My R9700 experince and Aniso still "flawd"?

[KimB]

"Extremely noticeable"? Yet it still gives quality better than the GeForce 4 (by your own words below)? I guess that means that the GeForce 4 aniso isn't good at all. And look at the performance hit...

No, I didn't say it "gives better quality than the GeForce4." It appears to give similar quality at 45 degrees, which means, to me, that it gives worse quality at the 22.5 degree and 67.5 degree angles.

Funny thing is that people on Rage 3D are saying that the Radeon 9700 looks beautiful in FS 2002. Here's an example:
http://www.rage3d.com/board/showthread.php?s=&threadid=33634907

From a person that used to own a Radeon 8500. Of course it looks better.

 

[Donald]

I think that I can solve the "are these shots are taken with the same settings" problem. I have a GF4 running in my server box at home and of course the 9700 running in my workstation. I will setup identical settings tonight using 4x aa and 8x af on both cards in 1024 and take shots (or should I leave out the aa?). I will use latest official drivers for each card as well.

Will put them in a big side by side image or something like that so tis easier to check.

Oh and that floor is moving so I will do my best to capture the images at the identical angle.

 

[Bjorn]

Oh and that floor is moving so I will do my best to capture the images at the identical angle.

Isn't it easier to just do a couple of saves and then copy it over to the other machine ?

 

[Donald]

Yes it is actually but the floor moves kind of quickly. Like I said, I will try my best to make sure they are identical.

Thanks for the idea.

 

[OpenGL guy]

It appears to give similar quality at 45 degrees, which means, to me, that it gives worse quality at the 22.5 degree and 67.5 degree angles.

Prove it.

 

[KimB]

It appears to give similar quality at 45 degrees, which means, to me, that it gives worse quality at the 22.5 degree and 67.5 degree angles.

Prove it.

Heh, I think I'll need a bit of help with this one.

Donald, would you mind attempting to capture shots halfway between 45 degrees and the horizontal, as well as at 45 degrees? I'd like to see if previous posters on this thread were indeed correct in their assertions.

 

[Donald]

Np Chalnoth.

I think that the maximum degree of rotation in that room with the moving floor is in fact 45 degrees. I will take one with the floor level, the left and right 45 degree angles again, and one shot in between those on each side.

I have an appt. with a client to fix a dead computer tonight so I should be able to get it all done sometime later this evening.

 

[Oats]

Donald, could you also post a 16x AF shot to?

 

[Bigus Dickus]

While you're at it, could you do 0x, 2x, and 4x AA with each? And try to capture 5, 10, 15, 20, 25, 30, 35, 40, and 45 degree rotations. Also both 8x and 16x AF would be nice. Perhaps throw in Quincunx on the GF4 since it might be responsible for the bluriness in the previously posted pics. And maybe a set showing both 4X MS and 4Xs (MS/SS) for the GF4 to see if that help clear up textures...

Just kidding.

Seriously though, I've found that when I get into a comparison like that, it's always better (and easier) to take a whole crapload of screenshots the first time through and decide which ones are necessary later, rather than having to go back after the fact and try to get a certain condition that was missed. Both 8x and 16x would probably be helpful, but I don't think AA would be (or should be used).

 

[Xmas]

As I said before and will say yet again: There will never be a time when anisotropic filtering is not applied. The obvious exception being when anisotropic filtering is not needed.

So could you tell us precisely when the chip decides that anisotropic filtering is not needed?

 

[OpenGL guy]

So could you tell us precisely when the chip decides that anisotropic filtering is not needed?

When there is no anisotropy?

 

[Donald]

While you're at it, could you do 0x, 2x, and 4x AA with each? And try to capture 5, 10, 15, 20, 25, 30, 35, 40, and 45 degree rotations. Also both 8x and 16x AF would be nice. Perhaps throw in Quincunx on the GF4 since it might be responsible for the bluriness in the previously posted pics. And maybe a set showing both 4X MS and 4Xs (MS/SS) for the GF4 to see if that help clear up textures...

Just kidding.

Seriously though, I've found that when I get into a comparison like that, it's always better (and easier) to take a whole crapload of screenshots the first time through and decide which ones are necessary later, rather than having to go back after the fact and try to get a certain condition that was missed. Both 8x and 16x would probably be helpful, but I don't think AA would be (or should be used).

You had me going for a moment there, I thought my whole night was screwed. Although I could bust out the ole' protractor and try to get those angles down for you

Ok, I will take shots at 8x & 16x on the 9700 with the angles that Chalnoth wants. Man I haven't had to touch the GF4 controls in ages once I set them up eons ago. It can only do 8x IIRC so the 16x is for the 9700 only. I will leave AA off so there is no chance of it causing any blurring.

 

[Xmas]

So could you tell us precisely when the chip decides that anisotropic filtering is not needed?

When there is no anisotropy?

So you mean only when du/dx and dv/dy as well as dv/dx and du/dy are exactly identical? Well, that's great

 

[Joe DeFuria]

Donald...

One last suggestion. For the GeForce, do both 4X and 8X. There has been debate / question on the difference in quality between those two modes.

So, that would be 4X and 8X on the GeForce, and 8X and 16 X on the 9700. Thanks, and may the force be with you.

 

[OpenGL guy]

So you mean only when du/dx and dv/dy as well as dv/dx and du/dy are exactly identical? Well, that's great

You can do better than that. Bilinear filter does work sometimes, you know.

 

[Xmas]

Ok, I will take shots at 8x & 16x on the 9700 with the angles that Chalnoth wants. Man I haven't had to touch the GF4 controls in ages once I set them up eons ago. It can only do 8x IIRC so the 16x is for the 9700 only. I will leave AA off so there is no chance of it causing any blurring.

Oh, and if you have the time, please also try to adjust the LOD bias of one card to get about the same texture quality on the far wall.
This wall shouldn't be affected much by AF, so it's obvious that the cards are using different LOD settings.

 

[Donald]

Donald...

One last suggestion. For the GeForce, do both 4X and 8X. There has been debate / question on the difference in quality between those two modes.

So, that would be 4X and 8X on the GeForce, and 8X and 16 X on the 9700. Thanks, and may the force be with you.

np, and I will do all captures in OpenGL using Grab Clip Save.

Also, I must know more about this force you speak of....

 

[Xmas]

So you mean only when du/dx and dv/dy as well as dv/dx and du/dy are exactly identical? Well, that's great

You can do better than that. Bilinear filter does work sometimes, you know.

Well, bilinear filtering "works" if the projection of a pixel into texture space is about square. If it's not, you should use more samples representing the direction in which the projection is "stretched".

 

[OpenGL guy]

Well, bilinear filtering "works" if the projection of a pixel into texture space is about square. If it's not, you should use more samples representing the direction in which the projection is "stretched".

Yes, I know the definition of anisotropy.

 

[Randell]

Donald whilst you're at it could you take those screenshots standing on one hand gargling the UK National Anthem whilst juggling a football with your feet?

Thanks