diff options
author | Søren Sandmann Pedersen <ssp@redhat.com> | 2013-02-03 12:57:54 -0500 |
---|---|---|
committer | Søren Sandmann Pedersen <ssp@redhat.com> | 2013-02-15 18:04:23 -0500 |
commit | 3f8e03240a6f194eb51f834a909f3c45964654b3 (patch) | |
tree | 1e8e322d72df96789aa89237efac2e9623b586eb | |
parent | 3c7f5c8af89386f354729de004dbc72fc185444f (diff) |
roundingdither
-rw-r--r-- | pixman/pixman-bits-image.c | 7 |
1 files changed, 7 insertions, 0 deletions
diff --git a/pixman/pixman-bits-image.c b/pixman/pixman-bits-image.c index 372c0b2f..cbf2178a 100644 --- a/pixman/pixman-bits-image.c +++ b/pixman/pixman-bits-image.c @@ -2598,6 +2598,13 @@ dest_write_back_narrow (pixman_iter_t *iter) * deeper. For example, compare: #142b3d -> #457262. Or * #000000 -> #39213e. Gimp renders these without banding, but when * evaluated in 8 bits, they look ugly. + * + * - When converting the dither signal to the correct number of bits, + * that should ideally be done with rounding, not bitshifting. + * With that, the 565 gradient becomes close to indistinguishable + * from the 8888 one. Note that one bug that can happen is a wavy + * colored pattern; I think this happens if the algorithm for g is + * different from the one for r and b. */ if (image->common.dither == PIXMAN_DITHER_ENABLE) { |