PDA

View Full Version : Need help on cleaning up for 3D printing



sublimationman
09-13-2014, 01:59 AM
So, I have this new 3D slicing software (Simplify3D) that I paid for because it prints better looking objects than the free ones and it's support (scaffolding not technical support) is the best.

Problem is it's picky about the STL file being all proper.

So, if you look at the attached image (just a quick mock up sample) what is the proper way to merge these 3 shapes so they are one (water tight or manifold) and not have any internal polygons?

Slic3r, Cura.... where all tolerant of this and I would be able to slice this object as is. The final product is so nice with the new software though that I want to use it (plus I paid $140 for it). I asked them and they basically said the model is not made properly.

So I really need to learn to make these things properly.

I have tried different boolean functions but none seem to fix both issues. Yes, I could manually select and delete internal polys but just tonight I was working on a model that was large and had some 60 or so internal polys. And they were in such a position that it was near impossible to select them. I ended up using Netfabb to fix it but I should not have to do that, I need to learn the proper way to merge these items cleanly.

Thanks!

spherical
09-13-2014, 03:33 AM
So, I have this new 3D slicing software (Simplify3D) that I paid for because it prints better looking objects than the free ones and it's support (scaffolding not technical support) is the best. Problem is it's picky about the STL file being all proper.

Well, depending upon your point of view, that's not a problem. It's a Good Thing.


Slic3r, Cura.... where all tolerant of this and I would be able to slice this object as is. The final product is so nice with the new software though that I want to use it (plus I paid $140 for it). I asked them and they basically said the model is not made properly.

It isn't that they're "tolerant", they don't know that the model is Borked to begin with. The better slicers do know.


So I really need to learn to make these things properly.

Good approach! You'll become a better modeler because of it and will also know how to go into a model and manually fix degenerate tris, edges shared by more than two polys and other stuff that Netfabb won't. Netfabb can introduce model errors while it is trying to fix what it thinks is wrong. I never bother with it.

First thing is that Booleans do not work on points or edges or polys, depending upon the axes of the latter two. Think "Flatland". Points have no dimension in any axis, edges have dimension in only one (there is no height), polys have dimension in only two (there is length and height but no depth). Trying to split a point in two will not work, because half of nothing is still nothing. That is why you end up with a lot of junk, because the Boolean algorithm is confused as to what is the object and what is the cutter.

The cylinder is placed correctly but the long angled rectilinear isn't. It should extend fully into the cube, not have two vertices coincident with the cube's side poly. Otherwise, the example model you gave should Union just fine and have only one outside and one inside after having merged points and deleted any that are left hanging where they shouldn't be. Sometimes the Boolean operation leaves point cruft hanging around. Usually these can be identified as points on 0 polys, points on 2 polys or just manually select and nuke.

124223 124224 124225

Try the STL I made and see if it works.

124222

MicroMouse
09-13-2014, 07:21 PM
The mesh that spherical made contained 4 interior triangles that should have been deleted.

I have deleted those triangles and included both ASCII and Binary STL files in this zip file.

124239

Wayne

spherical
09-13-2014, 08:04 PM
Confirmed. Actually, they shouldn't have been there in the first place. In a hurry to crash. Had I taken the time to run the STL through my slicer, it would have shown that I had converted the layered LWO to STL, instead of copying the Unioned part to a new object and then exporting. Parts of the original cube were in there, so it threw an error flagging edges and saying that more than two tris shared these edges.

One thing to note. At least with my slicer, placing the geometry centered on LightWave's origin places half of the part into the bed and you have to move it. I always rest the part on the ground in LightWave before saving and then exporting to STL or using an external converter (preferred). Then it automatically sits on the bed as it should, with no messing around.

BigHache
09-13-2014, 10:13 PM
So you're at the point where I was when I began experimenting around. Truth be told, I'm not convinced that any OTS 3D package's stock Boolean algorithm is satisfactory for what we need to do for 3D printing, efficiently.

Speaking on Modeler specifically, I have spent a LOT of time cleaning up Boolean operations on fairly complex meshes. Not efficient enough. So I invested in 3rd Powers' Boolean plug-in. It does a pretty good job and has some options for controlling the joint areas.

But what you're getting into is rapid prototyping and this is not what most modeling packages are built for. ZBrush and 3D Coat currently dominate this industry, mostly the former. Hasbro presented at the ZBrush Summit and mentioned they are now 98% digital and basically use that package to create assets that eventually turn into toys. I went 3D Coat myself because it was cheaper for me to get my feet wet with sculpting. What's nice about it is 3D Coat opens LWO files.

Probably not the answer you were looking for but this is my view after what I consider fully exploring Modeler's Boolean operation and finding myself left with non-manifold models that I have to manually clean up. If you simpler meshes it's not as big a deal and might not matter.

spherical
09-13-2014, 11:36 PM
I usually don't encounter problems with LW's Bools, if it works. Sometimes the Speed variety will work when the original won't. Just have to be careful about how the cutter(s) are placed. I also found that cutting with multiple cutters seems to work better (or at all) when they are all done at the same time. Sometimes, adding a cut later just screws the whole thing up. I did have one insanely complex high rez model that I had to slice, in order to create negative molds and the cutter slicing through points was inevitable. Left a lot of cruft that I had to manually fix, but that was to be expected.

I have 3rdPowers, both ZB and 3DC and MeshFusion in Modo, so will try a few models in them. Hadn't really considered ZB and 3DC for FDM.