Am 08.02.2022 um 16:41 schrieb Henri Verbeet <hverbeet(a)gmail.com>:
On Sun, 6 Feb 2022 at 18:15, Stefan Dösinger <stefan(a)codeweavers.com> wrote:
+ /* SPECULARENABLE shouldn't matter in this test, but WARP begs to + * differ. In the event that color keying is randomly on(see comments + * in test_ck_default for reference), WARP will randomly discard + * geometry based on something, even though texture anddiffuse color + * alpha components are non - zero.Setting SPECULARENABLE to FALSE + * prevents this in some cases - presumably WARP multiplies the + * specular color "alpha" channel into the final result and then + * alpha tests the result.Since the specular property normally does + * not have an alpha component the actual specular color we set in + * the vertex data above does not matter. */ + emit_set_rs(&ptr, D3DRENDERSTATE_SPECULARENABLE, FALSE); + I'm not quite sure what happened above, but that comment is full of white-space errors.
Ah, I typed that thing in notepad, copypasted it into Visual Studio later. Either my butterfly keyboard is going bad again or something decides to be smart about whitespace when copypasting...