I am trying to access a regular RGBA32 texture in a compute shader, and wondering if there is a way to get the data as 4 bytes. I tried used “fixed4” in the compute shader, but it doesn’t seem to recognize the keyword.
If I declare the texture as
RWTexture2DArray<float4> colorTexture;
it will compile to
writeonly layout(binding=0, rgba32f) highp uniform image2DArray colorTexture;
And if I declare it as:
RWTexture2DArray<half4> colorTexture;
it will compile to
writeonly layout(binding=0, rgba16f) mediump uniform image2DArray colorTexture;
However, if I declare it as:
RWTexture2DArray<fixed4> colorTexture;
then I simply get a compile error.
I would ideally like it to be defined in the glsl compute shader code as “rgba8” or “rgba8_snorm” because I am worried about extra bandwidth or conversion costs of defining it as a higher precision.
I haven’t been able to find any examples or documentation about whether this is possible, or whether it is something I shouldn’t be concerned about.
Thanks!
