On 6 September 2011 13:40, Emmanuele Bassi <[email protected]> wrote: > oh, you wanted *actual* proper feedback? ;-)
Yes! :-) > how is the shader approach structured? Well, at the moment the test code is a fragment shader which is attached to each actor using clutter_actor_set_shader (which I notice is now deprecated). I was going to wrap all the color stuff up in a GcmIccEffect and just add an effect to each actor. The question then becomes what to do with an actor that straddles two outputs in a multhead setup. Logically it makes sense to cut the actor up and use the correct screen profile on each part (using a 1bit bitmask or a uniform xoffset in the shader perhaps?) but it might be simpler just to choose the output profile that the actor most covers. As for masking out areas that already have color correction, I'm not sure. I'm tempted to just opt-out whole actors at the moment, although from a mutter point of view that means you don't get color corrected toolbars and that kind of thing. This can be quite a gradual thing, we don't have to do everything at once as long as we've got an good opt-out story for the Pixars and Dreamworks of this world. Richard. p.s. the test code I'm using is http://git.gnome.org/browse/gnome-color-manager/tree/tools/gcm-glsl-demo.c although this will change quite a lot in the next few weeks. _______________________________________________ desktop-devel-list mailing list [email protected] http://mail.gnome.org/mailman/listinfo/desktop-devel-list
