Superfluous top layer inside model

[pemur 1]

I have been printing hive modular drawers and outers from Thingiverse for a while, without problems,  Due to printer changes I've had to modify the start code in the gcode file (generated in Cura 4.9), and rather than edit each file I thought I'd just re-slice.  The outers are printed at 101%, to give some leeway in inserting the drawers.  But when I do this with Cura 4.10 it introduces an extra top layer at one particular row in the file:

 

 

 

 

All other rows in the file are correct, just walls. 

 

This doesn't happen at 100%, but trying 99% puts the extra layer at a different point, further up the object.  This appears to be a top layer (shown by swapping patters for top/bottom layers).  I have no idea where this is coming from - and equally cannot see how to remove or avoid it, other than printing at 100%, which won't match my previous drawers.  Any thoughts?

[pemur 1]

A bit of further information - the anomaly only seems to happen with a Standard Quality (0.2mm) profile, switching to Dynamic (0.16mm) yields a perfectly formed product, albeit predicted 30% slower to print.

[nallath 1,124]

Without the actual model / settings, we can't really help you any further. There are a number of potential causes to this.

[pemur 1]

OK, glad to hear there may be some answers, because it's puzzling me intensely.  Give me a run down of the possibilities and I'll work through 

 

The model is from Thingiverse, 1743145, though I think it's present all over the place.  I've been using Cura 4.10 at Standard Quality profile (for Creality Ender-3 Pro) with no modifications, expanding the outer case to 101% to give some freedom of movement for the drawer.

 

Playing with different quality settings shows the effect does not occur at anything other than a 0.2mm setting, but it is reproduced with CHEP's 4.10 Good profile (also 0.2mm).  Using different scales indicates it's perfectly all right at 100%, but 101% gives an extra skin at layer 58/389, whereas 99% gives the extra at layer 329/381.  Other scalings may or may not produce the anomaly, but usually at different layers depending on the scale - 97% for instance introduces it at layer 62/373.

 

When slicing there's a definite pause at around the point the extra skin is generated.

 

I've tried with nothing enabled in mesh fixes (though disabling Union Overlapping Volumes did mean significantly extended slice times), no change.

 

What I did note is that the file includes an extra little bit of inner wall where the layer is added, not present at any other layer:

 

 

Worst case is that I just dig out my old gcode, generated with Cura 4.9, and make manual edits in there (specifically to enable mesh levelling on the printer) so even if it isn't solvable nothing is really lost.

 

TIA.

[GregValiant 1,410]

A work-around seems to be changing the layer height to .21.

I didn't find any errors in the model, but in wireframe it looks like this.  Of the 6 inside faces it's the only one that has this whatever-it-is feature (EDIT:  It's a logo).

I think the odd slicing (dependent on scaling or layer height) is either a bug, or Cura is being fooled by this model and it's acting like a bug.  Either way I think maybe it should go up on GitHub.

 

HIVE_Module_V5.stl

Edited August 31, 2021 by GregValiant

[nallath 1,124]

8 minutes ago, GregValiant said:

A work-around seems to be changing the layer height to .21.

I didn't find any errors in the model, but in wireframe it looks like this.  Of the 6 inside faces it's the only one that has this whatever-it-is feature.

I think the odd slicing (dependent on scaling or layer height) is either a bug, or Cura is being fooled by this model and it's acting like a bug.  Either way I think maybe it should go up on GitHub.

 

HIVE_Module_V5.stl 13.28 MB · 0 downloads

If the issue only occurs with a certain layerheight, it's often because of something in the topology of the mesh. After all; if you change the layerheight, you also change where the slice happens (and thus change the shape of the 2D polygon we then work with). It could be that some specific geometry here is causing this issue. But yeah, please report this on GH. You can do that here: https://github.com/ultimaker/cura/issues

[pemur 1]

55 minutes ago, GregValiant said:

A work-around seems to be changing the layer height to .21.

I didn't find any errors in the model, but in wireframe it looks like this.  Of the 6 inside faces it's the only one that has this whatever-it-is feature (EDIT:  It's a logo).

I think the odd slicing (dependent on scaling or layer height) is either a bug, or Cura is being fooled by this model and it's acting like a bug.  Either way I think maybe it should go up on GitHub.

 

HIVE_Module_V5.stl 13.28 MB · 0 downloads

I am using a logo-less V3 version, from some months back, where five of the faces are plain.  There's some pattern nonsense at the base of the sixth face, but no indication of variation anywhere near the problem layers.  I'll see if I can find any other pattern in the behaviour and, as suggested, fill in a github report - and revert to my old gcode for further prints.  Thanks.

[pemur 1]

45 minutes ago, nallath said:

If the issue only occurs with a certain layerheight, it's often because of something in the topology of the mesh. After all; if you change the layerheight, you also change where the slice happens (and thus change the shape of the 2D polygon we then work with). It could be that some specific geometry here is causing this issue. But yeah, please report this on GH. You can do that here: https://github.com/ultimaker/cura/issues

Will do, thanks!

[GregValiant 1,410]

After playing with the model some more I'm pretty sure it's the model file (this is the one with the logo).

I opened the file in MS 3D Builder and scaled it to 101% and did a Save-As and then that file sliced OK in Cura.

I re-opened the original file in 3D builder and immediately saved the file (over-wrote the original) and the model appears to slice correctly.

There is something odd about the file and it's not showing up as a discernable error in the model.

My expert professional opinion is that "sumpthin' ain't right".

[pemur 1]

Just tried with the newer V5 version here and that one is weird - still gets an extra skin at 101% scaling, but 95% gives me four extra skins at different layers. Have added info to the github report.

[pemur 1]

And what may be my final contribution here (because I am now weirded out entirely) - the location of any extra skin depends on the position of the model on the build plate.  Five identical copies, side by side, sliced simultaneously, all give different final results.  How that can be a feature of the model escapes me completely.