Quake3e
Quake3e copied to clipboard
ec-
Solved: cg_shadows 2 doesn't work under vulkan renderer
Hi everyone: I believe some people have the issue that cg_shadows 2 can't renderer properly. Because this issue is triggered by another command which is "r_stencilbits". Just input this command in the console and change the value from 0 to 8. Then the stencil shadows will be back on your screen. By the way, cheers to the author of 1.32e.
0 stencilbits implies pick desktop stencil bits afaik?
Maybe, cause I suspect that Q3's technology is pretty old, the latest hardware can not be detected by its system, then this value has to follow the desktop default one. Also, you can input "com_hunkmegs", and change its value from the default one to 1024, this method can provide at least 1G buffer RAM for Q3 utilisation. By the way, I think cg_shadows 2 might be the experiment that was used by John Carmack for the preparations on his next engine, which is DOOM 3.
0 stencilbits implies no stencil buffer even for opengl renderer, the only solution to this "bug" is to hardcode it always enabled (along with color and depth bits)
Something else: I dunno if it's the same in all renderers, but cg_shadows 2 in Vulkan often yields weird results like shadows in the middle of the air or clipping through floors, far away from where you would expect them. Looks glitched like that.
Something else: I dunno if it's the same in all renderers, but cg_shadows 2 in Vulkan often yields weird results like shadows in the middle of the air or clipping through floors, far away from where you would expect them. Looks glitched like that.
This "glitch" is how it should be. Considering the tech in 1999 before volumetric shadows with dynamic lighting system, Q3 was the experiment of this tech. John Carmack said that without the constrain, the shadows would project on any surfaces, which are on the way of the volumetric lights. This is why you saw those "glitches". But lately, in DOOM 3, this issue had been resolved.
I see. Well, if it's due to a limitation imposed by the engine, there's little to be done about it. Technology wasn't ready yet and it was just supposed to be a preview for what would be possible later on.
Just M2C and Trivia about the "weird results" mentioned above: The fact that cg_shadows has got "cg" prefix makes me think much of it is managed by Client Gamecode, so probably one would have to modify the QVM code to fix it, something which is outside from Q3e scope which is just engine, I guess. AFAIK, cg_shadows >1 behavior has always been buggy. IIRC, to avoid the risk of exploiting buggy "through wall" shadows to discover enemy presence, OpenArena limited its usage to g_gametype 2 (single player ladder deatchmatch) only.
