search

Ultimaker / Cura

3D printer / slicing GUI built on top of the Uranium framework
GNU Lesser General Public License v3.0
6.16k stars 2.07k forks source link

Tree support completely omitted for one side of symmetric model #8701

Open StarNamer opened 3 years ago

StarNamer commented 3 years ago

I got this the first time I tried slicing the model. Moving it slightly in the Y direction and reslicing gave the expected behaviour of supports in both ducts, but since the is 4.8.0-Beta I thought I'd log it in case there's something that can be done to avoid completely omitting a tree support.

Application version 4.8.0-Beta

Platform Windows 10 Insider Preview Build 20246

Printer Ender-3

Reproduction steps Sliced symmetric model

Screenshot(s) Duct on the right has an internal support, the one on the left doesn't. image image

Actual results Cura inserts tree supports for one part of model, but not for the identical part on the other side, as shown in screenshots.

Expected results Roughly similar supports on both sides (within the randomness of tree supports).

Project file CE3_6fd0e370-37f0-4c17-8ae1-05368710ba89.zip

Log file cura.zip

Additional information I'm printing it anyway, since I didn't notice until I'd started. I will update on success or otherwise.

Ghostkeeper commented 3 years ago

Hmm, this doesn't seem to reproduce any more in the stable release. It's creating support on both sides now.

image

We didn't make a fix aimed at this in particular in between. We've changed the Maximum Resolution setting for your printer since the beta, but changing that back to 0.05mm doesn't reproduce the issue either.

Is it still occurring in 4.8.0 stable for you?

StarNamer commented 3 years ago

Model printed successfully. Further slicing attempt withs 4.8.0-Beta could not reproduce the problem. I'll try a few times with 4.8.0 later, but suspect it was simply an extreme random case for tree supports. Next time I'll check and re-slice if I see this.

Ghostkeeper commented 3 years ago

Cura's slices are expected to be deterministic by the way, unless you're using random Z seam placement or fuzzy skin (which both use explicit random numbers). The placement of the tree's branches is according to a grid. Before 4.7 there was a bug with tree support that made it nondeterministic (by reading uninitialised memory) but since then we don't know of any randomness.