Simple question, but one I'm not really qualified to answer.
And, if the answer is "no", then... why not? This is a huge performance-enhancing area, and if we're going to make more extensive use of shaders in the future, then it's downright vital.
I know I saw a fairly large improvement in performance in NanoBlobs, when I went over to DDS. That's not even arguable. But, if the engine isn't using mipmaps, then I'm wasting fairly-significant amounts of texture RAM in my current game... and frankly, I think it's more important to fix this and make it work, than to avoid it, if we're going to use the Very Fancy Stuff that's coming, due to Trepan's work.
Mipmaps. Do they get used by Spring?
Moderator: Moderators
Mipmaps aren't for improving performance, they're for improving the look of textures that are further away by reducing the number of texels per pixel and applying proper blurring. AFAIK mipmaps actually decrease performance and increase the memory footprint of a non-alpha texture by 1/3rd (not sure about alpha,we didn't have that in the CG course at uni).
I thought they improved performance, by lowering the size of the texture that was being wrapped around the model during a rendering pass...
At any rate, if Spring doesn't use them, and will not be using them, that's fine, but it means I should just take 'em out of all of the DDS files in my game (yay).
At any rate, if Spring doesn't use them, and will not be using them, that's fine, but it means I should just take 'em out of all of the DDS files in my game (yay).
Almost the same 
(1/4)^1+(1/4)^2+(1/4)^3+..+(1/4)^n...=1/3
Except for non-power-of-two textures, nvidia supports mipmapping on those and that increases the memory footprint with 1/2 i think.

It doesn't really matter which format, its just math:increase the memory footprint of a non-alpha texture by 1/3rd (not sure about alpha,we didn't have that in the CG course at uni).
(1/4)^1+(1/4)^2+(1/4)^3+..+(1/4)^n...=1/3
Except for non-power-of-two textures, nvidia supports mipmapping on those and that increases the memory footprint with 1/2 i think.
I wasn't sure about the alpha handling since the way it was explained to us was that it makes an 8-bit image of 2x*2y pixels with three quadrants being the color channels and the fourth the mipmaps (with each mipmap having three color quadrants and one quadrant for further mipmaps), that wouldn't fit into a square with alpha added so I wasn't sure if it won't leave some blank room like a filesystem would with its blocks.jcnossen wrote:Almost the same
It doesn't really matter which format, its just math:increase the memory footprint of a non-alpha texture by 1/3rd (not sure about alpha,we didn't have that in the CG course at uni).
(1/4)^1+(1/4)^2+(1/4)^3+..+(1/4)^n...=1/3
Except for non-power-of-two textures, nvidia supports mipmapping on those and that increases the memory footprint with 1/2 i think.