On 8/4/19 2:55 PM, Kima wrote:
> Consider a simple cube as
> 
> I have two problems:
> 
> 1. How can I keep the pigment color underneath to have a closed where has not
> been wrapped by the image?
> 
> 2. Wrapping starts from the left corner of the object, and I lose the curvature
> on the left side, but it is OK on the right side. According to
> http://www.f-lohmueller.de/pov_tut/backgrnd/p_sky9.htm the image covers +Z and
> goes to +X. Is it possible to put the image on the center of +Z to let the
> excess image covers both -X and +X?
> 

If I'm understanding your questions,

With (1) the object pattern is perhaps one way where you use another 
unseen/defined-only object along with a second solid color texture to 
limit where the image_map texture is applied.

With (2) move the image_map using translate <-0.5,-0.5,0>. You'll likely 
still not get the ideal map to the curvature for the superellipsoid at 
the corners. I don't know how to easily correct for that distortion.

Aside: For some shapes there is the shape uv_mapping mechanism(1), but 
as I recall the superellipsoid isn't supported. Further, there are today 
general issues with it even for shapes where it is implemented.

A question bouncing around in my head is whether with the new 
user_defined camera there isn't some way we can implement a more general 
3D point to shape mapping mechanism. I have a crude version of the idea 
going in my density file patch branch. It's how the warps of the posted 
maze images warped both the iso-surface shape and it's texture(s) 
together. Not sure though how to do the point to surface association and 
interpolation generally for any shape including csg - but that's sort of 
the thought.

Bill P.

(1) - Not here speaking of the image_map or warp type mapping.

Post a reply to this message