1
$\begingroup$

I'm a newbie learning to use Blender to produce 3D printable objects, since it is the most flexible software tool on the planet for 3D printing. However, in doing so I realise that I am using only a few percent of its capabilities (e.g. spending all my time in Edit Mode on objects that will never move) and, in particular, creating manifold objects is proving difficult. My object looks like it should look but it is non-manifold in various ways (non-manifold edges, non-intersecting faces, etc.). The original Blender file is here:

I realise that I can improve my construction technique to be sympathetic with the ways of Blender and I can use the 3D Printing add-on to slowly and carefully correct each minute mistake but that seems wrong to me: if the object looks correct then software should be able to generate a 3D mesh that is, near as dammit, printable. Maybe not quite what I drew but corrected to be a printable thing. A modifier of some form: the be-slightly-fatter-more-rounded-and-more-connected-but-don't-bother-me-anymore modifier :-)? FYI the "Make Manifold" option in the 3D Printing add-on doesn't produce a sensible result in my case.

Is such a thing possible and I've just missed it in the many very useful tutorials, or do I need to go through very carefully finding the slivers of errors in my (beginner's) work, which is not much more than a cup-shaped thingy with drill holes? See a acreen shot of the object below, along with one of the manifold-corrected version according to the 3D Printing add-on, the 3D Printing add-on's list of errors and the view of my 'orrible object as seen by slic3r-3d (Prusa edition).

enter image description here

enter image description here

enter image description here

enter image description here

P.S. apologies in advance if I'm the Nth person who's asked this.

$\endgroup$
  • $\begingroup$ Hi, welcome! The fact is that Blender (as most modeling tools) does not print itself... It has tools that help in that respect, but in my experience (used Cura) the printing software has lots of possible improvements and fixes for "wrong" geometry, and most of the times they "sort out" a possible way to print your object (of course I would prefer to be sure in advance of the possible result), so it could work anyway, it depends. Try to detail more a sample issue, and it will be easier to suggest how to fix it if necessary, or not. $\endgroup$ – m.ardito Mar 13 '18 at 23:18
  • $\begingroup$ Thanks for the swift response: I'm afraid my technique is so bad that there are a lot of errors (3312 auto-repaired manifold errors in slic3r) but I've edited the post above to show what the 3D Printing add-on is telling me. $\endgroup$ – Rob Mar 13 '18 at 23:27
  • 1
    $\begingroup$ Sometimes object is better to remodel if fixing it requires more time than modeling again. It's hard to tell from screenshots but likely there are either inverted scale or non-recalculated normals, internal faces on every fraction of the rounded parts, and maybe duplicated / unconnected edges / whatnot. If you want that model to be downloaded and checked upload it on blend-exchange.giantcowfilms.com and edit the question with link $\endgroup$ – Mr Zak Mar 13 '18 at 23:47
  • $\begingroup$ Agreed, original .blend file uploaded and post edited. The only things that matters here are the external dimension, which needs to stay where it is to match that of another 3D object I printed previously, and the position/size of the holes, as they need to match the object to which the shape is being mounted. $\endgroup$ – Rob Mar 13 '18 at 23:58
  • $\begingroup$ On the remodelling topic, though I have been through the initial Blender tutorials they tend not to address 3D-printing issues and so, not knowing what produces the least non-manifold result, I think that starting again might not produce a better result in my case. $\endgroup$ – Rob Mar 14 '18 at 0:06
3
$\begingroup$

The object is non-manifold and the main reason why are these edges holding the top part because technically they aren't holding it:

screenshot-1

For now printer won't be able to decide where is the inner part of that, where is outer, just as Blender can't calculate normals for the object correctly.

If you really want to print this object the easiest way to work precisely is to duplicate and separate to another object the outline for those curved extruded parts in order to cut them with Knife project / Boolean once they are part of another object.

First copy the top part of the object with which is the profile of the curved part extruded from the bottom. Then either hide those vertices with Mask modifier or delete them. Then dissolve the vertices corresponding to the ones removed from the top, - they are just going through the bottom part faces and are creating intersecting geometry.

enter image description here

In the end you want to get the bottom part only without any additional loopcuts:

enter image description here

From here on get duplicated curved profile of the deleted geometry, extrude it to get a line of faces in the top and line of faces going down:

enter image description here

Add Boolean modifier to the original object shown above, set it to Difference and use this profile as cutter. Below the cutter is hidden and new geometry create with difference is shown:

enter image description here

Do the same for the other part. Then either extrude top part once more or (as is done in the file below) copy the one from the original mesh, clean it up from doubles and unnecessary edges to get even quads and join meshes (Boolean has to be applied first). Then select both lines of faces and run Bridge Edge Loops.

enter image description here

Here you can convert tris to quads or not - it's not quite important for printer, the geometry here is more or less clean. What matters for printer here are quads / ngons which might not be flat (in your model this is the case), as well as overhanging faces which might or might not require fixing though this is more printing related question than Blender.

Download the file

$\endgroup$
  • $\begingroup$ Wow, that's excellent, thanks for taking the time to do this. I will look into it all this evening and confirm my success/failure with it. $\endgroup$ – Rob Mar 14 '18 at 16:07
  • $\begingroup$ I've had several goes at this now but, on my latest attempt, I'm stuck at the Boolean difference stage. I have the tidied-up base as one object and another object that is the two "rim" parts but, when I apply the Boolean difference modifier to the base object, using the "rim" object as the difference, nothing happens, the base is unchanged. I have uploaded the file here: <img src="https://blend-exchange.giantcowfilms.com/embedImage.png?bid=4666" /> What am I doing wrong? $\endgroup$ – Rob Mar 15 '18 at 21:38
  • $\begingroup$ Sorry, I was being stupid, differencing done now, however after extruding the "rim" back up to size I am left with 16 non-manifold edges and, try as I might, I can't figure out what is non-manifold about them. The Blender file is uploaded here: <img src="https://blend-exchange.giantcowfilms.com/embedImage.png?bid=4668" /> if anyone can figure out how I should go about resolving them. $\endgroup$ – Rob Mar 15 '18 at 23:58
  • $\begingroup$ I did it, finally: spent the last 45 minutes crawling inside the object deleting things and that seemed to work. I need a better process next time... $\endgroup$ – Rob Mar 16 '18 at 0:46
  • $\begingroup$ @Rob fixing this kind of object is non-trivial process since it requires matching to already existing geometry hence I only roughly outlined the process and added a file, I'm afraid it's not quite clear. Glad you found the way out. Instead of Boolean you can use Kinfe project if needed, grab the outline (without any loopcuts going down, only edge loop), enter top view and use Knife project to cut the bottom plate, see knife project explanation $\endgroup$ – R El Clein Mar 16 '18 at 13:09

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.