I was recently approached by a guy with a Horner Brothers table.
The table was missing the left wings, and he needed it reproduced. He did have a right wing. I used 123D catch to generate the mesh and made the cam paths off of that. Is there a more accurate way to generate mesh? You can see the detail isn’t quite good enough.
Low end scanners and software will not come close to good enough to make the effort worthwhile. Even with a high end scanner, the clean up is going to take some time. Multi-spectral scanning leads to the best results fastest.
Another approach is a “touch probe” - mechanical scanning - to generate a “point cloud” and an STL from that. Here is one example:
I know you can save the probe data from bcnc, but have never tried converting that to an stl. Have you done that before?
I used the NextGen software. My CAD program can handle point cloud conversion to mesh and on to an STL file. I would check your CAD program or look for software; there are many.
If you have a good DSLR or mirrorless camera with high-quality glass (minimal lens distortion and abnormalities) you could try doing photogrammetry using VisualSFM and then reconstruct a mesh with CloudCompare to run your CAM on.
It’s a free workflow if you have the camera and computing hardware, and you can get some really great results, but it’s neither easy nor obvious, so get ready to dig in and learn a lot of new stuff to do it.
Here’s an example scan-reconstruction (with color):