http://bugs.winehq.org/show_bug.cgi?id=31071
--- Comment #70 from Matteo Bruni matteo.mystral@gmail.com 2012-07-24 10:46:10 CDT --- (In reply to comment #69)
Created attachment 41135 [details] +d3d log with 1st, 2nd (v2) and 3rd patches applied
This patch eliminates original issue. However, i don't see any actual visual difference between applying these 3 patches and reverting commit which causes regression. I think i'll make couple of side-by-side screenshots later to confirm it.
It should be the same indeed. Before the regression commit we weren't testing and enabling the "post-pixelshader blending" feature on some float texture formats, after the patch we are. These new patches I've attached to the bug actually test harder for that and now the test fails, so again removing the feature flag.
Earlier i posted couple of screenshots: http://bugs.winehq.org/attachment.cgi?id=41100 (wine with reverted commit) http://bugs.winehq.org/attachment.cgi?id=41101 (windows)
besides lack of antialiasing, wine version lacks some glow, could it be something connected with this bug which you're trying to fix?
I suppose the game checks the post-pixelshader blending flag to enable those additional glow effects, which, I guess, don't work correctly for the same reason the stricter test in these patches fails. But I still don't understand this fully, so I can't say for sure.