1
$\begingroup$

I have an object which is going to be 3D printed to which I would like to add lot of small rivets, and to do this without losing its manifoldness. The way I've decided to do this (stupid it may be) is to place spheres in the appropriate place so that a portion of sphere is sticking out, the spheres being separate objects to the thing they are "riveting":

enter image description here

enter image description here

However, keeping them as separate objects is kinda risky (one false move...) and so I'd prefer to incorporate them into the main object. But I need to do this without destroying my manifoldness. I've tried a Boolean Union but that leaves stuff hanging over on the inside of my object, which makes it non-manifold. Is there a way to achieve my aim, or maybe an entirely different rivet strategy which I should adopt?

$\endgroup$
2
$\begingroup$

You could model this, if you're starting with a topology something like the one illustrated here:

With Pivot set to 'Individual Origins'

  1. CtrlR add three edge-loops
  2. AltP poke the two middle face-rings
  3. Select a couple of consecutive rivet-vertices (not necessarily just one apart) on the central loop - use CtrlShiftNumpad+ to select the rest, and then CtrlShiftB bevel those vertices, in 'Offset' mode
  4. Space .. look up the 'Circle' operator, and use it on your newly created faces. (They should still be selected)
  5. perform whichever series of insets, extrudes, and scales create the kind of rivets you want.

enter image description here

I don't know whether you're sub-d modelling, but this is the result after putting on a level of subdivision.

enter image description here

You could use the Loop Tools add-on > 'Space' option at the vertex stage, in various ways, if you want to improve the distribution of rivets.

Although the result is not all quads (before subdivision), it's in the flat, for printing, and is not expected to deform, so that's probably OK?

$\endgroup$
2
  • $\begingroup$ Ooh, yes, lets of things I've not tried before there. I don't quite have the arrangement you started out with but I have something similar and through an insane number of operations and a little repairing of non-contiguous surfaces from the 3D printing tool I can achieve a manifold riveted object. Cool. :-) $\endgroup$
    – Rob
    Sep 2 '18 at 23:21
  • $\begingroup$ You just need a radial edge for each of your rivets, and quads all round, to speed up selection. And, of course, you can 1/4 the amount of work you have to do at that stage by using a mirror modifier in 2 axes. $\endgroup$ Sep 2 '18 at 23:34
0
$\begingroup$

You can use a boolean set to difference to create a hole in the shape of the sphere and then press J to merge the objects into 1.

$\endgroup$
2
  • $\begingroup$ I can do that but it ends up non-manifold because one side of the hemisphere is still there. Is there a simple way to delete it? $\endgroup$
    – Rob
    Sep 2 '18 at 23:25
  • $\begingroup$ It is better to avoid boolean operations when possible. They may introduce messy geometry that is harder to work with. $\endgroup$ Sep 3 '18 at 10:34
0
$\begingroup$

I tried the true modelling method which I think would have been the right way if I were starting from a clean shape but, since I'm applying rivets to an existing shape which is not necessarily always planar, I found that the subdivide/poke/bezel/loop procedure ended up producing shards that were difficult to get rid of. So I adopted a more laborious but less "shardy" method as follows:

  1. Create a cylinder to use as a rivet-punch.
  2. Position this at each rivet position and Boolean difference it with the base shape.
  3. Select the edge of the resulting hole, circlify it with loop tools, extrude it as a region and add a face.

Provided I'm careful with the rivetter this always produces a manifold structure. Here's the model, a picture of the thing I'm modelling, rivets exaggerated in the model otherwise they wouldn't be visible (the length of the riveted section you can see in the picture is about 22 mm when printed to the correct scale), and a test print.

enter image description here enter image description here enter image description here

$\endgroup$
4
  • $\begingroup$ Hehe...... you always want to start from a clean shape.. ; ) The Boolean habit can often get you into trouble IMO, but maybe that's just me.... $\endgroup$ Sep 4 '18 at 12:56
  • $\begingroup$ Yeah, I understand that now, one for next time. Thanks for your help, I definitely appreciate pointers on process. $\endgroup$
    – Rob
    Sep 4 '18 at 18:08
  • $\begingroup$ FYI, I've now redone the model starting from a clean shape :-). $\endgroup$
    – Rob
    Sep 23 '18 at 23:00
  • $\begingroup$ Cheers! I've had a look.. It would still be difficult to use this method with your 'fan' topology, rather than generally quad face-loops and a more traditional sub-d approach. Despite its apparent convenience, I think your method will probably wind up being harder work than a much simpler framework , (basically hinting the sub-d how to make your curvatures for you, rather than explicitly laying in hi-res bevels). Many of Blender's tools are well-geared to it. Check out some subdivision-surface tutorials (any software) and see if it suits you. $\endgroup$ Sep 24 '18 at 19:45

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.