algorithmic modeling for Rhino
I woundering if it would be possible to improve bump preview a bit, I am not sure how this is implemented but it would be nice to make it repsonsive based on the z-buffer, so that if you move closly to where the point is the displacement is smaller... something like this:
vec3 mDir (direction from the vertex point towards camera),
float depth = 256.0 * log2(|mDir|+1) +1
newVertexPos = vertexPos * depth * factor(based on drawing tolerance.. or some other setting)
Does that make sense?
it has to be a single Transform though, it cannot be implemented as a pixel or fragment shader. At the moment the bumping is just some fixed translation + scaling towards the camera in clipping space (which is a normalised space which already contains the current camera transform).
Bumping isn't exactly elegant (and indeed has weird side-effects once you zoom in close) and I'm hoping to get something better going for GH2. But it has to be either a 4x4 transform applied on a per-object basis, or some sort of clever shader which may or may not take the existing z-buffer into account. I'll have to ask Steve Baer about what the possibilities here actually are.