A little slicing help for a newbie

[Robert Flemming]

I'm new to KISSlicer and 3D printing in general so forgive me if the answer should be obvious. I'm trying to slice the train body here:

http://www.thingiverse.com/thing:11671/#files

Unfortunately what results simply falls apart. The issue is where the two different height volumes meet at the connector: KISSlicer is treating them as two different volumes and thus giving them each their own perimeter which would be okay (though sub-optimal) if the perimeters continued all the way up, but it creates some funky paths which results in the two sections being almost completely disconnected. See the attached slice image.

I've tried tinkering with various perimeter, loop, and infill settings but the results appear to be more or less the same. Even at 100% infill there's still a strange void where those two volumes intersect. I aborted the print once I realized it was going to fall apart, but the from the look of the paths in KISSlicer, the same problem might exist on the opposite connector as well.  Is this a model issue, KISSlicer bug or me just not twiddling the correct knobs?

Any help is appreciated.

Robert

[Rick Zehr]

Try it with Supports On?

[funBart]

Hi, welcome here!

The problem with that behavior is not Kisslicer, but the model. It is -as many models on thingiverse unfortunately- not manifold, or 'watertight'.

In general is a stl a bunch of triangles, each of the sides of that triangles has at least exactly share a side with another triangle. The complete model must be covered with triangles: that's what's called a 'watertight' model. Important as well: each triangle has an inside and outside: of course for a good stl, the triangles have to be oriented well: so the 'defined' inside of a triangle facing the inside of the model. There are more rules for it but that are the most important ones.

So when triangles are overlapping, or leaving a gap, or double, or insight out, it gives errors as you have seen.

I have checked it, but stl has some errors, therefore KS can't slice it ok as well.

Kisslicer is more picky regarding an error free stl then other slicers, first I disliked it, but changed my opinion. When providing an error free stl, there is a much more predictable result.

You can repair it in the free netfabb http://www.netfabb.com/downloadcenter.php?basic=1 , but that wasn't enough for this model. I have repaired it in a professional program for you and attached it.

You can try the free Meshmixer from AutoCad as well. It's not stable, but I had some descend results / repairs with it sometimes. But hey! It's free! http://www.123dapp.com/sandbox

I attached some screenshots of the repair program as well the comparison in KS of the behavior with stl's with errors.

I saw a similar posting at: https://groups.google.com/forum/#!msg/makerbot/w7p_NQBBrW4/aeCrcrhARD8J

Bart

a copy of PenskeGuys post on a similar topic:

Models from Thingiverse are often junk. Netfabb will usually not fix a model to perfection. Most times, it introduces more errors than it finds or fixes, if it fixes any.... or finds any. usually, the only "mesh fixer" that is reliable has a pulse, because it takes intelligence to make correct decisions on something as complex as a 3D model having a multitude of points and polygons. No free lunch.

You need a mesh modeler, a converter or importer that will convert the STL mesh into the modeler's native format. Then look at the Mesh Error Key that shows if KS identifies errors to point you to where on the geometry the problem lies. Find the issue on the model and remove the usually extraneous geometry by welding points, flipping polygons, closing holes where polys are missing, locate two polys that may or may not be coincident (exactly same shape in same location), etc. The model needs to be what is referred to as "water tight", or in the technical term: "manifold". If it isn't the slicer will not be able to figure out what is the inside and what is the outside and you get garbage that is the evidence of this confusion.

Many CAD applications will not produce manifold models, nor do they convert to STL meshes very well. That isn't their focus. They aren't mesh applications by nature. They will have lots of coincident points, 2-point polys, points not on a poly, points on only one poly, etc. STLs need to be fully triangulated. This means one poly having three, and only three, points describing its shape. When you see "Degenerate Tri" in the Mesh Error panel, this usually will lead you to a "tri" that has four points. Selecting two that are near each other and welding them will clear the problem.

Other than that, it's too long to go into here. Jump in and you'll catch the drift.

[Robert Flemming]

Wow, thanks so much for the detailed response and especially the fixed model.  No doubt my 3 year old will appreciate it as well.  The last 3D modeling work I did was back in college and since I'm unlikely to get 3D Studio 4 running on DOS any time soon, I'll need to brush up my skills a bit :)  Thanks again!

Robert

[frozen...@yahoo.com]

Hey Bart,

What did you use to repair this model with?

[funBart]

Hi, that's Magics 17 of Materialise.

http://software.materialise.com/magics

[frozen...@yahoo.com]

Thanks, never ran across that one.

[funBart]

Probably because it costs 4000 dollar a year ;-) Stupidest investment I have ever made as my CAD software was little time after it able to produce good stl's...