side you have an filtering via the optimised mode and on the right to have "optimisation defeated" (i.e. Traditional) trilinear.

How is it defeated because if they are using a different filter method on the mip-map the results can look vastly different I was showing off that point in another thread http://www.beyond3d.com/forum/viewtopic.php?t=12531 where one filter would produce a white mip and another filter would produce a grey mipmap. And as such that would invalidate any side by side comparision of the filtering methods.

 

[Bouncing Zabaglione Bros.]

How long will it be before neither company gives us FULL TRILINEAR?

Well look at it this way: Trilinear is solution to the problem of visible mip-map transitions, but it does have the drawback of blurring textures (also undesirable). As soon as that mipmap transition problem is solved by a better solution, why would we need trilinear at all? Maybe this adaptive method is a step along the way. The ATI transitions are not visible in the same way as they are on Nvidia's solution to the same problem, so it certainly seems like a step in the right, as opposed to the wrong direction.

One day, maybe there won't be multiple texures and mipmap transitions at all, so we won't be lamenting the loss of trilinear filtering to hide these transition artifacts then, will we?

 

[Xmas]

I don't think the options in the vote fit here. I couldn't tell which one was true trilinear, or whether there's true trilinear at all. But I could tell that they are different, and that I feel the right one looks better. There's a slight variation in brightness (depending on your gamma setting), and seen from a distance the right one looks a bit smoother.

 

[Ylandro]

Yep... there should be an option: "after ruining my eyes for 15 minutes staring at the bloody picture, I tend to think I see a difference, but don't have a clue which one is better".

For me, I voted both similar. Then looked again, and again, and again....
And eventually, I think the left looks sharper while I see more color on the right. Could also be that the tears in my eyes are starting to distort it...

 

[Sxotty]

Do you think it would work better if you had thicker lines? The lines here are like those illusions you look at that dance around before your eyes. At least that is what I think, the number of visual artifacts your brain makes is quite large

 

[Drak]

So whats interesting is that I've now gotten a chance to look at this on 3 different displays. My first response was based off of the display on my laptop (Thinkpad T40 with 1024x768 display) where I could see a little more moire on the left hand side.

Since I've gotten home, I've looked at it on my Dell 1901FP and a Compaq P110 (rebadged 21" sony trinitron). I really can't tell much of a difference on either of these displays. Interesting to say the least.

Aaron Spink
speaking for myself inc.

You're bang on the money. Moire patterns occur when sampling a high frequency pattern at too low a resolution. That gave me a hint as to which one was trylinear. I couldn't see any moire on the right image on a 21 inch Vision Master Pro 510.

edit: Now, I can see some slighter moire on the right after staring at the image for two-three minutes. But the moire on the left was more easily visible when scanning the image the first time.

 

[Drak]

Dave, I think your coloured grating messes with the human visual system and causes some nasty adaptation effects. You don't want to move into the field of psycho-physics research, do you?

 

[digitalwanderer]

(and because this app has only just been provided to me, by ATI, something that NVIDIA probably wouldn't have done - they have test apps in house that I've seen and asked for and never got).

Way cool of 'em. 8) (At least they're returning someone's e-mails...unfortunately now I know it's a personal issue with me that they ain't returning mine anymore. *sadpanda* )

Anyway, the point of the application is to provide textures that will defeat the optimisation - so, on the left hand side you have an filtering via the optimised mode and on the right to have "optimisation defeated" (i.e. Traditional) trilinear.

Damn, it got me.

I could tell some subtle differences 'tween the sides, but I couldn't decide which one was "better". (I didn't zoom in or nothing, just looked at the pictures as I thought that was the point.)

Thanks Dave, I get it now. 8)

 

[Ostsol]

Heh! I just checked the shot on an LCD monitor and can see zero difference. Given that this texture is in the area of "worst case scenario", I doubt that many game textures would show any noticable difference at all.

At home on my 17 inch CRT I could definitely see that there was a difference, but it was quite subtle. Only when stepping back about a metre from where I normally sit was I able to really see the difference.

 

[Drak]

Because there would have been no difference between left and right (and because this app has only just been provided to me, by ATI, something that NVIDIA probably wouldn't have done - they have test apps in house that I've seen and asked for and never got).

Anyway, the point of the application is to provide textures that will defeat the optimisation - so, on the left hand side you have an filtering via the optimised mode and on the right to have "optimisation defeated" (i.e. Traditional) trilinear.

Thanks again for that Dave. Do you know if ATI is defeating the optimisation by the content of the textures or some other means? I'm wondering if the more promiment moire patterns on the left (than the right) have anything to do with the optimisation-defeating mechanism or the different filtering method itself?

 

[digitalwanderer]

I just checked out Bloodbob's picture after reading Kruno's link and if I had studied that picture really carefully for a very long while I'm pretty sure I STILL would have chose the left side.

 

[Mephisto]

There is no sense in comparing still images (especialy if they're provided by ATI) - mip map transitions are very annoying in the moving image, thats where brilinear has its flaws.

 

[Rodéric]

I can see the difference and find the "correct" trilinear on a LCD screen.

The only problem I have is that I find the left part better than the right...