Hi all On Thu, 2011-01-20 at 22:17 +0200, David Griffiths wrote: > > Both are 750x160 .png files (like the one already send). One glitches, > > the other doesn't. I don't know enough about the anatomy of .png files > > to say anything sensible about this but I do know that this means > > there is some non-trivial weirdness here or at the very least that > > dimensions in pixels aren't the only thing that affects whether or not > > a given texture displays correctly. > > Just a quick guess - does one have alpha and the other not?
Oh yes, you attached the textures to the bug report :) Yeah, I think that could be the problem. The annoying thing is that it looks like it's passing the right thing to OpenGL (desc.InternalFormat, TexturePainter.cpp, lines 266 and 272). I'll have a closer look. Also, on the power of two issue - my thinking on this was that if you were generating mipmaps (the default in fluxus) gluBuild2DMipmaps[1] should take care of this for you by scaling up/down and correcting. If you have mipmap generation off then it's going a much lower level route, so you are on your own with these calculations. All of this gloriously missing from the documentation I suspect... cheers, dave [1] http://www.opengl.org/sdk/docs/man/xhtml/gluBuild2DMipmaps.xml
