diff options
Diffstat (limited to 'source/blender/windowmanager/intern/wm_draw.c')
| -rw-r--r-- | source/blender/windowmanager/intern/wm_draw.c | 15 |
1 files changed, 14 insertions, 1 deletions
diff --git a/source/blender/windowmanager/intern/wm_draw.c b/source/blender/windowmanager/intern/wm_draw.c index 5dbbf35796f..899c7728af9 100644 --- a/source/blender/windowmanager/intern/wm_draw.c +++ b/source/blender/windowmanager/intern/wm_draw.c @@ -389,6 +389,7 @@ static void wm_draw_triple_free(wmWindow *win) wmDrawTriple *triple= win->drawdata; glDeleteTextures(triple->nx*triple->ny, triple->bind); + MEM_freeN(triple); win->drawdata= NULL; @@ -560,7 +561,8 @@ static void wm_method_draw_triple(bContext *C, wmWindow *win) else { win->drawdata= MEM_callocN(sizeof(wmDrawTriple), "wmDrawTriple"); - if(!wm_triple_gen_textures(win, win->drawdata)) { + if(!wm_triple_gen_textures(win, win->drawdata)) + { wm_draw_triple_fail(C, win); return; } @@ -675,13 +677,24 @@ static int wm_draw_update_test_window(wmWindow *win) static int wm_automatic_draw_method(wmWindow *win) { + /* Ideally all cards would work well with triple buffer, since if it works + well gives the least redraws and is considerably faster at partial redraw + for sculpting or drawing overlapping menus. For typically lower end cards + copy to texture is slow though and so we use overlap instead there. */ + if(win->drawmethod == USER_DRAW_AUTOMATIC) { /* ATI opensource driver is known to be very slow at this */ if(GPU_type_matches(GPU_DEVICE_ATI, GPU_OS_UNIX, GPU_DRIVER_OPENSOURCE)) return USER_DRAW_OVERLAP; +#if 0 + /* also Intel drivers don't work well with this */ + else if(GPU_type_matches(GPU_DEVICE_INTEL, GPU_OS_ANY, GPU_DRIVER_ANY)) + return USER_DRAW_OVERLAP; +#endif /* Windows software driver darkens color on each redraw */ else if(GPU_type_matches(GPU_DEVICE_SOFTWARE, GPU_OS_WIN, GPU_DRIVER_SOFTWARE)) return USER_DRAW_OVERLAP_FLIP; + /* drawing lower color depth again degrades colors each time */ else if(GPU_color_depth() < 24) return USER_DRAW_OVERLAP; else |