summaryrefslogtreecommitdiffstats
diff options
context:
space:
mode:
-rw-r--r--src/mesa/drivers/dri/r200/r200_texstate.c8
1 files changed, 4 insertions, 4 deletions
diff --git a/src/mesa/drivers/dri/r200/r200_texstate.c b/src/mesa/drivers/dri/r200/r200_texstate.c
index b57657c32b4..2eb6962189b 100644
--- a/src/mesa/drivers/dri/r200/r200_texstate.c
+++ b/src/mesa/drivers/dri/r200/r200_texstate.c
@@ -1038,10 +1038,10 @@ static GLboolean r200UpdateTextureEnv( GLcontext *ctx, int unit )
case GL_DOT3_RGBA:
/* DOT3 works differently on R200 than on R100. On R100, just
* setting the DOT3 mode did everything for you. On R200, the
- * driver has to enable the biasing (the -0.5 in the combine
- * equation), and it has add the 4x scale factor. The hardware
- * only supports up to 8x in the post filter, so 2x part of it
- * happens on the inputs going into the combiner.
+ * driver has to enable the biasing and scale in the inputs to
+ * put them in the proper [-1,1] range. This is what the 4x and
+ * the -0.5 in the DOT3 spec do. The post-scale is then set
+ * normally.
*/
RGBshift++;