On Mon, Dec 16, 2002 at 01:04:18PM -0800, Andy Ross wrote: > Keith Whitwell wrote: > > Andy Ross wrote: > > > Is it possible [...] to get the client-side libGL to look somethere > > > *other* that /usr/X11R6/lib/modules/dri for drivers? > > > > It's LIBGL_DRIVERS_PATH > > Perfect. Thanks. > > > > You also don't exhibit a texture border bug that the ATI drivers > > > have > > > > We fallback on some texture border cases. > > Not this one, I suspect. The runway markings are built up of tiled > quads, drawing the whole texture into each quad. This makes up > probably 20-30% of the screen pixels at startup; if you were rendering > them in software I'm sure it would be noticeable. > > Actually, I'm fuzzy on the spec here. The DRI (and NVidia) drivers > appear to *not* sample the texture border color by default when > drawing quads with texture coordinates of exactly 1 and 0. The ATI > ones seem to do so, which results in a dark stripe between tiles > (FlightGear doesn't bother to set the border color and leaves it as > black). Problem is, my cursory reading of the spec says that ATI is > right and the rest of the world (!) is wrong, even though that would > make life hell for people who simply want to build up tiled images and > don't care (much) about border artifacts. What's the truth?
Software Mesa does this (correct) behavior too. You're supposed to either use GL_CLAMP_TO_EDGE or a texture border if you don't want the border color to be factored into the texture filtering. GL_CLAMP_TO_EDGE is a lot easier to implement in hardware (and it follows the semantics of DirectX teture clamping), though, so I'd suspect that most GL drivers cheat and just do CLAMP_TO_EDGE instead. Personally, I don't know why SGI put in the texture border color to begin with... it seems horribly counterintuitive, and not all that useful. -- http://trikuare.cx ------------------------------------------------------- This sf.net email is sponsored by: With Great Power, Comes Great Responsibility Learn to use your power at OSDN's High Performance Computing Channel http://hpc.devchannel.org/ _______________________________________________ Dri-devel mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/dri-devel
