In a message dated 98-05-06 04:53:36 EDT you write:
> When I read about the "project and rubber-stretch image data on facetted
> geometry" - idea, I feel uncomfortable.
> The problem is like in 3D scanning - you have to solve the problem for
> all geometrical cases, including undercuts, hollow parts, complicated
> technical geometry or even medical files derived from CT data.
> You are only able to apply this technique in a very limited region where
> the surface is convex, i.e. one bitmap for each STL triangle.
> What do you think ?
I'm just exploring possibilities which might allow efficient 3D printing of
colored computer hardcopy. I certainly hope there is something which has the
theoretical appeal of a voxel-like 3D bitmap with customer determined
attributes for each point in space - but with the efficiency and flexibility
to work for a variety of purposes, etc.
Short of such an ideal, it seems that we must compromise to find ways to make
progress. I'm not saying this is THE answer, but it's certainly possible to
communicate all the external colored information on a physical part with a
number of "views" or "planar projections." The particular views which are
required would be determined by the shape. Six would be a typical minimum and
would be capable of handling convex and concave but would need supplemental
views for "hidden" areas. The number of views would be a "project driven"
The "rubber sheeting" I proposed would be used only to the degree a customer
desired its efficiencies. It just doesn't seem to me that it would be
absolutely necessary to have each facet individually "bit mapped" (couldn't
that result in millions of "views"?) In fact, it would be highly desireable
in many cases, to be able to change the facets (generally an approximation of
shape) and not to have to worry about how the color information precisely
For better of worse, the requirements of "visualization" might just persuade a
portion of this industry to utilize a 3D equivalent of some of the "tricks" of
computer graphics. By this I mean, using "procedures" which accomplish the
customer's needs with the most efficiency. This shouldn't cause a problem -
as a segment of a more diversified industry.
Any constructive suggestions? Slicing CAD data with surface info? internal
info? Slicing FEA models?
For more information about the rp-ml, see http://ltk.hut.fi/rp-ml/
This archive was generated by hypermail 2.1.2 : Tue Jun 05 2001 - 22:45:31 EEST