Quincunx & FSAA: how is it changed between Geforces

[Jerry Cornelius]

That's an interesting idea, just point sample at a ridiculously high res and down sample. You wouldn't have any anisotropy in a sceme like that tho and you're frame/buffer buffer bandwidth requirements would be high.

Eventually I think TMUs will be programable to take a patch of samples instead of just blending the four nearest.

16 or more samples, 256 possible patterns and throw the weighted blending out the window.

You could make the density of the sample patch a variable depending on the z value and do away with mip levels as well. Sort of "sparse" texture filtering for textures on far away polygons. It would be hell on memory, but should avoid a lot of texture aliasing.

 

[KimB]

What form of texture filtering reduces aliasing very well?
Aniso does a so-so job at it, but is more designed to make the texture clearer at angles and further in the distance than it is designed to reduce aliasing.
Aniso does a poor job of decreasing texture aliasing, imo. It does make textures look better (esp far away textures)

I do know of one...I believe it was the SGIS_Filter4 extension, which always uses a 4x4 texel grid (as opposed to a 2x2 for normal bilinear) when filtering a texture. Such filtering methods, especially when used in conjunction with anisotropic and MSAA, could provide the same image quality as SSAA, but with higher efficiency.

Regardless, I doubt that many people would like the improvement in reduced texture aliasing that such higher-order filtering methods would bring, given the significant performance hit (Except, of course, in older games where the fillrate would otherwise go unused...).

Yes, it is very true that provided that the texture LOD is adjusted accordingly, good anisotropic should not affect texture aliasing at all. It just improves the filtering in certain places, allowing more detailed MIP maps to be used.