I’m getting a weird problem with just a few files*.
The files load cleanly with no errors. When slicing them Bambu studio takes a longer time than usual, pausing at 60% complete, then finishes. Scrolling through the sliced model BS has ignored the instructed wall count (4walls) and infill data (45% gyroid) and sent instructions to print a single wall, no infill at all and fill the model with a support structure instead.
This is completely repeatable with 2.3.1.51, also with 2.03.01.51, 1.10.02.76 and even Orca 2.3
Even making a small edit to the file in Blender and exporting it as a new STL will make BD make the same error.
Any ideas what’s causing this ? Has anyone else bumped into the same problem ?
*The files are small human figures 90mm high generated by Meshy AI. They’re then loaded into MS 3D Builder to repair any non-manifold errors and saved.
The problem has only occurred in a tiny number of the files I’ve tried to print.
No, the body is at least 10mm across. Other similar models will print with 4, 6 or 8 walls without problems, but a couple of files slice to a single wall thickness with no infill at all regardless of the settings in BS.
Changing model orientation makes no difference.
Quite bizarre, never encountered a problem like it before in over 1000hrs on the A1 or my pervious Creality printer.
The same issue with arachne selected.
If you look at the screen shot included; You can see that I’ve selected 4 wall loops and 45% infill, but the slicer has just put a single wall loop and no infill.
It has given you a hint: a single OUTER wall, and it grew supports in the cavity - supports cannot be generated inside the walls of a model, only ever in outside space
It thinks the geometry is hollow. It may be? This sometimes can happen with nonmanifold (error) stls, is the slicer prompting you to Fix/Repair the file?
As I said in my first post;
“The files load cleanly with no errors”
“They’re then loaded into MS 3D Builder to repair any non-manifold errors and saved.”
When I see support in your object, it is hollow. Can still be a valid model from a topology view point, but that is definitely the reason for your problem.
So in my opinion, you need to look at the source. Not at Bambu Studio.
But how can I diagnose the model and any future ones ? I can’t see any discontinuity in the model’s wall scrolling through the slices. Is there any software that can detect an STL as hollow and/or highlight where any hole might be ?
While I think the problem is the model file itself, it’s always worth trying to reposition/rotate the model and slice again. I’ve seen plenty of examples over the years where that produces a different result when slicing complex geometries.
But I think this is a problem with your model file. It may be non-manifold even if the slicer doesn’t pop up an error. It takes longer to slice because even though it looks the same, there’s a spot where the error significantly complicates the work the slicer has to do.
You should try MeshMixer to see if it spots the problem.
If it doesn’t, to figure out where the problem is, I’d start by chopping the model in to two segments. Halfway up, split it and see if one half slices differently than the other. If one is solid and one is hollow, for example, the side that’s wrong has the problem. Split that one in half. Lather-rinse-repeat until you’ve zeroed in on the area with the defect. Then you can fix it back in Blender and re-export for slicing.
Load it up in blender, enable vertex mode, and zoom to infinity
Is the model an ultra thin shell? Vertex on one side, vertex on the other side, with two distinct faces on the “wall”?
People wouldn’t generally make a model like this, but an AI might so you do kinda have to keep this in mind…
Yes, went all through that before posting with no change.
The problem is that MM doesn’t flag up an error as such*.
If you start zooming in hugely on the figure it’s possible to see very small holes. However fixing them seems really tricky, especially given Meshmixer’s dreadfully small UI and complex options.
*I guess for some meshes being open(hollow) is planned, so I guess it’s a bit optimistic to find a package that will flag it as an error. A complete PITA in my use case.
