Using Slic3r 1.1.1.-dev (c4bfe64) to slice a thin wall calibration piece (https://raw.github.com/ahmetcemturan/SFACT/master/calibration/_40x10.STL) i noticed there is a significant overlap of each layer start and end points. Measured on the physical print, the overlap is about 2.5 mm wide, which seems excessive. Expected behaviour would be an overlap approximately the size of a single wall end radius or so (about 0.25 mm in my case with 0.5 mm perimeters).
I've used http://gcode.ws/ to visualize the problem, you can see the overlap in the attached screenshot.
My Slic3r config is available here: http://pastebin.com/zZp2Ezns
(I'm not using wipe on retract)
The same problem is discused in #1875. This is problem with medial-axis code. Maybe you can avoid using using thin-walls detection (it doesn't work for some closed shapes) and use tricks to print thin wall vase (make the object solid and use zero infill).
Ok, so it's mainly affecting thin wall objects? I guess i can live with that for a while. (I stumbled on this while trying to calibrate away seams as much as possible, i'll switch calibration object to something more solid and less rounded for that then). Thanks for the pointers.
PS Is there any way to mark an issue as a duplicate?
It is probably only affecting objects with thin wall that forms a loop. If the thin part has a start, then the algorithm produces correct result.
Bugfix: thin walls forming a closed loop had overlapping segments at …
…their endpoints. #1948 #1875
Fixed! And added a regression test.