My R9700 experince and Aniso still "flawd"?

[OpenGL guy]

While it is definitely better than what the Radeon 8500 offered, I still consider it a flaw.

Good for you.

In particular, in situations such as the one shown in Serious Sam, it's extremely noticeable. Flight sims would almost certainly show the issue more commonly (as has been repeated many, many times).

"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...

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

Now, if you want to attempt to make the argument that at its worst, the 9700 still isn't as bad as the GeForce4's max aniso, here's what I have to say:

1. The 45-degree angle is no longer the worst-case for the 9700. The worst-case is apparently at 22.5 degrees (according to previous posters on this thread), and every 45 degrees after that, which means to me that the 9700 shot posted at the beginning of this thread isn't necessarily a worst-case scenario.

You're basing all your comments on what other people have posted. When you do your own research, maybe I'll take you seriously.

2. While the approximation does make it so that the wort-case isn't as bad as the 8500's worst case, the flaw is apparently more common (I still consider it better than the 8500's, mind you...from sort of a "purity of rendering" standpoint...).

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.

When those scenarios where the 9700 shows this flaw obviously come up, they will almost certainly be very obvious. I would go so far as to call the problem distracting.

How would you know since you have never owned or used a Radeon 9700, or Radeon 8500 for that matter? You've never owned a Radeon 8500 or 9700, yet you are a self-proclaimed expert on their anisotropic filtering. And what's distracting? The fact that the Radeon 9700 gives better anisotropic filtering, with better peformance to boot, than the GeForce 4?

 

[andypski]

2. While the approximation does make it so that the wort-case isn't as bad as the 8500's worst case, the flaw is apparently more common (I still consider it better than the 8500's, mind you...from sort of a "purity of rendering" standpoint...)

This is not an assumption you can make from the current evidence, or at least it is a poorly thought out conclusion.

If you remember - the decrease in anisotropy on R8500 is a continuous process as the plane of the polygon is rotated.

If we are to assume the case where the falloff in R9700's anisotropy matches that on R8500 until you reach 22.5 degrees (or wherever the minima is) and then climbs back up again as you approach 45 degrees then this is not a 'more common' flaw. What is happening in reality is that the minima 'nodes' occur more frequently, but the error itself is still reduced in all cases and is at most equal to that of R8500 at any given angle. Now in addition it may be that R9700's anisotropy actually falls off less rapidly, in which case the error is then reduced in all cases.

I think that stating that a 'flaw' occurs more frequently presents this information in an unnecessarily negative light, as I'm sure you'll agree

- Andy.

 

[demalion]

While it is definitely better than what the Radeon 8500 offered, I still consider it a flaw.

In particular, in situations such as the one shown in Serious Sam, it's extremely noticeable. Flight sims would almost certainly show the issue more commonly (as has been repeated many, many times).

...

3. Visual quality is relative. Similar to how people have said they'd rather take 30fps constant over 20 fps minimum, 60 fps average any day. When those scenarios where the 9700 shows this flaw obviously come up, they will almost certainly be very obvious. I would go so far as to call the problem distracting.

I looked at those shots a few times. To me, it quite clearly looks like th "angled" shots are "bad"...when compared to the other 9700 shots. They also look identical to me to shots I remember for the GeForce cards in "non-angled" viewpoints in the quality displayed. With you being a GeForce owner, it just strikes me as odd that you make the comments you do regarding this. But that is just based on my experience with screenshots, not with actually owning a card, exactly what you've been criticized for doing....though I don't recall anyone disagreeing with that sentiment among the GF owners (maybe I read too hastily?).

What I will say is that I don't find your viewpoint wrong or worthy of attack in itself (atleast in general...the specifics have already been addressed by others). It seems a valid viewpoint (again, in general) to have and hold...but perhaps you should take this opportunity to go through the recent threads, as I have after coming here to catch up, and to see how incredibly consistently and (IMO, of course) weakly your posts fit the mold of "saying anything and tailoring all beliefs to favor nVidia no matter what". This is more than just bias, it is applying the bias to seemingly everything you write. You don't get emotional or "loud" in doing it, but the volume (to me) goes a long way towards making up for it.
It seems to me that your comments fit this criteria moreso than an objective observation about image quality, and that is basically the only real concern I have with your outlook...in other words, if someone rejects my view of your posting style and "bias", they can consider this a post "defending" your viewpoint.

 

[Fuz]

First, let me say sorry to the modem users on here. Sorry
This coming from a gamer’s point of view. I don't pretend to know how ATI's method differs from nvidia's, I just see what my eyes are looking at.
Also, I don't really care what brand hardware I have driving my games, and I hope the regulars on these boards will attest to that


I think its best to compare the pics side by side.

Here is the GF4 pic.

And now the 9700 pic.

In terms of Aniso quality, which one looks better? Do we all agree? Good! If you don't, then you need glasses!

Now, the angled shots.

GF4

9700

Which looks better now? Pretty hard to tell the dif isn't it?
So here we have situation when the 9700's "flawed" aniso kicks in, it looks like the GF4's "proper" aniso. In other words, ATI's worst case scenario is the same as GF4's best case scenario.

Yet, we still have people complaining about ATI's "poor" aniso implementation. Makes you wonder.

Fuz

 

[Nagorak]

Ok, in the second set of shots I agree the floor looks more or less the same. But the wall on the 9700 still looks way better. And, for that matter, the end of the hallway looks like a blurry mess on the GF4 and much clearer on the R9700. In fact everything in the picture looks like a blurry mess on the GF4 (look at the pillars at the top left, etc). How can anyone claim the GF4 has image quality anywhere near the R9700????

Anyway let's concentrate on the floor, it looks exactly the same on both. So R9700 Anisotropic is no worse than GF4... Meanwhile in the flat shots (and honestly how often do you play with the floor at a 45 degree angle) the R9700 looks way better.

So I more or less agree with Fuz...the R9700 always looks better or the same! In no situation does it look worse than the GF4...

So how can anyone complain????

 

[DemoCoder]

uh, I don't agree that the angled shots look the same. I agree with your overall point, but the 9700 angled pics near the far wall on the ground look alot blurrier, as I can see the "crack" in the stone wash out alot earlier in the 9700 pic.

The 9700 angled shot also seems to have a "harsh" transistion, almost like a mip-banding boundary between the acceptably sharp section, and the farthest section that is all washed out with no detail.

Finally, on the "flat" shots (the GF4 one is slightly angled), why does it look like the GF4 version has AA off, but the 9700 has AA on? Were the AA settings and LOD settings identical?

The real thing I'd worry about is how it looks in motion and whether or nor that badly filtered rear section in the angled 9700 pick looks like a visible mip-band sliding over the landscape as you walked forward. I hate visible banding. The blurry section wouldn't really be a noticable problem unless it became like a stable rectangle section that floated on the ground, like the ole mip-bands in the pre-trilinear days.

Ah well, I'l be able to tell soon as my 9700 PRO will be arriving.

 

[darkblu]

I think its best to compare the pics side by side.

<snip>

Which looks better now? Pretty hard to tell the dif isn't it?
So here we have situation when the 9700's "flawed" aniso kicks in, it looks like the GF4's "proper" aniso. In other words, ATI's worst case scenario is the same as GF4's best case scenario.

Yet, we still have people complaining about ATI's "poor" aniso implementation. Makes you wonder.

Fuz,
for the sake of subjective/gamer's viewpoint/whatever-you-name it argument, let me note that the side-by-side comparision of the rotated-case shots shows it quite clearly for me that on the 9700 the final 1-1.5 cm of the floor span look clearly worse than the same span of the floor on the gf, i don't see what's so "hard to tell" here (read: it's not hard at all for me). and if i'm right to assume that on those shots the gf is 8x aniso and the r300 is 16x aniso (right?) that indeed could make one wonder why people would complain about it..

ed: looks like DemoCoder didn't find it "hard to tell" either.

 

[Dave Baumann]

Fuz - if you're really bored today try typing in tex_bColorizeMipmaps=1 in the SS command console and take those pics again (and also a plain bilinear shot); again, only if you really have nothing better to do!

 

[Basic]

It would help if the shot was taken from the same spot. The 9700 shots are taken one "stone pilar" further ahead.
Nice way to fake the comparison.

 

[Galilee]

Fuz: Those pictures use different level of aniso right? Max on both?
Strange to compare 8 and 16-level.

FSAA on off? Same resolution?

The GF picture looks brighter on the gun.

The angle is much bigger on the GF picture. Radeon picture looks like 30-35degrees or something.

 

[Galilee]

I'd be glad to provide some GF4 aniso screenies in SSam : SE but it appears that aniso doesn't work (on my WinXP/40.41 system, at least) in SSam : SE thru either in-game-aniso-option or NV OGL control panel with my GF4Ti4600... any GF4Ti owners with the same drivers (40.41) having the same problem?

For me it worked with in-game control, but not from the control-panel.

 

[Fuz]

Let me clarify a few things. I didn't "fake" the comparison. Infact the pictures were posted by Donald and tEd on this thread. I just thought it would be better to have the images side by side to compare.

I should have stated that before.

To my undestanding, both cards were set to max aniso settings. Not sure about resolution, and not sure about AA settings either. I understand that comparisons like this is not exactly the best way to compare quality.

I should have thought about these things before I posted.

I will try to post some imges that are from the exact same position, same res, same AA and max aniso.

Fuz

 

[Galilee]

Test 8X aniso on both.

 

[Dave Baumann]

Test 8X aniso on both.

Personally I find the maximum each has to offer as a valid test - so be it that that happens to be 16x for one and 8x for another. Honestly, given the performance who would use 8X on 9700 anyway?

 

[Galilee]

Hehe, sure. But it is a little funny when people say "WOW easy to see what card has the best implementation", when the pictures show 8X vs 16X

You could also make a pretty strong argumentation that 6X FSAA(R300) looks better than 4X FSAA on GF4, but it doesnt way very much about the quality of the implementation.

 

[Mize]

All I can say is this thread is going beyond nitpicking. I looked at both the R9700 and the comparison shots from the GF4 and I still can't see any noticeable difference between them at all.

I disagree. Look at the far wall in both shots; the R9700 is much more detailed. Coupled with the comparable aniso at 45 degrees and the R9700 is the clearly superior rendering.

Score ATi.

Mize

 

[Sabastian]

All I can say is this thread is going beyond nitpicking. I looked at both the R9700 and the comparison shots from the GF4 and I still can't see any noticeable difference between them at all.

I disagree. Look at the far wall in both shots; the R9700 is much more detailed. Coupled with the comparable aniso at 45 degrees and the R9700 is the clearly superior rendering.

Score ATi.

Mize

I was just about to make the exact same observation. Even in the worst case scenerio the Radeon 9700 has superior IQ compared to the Geforce 4 ti. Its as plain as day.

 

[tEd]

Fuz - if you're really bored today try typing in tex_bColorizeMipmaps=1 in the SS command console and take those pics again (and also a plain bilinear shot); again, only if you really have nothing better to do!

unfortunately colormipmaps doesn't work on the floor

 

[Oratius]

I think that before going ahead in the discussion we should clarify what we want to discuss: which is the card that offers the better implementation of AF (and in this case I agree with Galilee in his idea of comparing the cards in the same condition) or which is the card that offer the best IQ.
In my opinion the second choice would be better from a gamer point of view, who looks at IQ he can obtain and not the way it'd obtained.

 

[Dave Baumann]

Hehe, sure. But it is a little funny when people say "WOW easy to see what card has the best implementation", when the pictures show 8X vs 16X

Errrm, if one is limited by the hardware and the others highest setting offers at least as good quality as that setting (hypothetical, not based off this thread) which would you say was the best implemtation?

Performance can also be a factor in deciding that, but we’ve got such an apples to oranges comparison here because 9700’s Aniso performance is substantially higher than GF4 and the difference between the Aniso levels on 9700 is so minimal.