Aniso becoming less useful?

[RussSchultz]

Hmmm...

[sarcasm]
It looks like nVidia is "cheating" here by not applying aniso to all texture stages. And ACK....disabling trilinear filtering? Surely, no self-respecting GeForce owner would enable such a thing, no matter what the performance boost. Can't nVidia engineer an anisotropic engine that is fast without having to resort to implementing all of these cheats and hacks? Man, that implementation sucks!

[/sarcasm]

So what happens if this treatment of anisotropic filtering gets put into the driver to improve performance?

That would suck, unless you could turn the optimization off. Or, better yet, it defaulted to disabled, but was enableable.

Otherwise, you have all the reviewers reviewing it in 'default mode' and performing higher than it would then in its useable state.

 

[Joe DeFuria]

wow, look at us, our sense of irony is so refined that we use SARCASM html tags!

Yeah, isn't that the truth!

So what happens if this treatment of anisotropic filtering gets put into the driver to improve performance?

There will unfortuntely be another round of debate and flip-flopping on the stance of "correctness" vs "performance" and accusations of "cheating" vs. "legitimate" optimization, etc. All the sudden nVidia fans will be talking about the great "improvements" in AA performance with "no noticable impact" on image quality even with the loss of trilinear and failure to aniso all textures in all cases. All the sudden ATI fans will be talking about how nVidia has to resort to "driver tricks" by not "anisoing everything" to cheat and get better performance at the expense of image quality.

Unfortunately, what probably won't happen, is someone doing an in depth view of the end result: the objective and subjective views of overall impact on image quality and performance.

 

[RussSchultz]

Unfortunately, what probably won't happen, is someone doing an in depth view of the end result: the objective and subjective views of overall impact on image quality and performance.

Actually, I think if you look up a few pages in this thread, somebody is working on exactly that. (It was mentioned in Alexsok's post)

 

[Galilee]

From my understanding of most NVIDIA owners they would like NVIDIA (or Rivatuner) to have "performance aniso" as an option. While some implementations of "performance aniso" can have some negative effect on the image quality I would very much like to have it as an alternative to the normal quality anisotrophic filtering.
If NVIDIA implements this kind of aniso I am pretty sure it's as an alternative, or else I would be dissappointed.

About the bilinear issue, it all depends on the image quality. It is here talk about maybe enabling bilinear on stage 1, 2 or 3. That would not mess up trilinear filtering on most textures. But the tweak I use as default now doesnt use bilinear, I think maybe trilinear filtering is THE most important part of this whole debate. I would rather have trilinear than a lot of aniso without trilinear, but that is just me.

 

[Joe DeFuria]

Actually, I think if you look up a few pages in this thread, somebody is working on exactly that.

I am indeed interested to see that "review." However "working on a comparison article" does not indicate the quality of the article itself....

From my understanding of most NVIDIA owners they would like NVIDIA (or Rivatuner) to have "performance aniso" as an option...

Yes, that's the difference between nVidia "owners" and nVidia "fans." Rational owners will tell you they want options in all cases....fans will tell you that any competing alternative option is bad...until that point in time where a similar option is made available to them.

As an Radeon 8500 owner, I have said many times that I would like to have the option for "quality" anisotropic with Radeon's drivers. However, if I had to be stuck with one, I'd prefer performance. I play primarily FPS games...if I were a flight sim guy my preference might be the other way.

 

[Galilee]

Who cares about fanatic fans anyway

(and I am with you. I would rather have performance than quality because I also mostly play fps-games. Therefor I am very happy with this initiative from the Rivatuner guy)

 

[Ailuros]

Joe,

On the flip side of things anisotropic filtering is not even now an option in D3D, and all you get in openGL is quality (default) aniso settings from 2x to 8x.

I have no idea if NV is working on those optimisations or will adopt them, but currently what Unwinder has done with Rivatuner, leaves the user more than enough available options to tune aniso as a user's heart may desire.

I'm not really crazy about the texturing stages optimisations that got in there lately, but the openGL performance/quality switch was a very nice extra. To the best of my observations so far one of the most significant differences between those two options is that quality filters lightmaps too, which is pretty unnessecary IMO (or for my imagery if you prefer).

What fans from either sides concerns: blah. Up until now it was the superior AA + cheat aniso vs. inferior AA + quality aniso argument. I don't believe that NV will sustain the same aniso algorithm in their next batch, so the message to either sides would be that to each algorithm/implementation there can be downsides, but for heaven's sake look at the positives first, before you call it "useless" or a "cheat".

edit: same of course goes for Multisampling too.

 

[Nagorak]

[sarcasm]
wow, look at us, our sense of irony is so refined that we use SARCASM html tags!
[/sarcasm]

Considering most people seem to have trouble detecting sarcasm, you can't really blame him.

 

[K.I.L.E.R]

I play on my G3 with 8x aniso alone and it runs fast at 800x600 32bpp so my G3 must be beating his G4 in aniso performance alone. (When I mean very fast I mean around 30-35fps lolz)

GF4 cards has a serious problem with anisotropic filtering.
The 2nd texture unit is disabled when AF is enabled.
nVidia probably simplified that 2nd TMU so it's not capable of AF.

I wonder if GF3 cards could beat Gf4s in these games with AF.

Might I add disabling aniso gave me next to nothing performance gain. Oh well, I must be pretty crap.

 

[Ailuros]

Might I add disabling aniso gave me next to nothing performance gain. Oh well, I must be pretty crap.

That means that there's some other bottleneck that comes forward through the application you use and not that you get anisotropic for free.

 

[K.I.L.E.R]

Maybe. CPU prehaps?