Shoelast

[microelly2]

we connect the main part with an independent cap using the seam and tangentsurface objects.
in the video is demonstrated how to split the shoe last into two segments red and blue and reconstruct the gap by seams and tangentsurface blue.
at the moment all segments must have the same number of poles, in our case 17 and all surfaces have degree 3.

phpBB [video]

[microelly2]

bp_673.png (285.08 KiB) Viewed 1888 times

to put the leather patches onto the surface I use geodesics for the borders. so the borders lie on the shoe last and the inner area can bend according to the material forces.

[microelly2]

with parametric geodesics it is possible to see how a rectangular patch is stretched.

phpBB [video]

[microelly2]

putting the leather onto the last:
the network of geodesics shows where the material is stretched or overlapped

bp_679.png (107.03 KiB) Viewed 1814 times

bp_680.png (131.8 KiB) Viewed 1814 times

bp_681.png (128.1 KiB) Viewed 1814 times

Now it's time to calculate the forces and simulate the real mapping.

[microelly2]

stick a leather piece (2D Bspline curve) onto a shoe last

phpBB [video]

[microelly2]

I have played with inventor LineSet to get colored wireframe objects (this will be a good way to get hatch patterns for illustrations too)
and added some scaling methods: calculating geodesics is time expansive ( I need 1 ms per point, so going down to 0.1 mm quality takes minutes)
We now can adjust the location of the patches on the face and we can rotate them.
The difference between "exact" positioning and 1 mm - 4 mm approximation is not significant for real time draft design (I hope).

phpBB [video]

link to model
https://www.dropbox.com/s/suoxkdz08fkx1 ... fcstd?dl=0

[microelly2]

create meta data by marker and geodesic connections

phpBB [video]

[microelly2]

the leather pieces go to their position on the shoe last relative to the markers

phpBB [video]

[microelly2]

create a bspline for a shoe last rib starting with 400 scan points and 80 error points (noise) with a script.
(the error points are added to test robustness of the method )
the result spline has only 12 poles.

bp_712.png (70.69 KiB) Viewed 1604 times

[microelly2]

I used the methods from the lidar importer (https://forum.freecadweb.org/viewtopic. ... 70#p216866) to improve the usability of the scanned data for the shoe last.
First I "unroll" the points to a elevation structure - think as a mapping of a cylinder surface with heights to a plane with heights.
Then I can run a scipy.interpolate.Rbf to get a elevation grid. I do this in 10 x 10 subsquares. If something goes wrong with such a square cm I can
use a simple quadrangle instead.
Then I map this back to the real 3D surface.
Still some artifacts have to be deleted. I think it's time to start the work on a sculpting tool for quadmeshes. Then the irregularities ca be
smoothed with the iron.

phpBB [video]