Keith Whitwell wrote: > > Michael wrote: > > > > On Wed, Apr 03, 2002 at 10:59:18PM +0100, Michael wrote: > > > On Wed, Apr 03, 2002 at 10:11:36AM -0800, Ian Romanick wrote: > > > > On Wed, Apr 03, 2002 at 01:28:46PM +0200, Timothee Besset wrote: > > > > > Not sure if that's related, but Q3 shaders don't do any bump mapping. > > > > > > > > > > TTimo > > > > > > > > Hrm..ok, so then it's not the bumpmaps that are missing. :) In any case, > > > > the walls look perfectly smooth, and they didn't used to. I'll send a > > > > picture out to the list, if there is demand. > > > > > > Looks like a few of the wall textures are magnified a few times, like the > > > texture coords are wrong. > > > > > > rtcw victory screen has the rolling texture thing (which is no doubt > > > texcoord related as well) > > > > It looks like varray.c doesn't correctly set ctx->Array.NewState in each > > function (it's always setting _NEW_ARRAY_VERTEX), the reason anything > > works seems to be that (Un)LockArrays_ sets _NEW_ARRAY_ALL, so it's > > usually 0x7fff by the time that state reaches the new > > radeonReleaseArrays. Currently, ReleaseArrays sometimes gets duff state and > > was leaving old tex coords in dma for some of the multipass textures > > quake3 is drawing. > > > > I'm testing my change here a bit wider before committing, but it looks > > like changing each <xxx>Pointer function in varray.c to use the > > corresponding _NEW_ARRAY_xxx state works. > > All my cut&pastes come back to haunt me.
I'll duplicate the fix in Mesa CVS, etc. Nice catch, Michael! -Brian _______________________________________________ Dri-devel mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/dri-devel