Planning and Partitioning

To efficiently use point cloud manipulation software, a thorough understanding of surface creation and the final modeling package is vital. Several scanning hardware manufacturers offer scanners that accurately digitize amorphous shapes. This allows accurate capture of ergonomic design that interfaces with humans, such as helmets, orthopedic braces, and prosthetic devices. Often, manufacturing software can be created for CNC milling right from scan data or from an STL file without taking the extra step of producing a surface model. This means that a prototype can be made and approved, scanned, and a mold made of any proportion quickly and easily. The entire process takes only a matter of days.

Because some objects cannot be scanned in a single pass, it is important to intelligently partition the object into small, discrete sections. After scanning, the discrete sections can be combined electronically into a model representative of the whole project. Surfacing tools are not able to handle overly complex scan data, so it is important to divide the entire project into simple patches that overlap and can be easily scanned and surfaced.

The contoured surface in the adjacent figure is partitioned into seven patches. Notice the simplicity of each patch. Each patch benefits by scanning technique unique to its geometry. The keys to dividing up the project are:
  • Separate Discontinuities in the Surface: Notice that regions 1 and 7 are separated by region 6. This is because region 6 would be considered a discontinuity if regions 1, 6, and 7 were combined.
  • Separate Discontinuities in the Surface Edges: Notice the unique definition of regions 1, 2, 3, 4, and 5. They are separated because, if any two of them were combined, there would be a discontinuity in their edges (a corner) or the trim line of the resulting surface. This is because even powerful surface modelers have trouble with edge discrimination.
  • Scan Each Region Meticulously: This is one of the most important steps. Often it is helpful to have a picture available and use the "coloring in" method. In the earliest implementation of this technique, this is when an operator watches the graphical interface while scanning and fills in the gaps seen on the screen.

In summary, over scanning and getting too much data is always good. The software can easily filter the data; however, missing data is always bad, and always requires another scan.