Completely different overhang / infill behavior when part is rotated 90 degrees

[DasBootU96]

Do not report Prusa3D Slic3r bugs here without confirming it is a problem on a development release of Slic3r, or your issue will be closed. Development builds of Slic3r. Prusa3D Issue Tracker

Describe the bug Completely different overhang / infill behavior when part is rotated 90 degrees

To Reproduce Steps to reproduce the behavior:

1. copy / download https://github.com/VoronDesign/Voron-2/blob/Voron2.4/STLs/VORON2.4/Gantry/X_Axis/X_Carriage/Printheads/TriangleLab_Dragon/printhead_front_dragon.stl

2. drag drop the printhead_front_dragon.stl in SS. Make a copy of printhead_front_dragon.stl or drag & drop printhead_front_dragon.stl again in SS. You now have identical printhead_front_dragon.stl in SS.

3. rotate one of the printhead_front_dragon.stl parts 90 degrees and slice

4. examine layer 130 (@0.2mm) with overhangs. For the rotated part, overhangs are printed in mid air and shows perimeter lines where there should be infill.

3D Model and Slic3r Configuration Export Please upload a ZIP archive containing a copy of the 3D model you are seeing this bug with and a CTRL-E export of the configuration used. printhead_front_dragon.zip

Expected behavior Expected behavior is that overhangs are not printed in mid air and infill is not printed as perimeters.

Screenshots If applicable, add screenshots to help explain your problem.

printhead_front_dragon.emf.zip

Desktop (please complete the following information):

• OS: Windows 20H2
• Version 2.3.56.0 (https://github.com/supermerill/SuperSlicer/actions/runs/804896051 - Nigthly windows x64 #299 - cf80732 Nigthly)
  • Do not report Prusa3D Slic3r bugs here without confirming it is a problem on a development release of Slic3r, or your issue will be closed. Only use normal Slic3r version IDs.
  • For -dev versions, use git describe --tag or get the hash value for the version you downloaded or git rev-parse HEAD
  • Use About->About Slic3r for release versions.

Additional context Issue seen in SS 2.3.56.0 and SS 2.3.55.2. Issue not present in PS2.3.1

[neophyl]

It seems to be caused by the Extra Perimeter on overhangs setting. One thing I've noticed with both SuSi and PS is that bridge infill direction can be a bit sub optimal when its automatic, often going the long direction rather than the shortest.

Notice how if you turn off extra perimeters on overhangs (a setting not in PS) the bridge direction relative to the part is different and goes the same direction for both part rotations. Ideally they should both bridge the shortest distance no matter which way the part is rotated but they aren't.
When you combine this with the Extra Perimeters on overhangs you get the 2 different results on the slice I suspect.

It doesn't happen in PS because it does'nt have that function/setting.

[DasBootU96]

Yes, turning off extra perimeters: [v] on overhangs gives the exact behavior you mentioned:

thanks @neophyl

[supermerill]

that's a bug in the bridge detection, as it shouldn't fill bridge area.

[supermerill]

I checked and it's because the bridge angle is set to 180°, so bridges can only have one direction, hence the different behavior if rotated.