constant bias scale just works by
y = (input + bias) * scale
to convert [0,1] (RGBA8 texture values) into a world space offset in the range -128, 128] you would use a bias of -.5 and a scale of 256
constant bias scale just works by
y = (input + bias) * scale
to convert [0,1] (RGBA8 texture values) into a world space offset in the range -128, 128] you would use a bias of -.5 and a scale of 256