hi just a question do pixelshaders support bitwise operators?

Here is Microsoft''s documentation, listing all of the instructions supported by each version of pixel shaders. Just in glancing through them, I didn''t notice any bitwise operators.

Shaders have 3 possible data types:

floats
bools
ints

ints are likely emulated using floats.

data can be (and is usually optimally) accessed as a vector, that is to say you can declare float2, float3, or float4 (or even float2x2, float2x3, float4x4, etc.).

bitwise operators would probably be nice, but in general most hardware would cry "mamma" if you tried to do enough branching as to require a bitfield.

I thought about BOOLMAPPING static scenes with them

a BOOLMAP is a shadowmap indicating which pixels are shadowed and which arent

so lets say we pass the bool maps without filtering and without mipmapping then you could theoretically access the Unsigned byte values in the pixel shader via a integer and check which pixels are shadowed and which not and accordingly apply ambient light and other lighting calculations

the advantage is that you can use very highres BOOLMAPs to create smooth shadows on the geometry

you could even interpolate between the shadowed and unshadowed pixels by altering the u,v coordinates inside the shader


But this is impossible to do when those damn pixelshaders don t support bitwise operators, since all the bool values shall be encoded horizontally into unsigned byte values so your BOOL map is actually [width/8][height]

this is just theory but i think it s worth the discussion imagine how smooth static shadows one could create
no more stepping effects
you could even implement a axisaligned interpolation and a 45° rotated grid interpolation within the shader

yes i know but i thought about this feature to render my static sceens with high resolution shadows which are precalculated but require very less memory

that s why i thought about this system but good to know that they at least plan it