Arachne Does Not Obey Slicing Resolution

[ruredi87]

Description of the bug

I noticed arachne causes some misalligment at walls, specialy on curved walls. If you look the pictures there is artifacts on outer walls not because of variable line width, because of generated printing paths and their random allignment. Also arachne generates walls with much less gcodes compared to classic slicing engine. I randomly picked a layer and analyzed it. With classic slicing mode the outer wall complating with 163 commands (between 174176-174339 lines). But with arachne it is only 34 gcode commands (between 60023-60057 lines)

Arachne's this behavior causes artifacts on walls and much less resolution on walls. plesase see the pictures. There is no problem with classic slicing engine.

Sliced file size with arachne around 1mb, With classic around 2.4mb

Project file & How to reproduce

bmg project (2).zip

Checklist of files included above

• [X] Project file
• [X] Screenshot

Version of PrusaSlicer

Build: PrusaSlicer-2.5.0+win64-202209060714

Operating system

Windows 11

Printer model

Ender 3 v2

[kubispe1]

Hi, thanks for report. We detected this problem yet. I think this is duplicate of #8524, so I am closing it. I also recommend to check this comment: https://github.com/prusa3d/PrusaSlicer/issues/8524#issuecomment-1255047681. Quite nice description (Increase of resolution of exported .stl is needed).

[hejllukas]

I reopen this issue. This is a different issue than #8524. This issue is caused by preprocessing of input polygons for Arachne and post-processing of perimeters generated by Arachne, which doesn't respect the value of slicing resolution.

[kubispe1]

Ok, @hejllukas thanks for correction.

[DragonFinlay]

It is however the same issue or at least very similar to #8700. @hejllukas Enough to close this one again?

[TalalMash]

@DragonFinlay Same issue here, OBJ file with very fine resolution, completely fine with Classic: Classic: Arachne:

[TalalMash]

*Thank you for your efforts! OBJ file fine edge:

They're visible in real life, even with small models especially with reflective filament

[Mark-Dunn-OIL]

I have, for some time, been trying to understand where the resolution of my printed gears have been removed from my design. After reading this I switched between the two algorithms. For my gear teeth: Arachne issues 7 gcodes Classic issues 39 gcodes I realise that in this case 39 gcodes may be too many, but I would like it to be under my control as to how much approximation (deviation) I am prepared to accept.

Is it too much to ask for the benefits of arachne with the resolution of classic?