Feature Request Add support for byte type for Sample Buffer

In the official documentation, we can see the list of types compatible for sampling structured buffers.

Do you think it is possible to add support type byte ? I would like to minimize the RAM usage in my data, data that can be encoded in 8 bits and not 32. I work a lot with the new [VFXType] attribute, I would like this to be possible with custom types as well.

 

VFX graph generates compute shaders, and the byte type doesn't actually exist in shaders. If you check HLSL and GLSL docs, you'll see the smallest type is 32-bit float/integer, with some GPUs supporting 16-bit floats (but not 16-bit int).

But you can use bytes in the GPU, just not directly. In RGBA8 textures, for example, each channel is stored using just one byte, but when you load/sample from it in a shader the GPU converts it into a 0-1 float. If you want to use structured buffers, however, the only choice is to manually pack/unpack bits using bitwise operations.

 

I will use bitwise operations for the moment. Thank you for your feedback