spirals instead of circles

slic3r 0.6beta from github, downloaded a couple of days ago.

running it with the circles option since I use Marlin, and when it works it is awesome.

In at least 3 very simple models, I got circles going nowhere. There might be an issue with the circle detection routine...




Edited 2 time(s). Last edit at 01/18/2012 07:56PM by Lanthan.

hello,
same here with slic3r 0.57, also fails on small rectilinear perimeters as in this
t_mount. first attachment is with arcs disabled, second with arcs enabled.

also with arcs enabled in this case slic3r outputs one single layer only.
on larger object it sometimes works fine...

grtx

Note to self: at least when the openscad source file is available, try increasing the number of facets $fn to see if this decreases the error frequency... I do not remember having seen this happen on high $fn cylinders...

BTW running standard current Ubuntu.

Arcs code is highly experimental, thus everything is possible

OK I think I found a quick fix: as predicted, when the cylinders are generated with a large number of facets (in openscad, $fn = 100) the circle-finding routine is less prone to generate artifacts. Same piece as previously, re-exported and re-sliced:



there is still a spiral artifact, always accompanied by an unusually large blob of plastic, but much smaller and manageable.

Suggestion: Maybe your routine should oversample a couple of times around any detected circle to confirm its exact radius

Edited 2 time(s). Last edit at 01/19/2012 06:24PM by Lanthan.