In lighting, always normalize zero to (0, 0, 1) #14231
Conversation
|
I wonder if that SSE4 path is really worth it. But probably doesn't matter much. |
|
Well, it was already there from #11425, I just figured might as well use it more places since it's there anyway. -[Unknown] |
|
Yeah, I know, just wondering out loud. |
| @@ -1056,7 +1062,7 @@ bool GenerateVertexShader(const VShaderID &id, char *buffer, const ShaderLanguag | |||
| break; | |||
| case GE_PROJMAP_NORMALIZED_NORMAL: // Use normalized transformed normal as source | |||
| if (hasNormal) | |||
| temp_tc = flipNormal ? "vec4(normalize(-normal), 1.0)" : "vec4(normalize(normal), 1.0)"; | |||
| temp_tc = StringFromFormat("length(%snormal) == 0.0 ? vec4(0.0, 0.0, 1.0, 1.0) : vec4(normalize(%snormal), 1.0)", flipNormal ? "-" : ""); | |||
There was a problem hiding this comment.
@unknownbrackets I think the bug is here. Two %s but one parameter.
You don't really need the sign in the length check anyway...
There was a problem hiding this comment.
Oops, I don't think I intended to have it twice. Bah.
-[Unknown]
There was a problem hiding this comment.
It's kinda ridiculous that this compiles. (I mean, I know why, but ... )
There was a problem hiding this comment.
Makes me miss D where this wouldn't. I'm sure rust has something similar.
-[Unknown]
There was a problem hiding this comment.
Indeed, this would break with Rust formatting macros.
Fixes #14167. We've seen several cases of this now, so I think it's better to just assume it's the case for all normalize usage.
-[Unknown]