Exoskeleton

Information

Exoskeleton is a free plug-in for creating meshes. It is currently comprised of two components: Exo Wireframe and Cytoskeleton.

Exo Wireframe thickens line/node into watertight meshes. It solves the nodes using a convex hull and stitches the hulls together with polygonal struts.
Cytoskeleton thickens the edge network of any existing mesh into a thickened mesh. The topology of the base mesh enables the production of a clean quad meshes whose vertices are all of even valence. Because It uses the half-edge mesh library developed by Daniel Piker and Will Pearson, it also requires the installation of the Plankton dll's and gha.

Exoskeleton remains a work in progress, and as bugs are fixed and new components and features introduced, updates will be placed here. Furthermore, Exoskeleton is an open-source library.

Download the most recent version here: Exoskeleton2_150904.zip

This has been updated to also include Plankton version 0.3.4

This library is distributed under the terms of the GNU Lesser General Public License (LGPL).

(The source is available on GitHub here)

Please use this discussion forum to post questions, describe issues, and provide feedback (and examples!).

Thanks to Will Pearson, for his work on Plankton and who also has given some invaluable contributions to Exoskeleton in terms of project organisation. Thanks also to Giulio Piacentino for Weaverbird, and for general knowledge and support, and Mateusz Zwierzycki for the same, as well as for sharing his code for convex hulls, which although not used explicitly here, was very helpful in many regards for the development of Exoskeleton.

Members: 464
Latest Activity: Aug 13

Discussion Forum

View All

Comment Wall

Comment

You need to be a member of Exoskeleton to add comments!

Comment by martyn hogg on June 7, 2015 at 8:05am: I got there eventually... the sl3dHull seems a bit buggy sometimes; I have to switch off parts of the model to get it to work.

Anyway, in theory I could make this, in practice I need to build a new CNC router first!

The overlapping mesh faces at the corners are just because the struts and nodes will interlock and the interlocking feature is not modeled

Comment by martyn hogg on May 28, 2015 at 12:14pm: Blimey, thank you! That would have taken me ages and will indeed take me ages just to understand it all and be able to re-create it but it is just what I was aiming for.

I always get stuck on managing data and notice you do some clever tricks with components like Propagate Ancester, PShift, Flip Last etc. which i really need to think about hard to understand and visualise what they are doing.

I like the smoothing end result. I'd not considered doing this but I will see if I can get some smoothing. The idea is to CNC the nodes from wood with a type of mortice and tenon joint and use approx 50mm x 50mm timber for the struts. I'd probably need to minimise the smoothing of the nodes where the struts will join but try and keep some smoothing in the centre of the nodes.

Thanks again!

Comment by David Stasiuk on May 28, 2015 at 5:37am: Here you go. I've also added at the end my Catmull-Clark implementation with creases...you'll need Plankton for that to work. If you don't have it, though, so long as you have [UTO]'s MeshEdit tools and Tree Sloth, the rest should be fine...it more or less follows the logic of the Exo plug in, just using components. ComponentExo.gh

Comment by martyn hogg on May 27, 2015 at 4:54pm: I'll have to spend a bit fo time on this I think!

Blog Post With File

I'm getting lost in data structures again!

Comment by martyn hogg on May 27, 2015 at 3:46pm: Thanks! I'll try that... I'm sure I got the Tree Sloth plugin but can't see the list comparisson component in the Tree Menu. The components should all be in Sets >> Tree right?

Comment by David Stasiuk on May 27, 2015 at 3:28pm: I don't know how much of a purist you're being in regards to components, but the Tree Sloth plug-in I wrote has a list comparison component that's really useful when you want to compare items against each other in a given list. You can choose for it to give all unique combinations, or you can set it to create for each object a data branch that matches it against every other one. So:

{0}

A

B

C

will return:

{0;0}

A - B

A - C

{0;1}

B - A

B - C

{0;2}

C - A

C - B

So if you had a branch with all the struts from each node, you could easily test and sort all of the angles for each one against its neighbors. Just an idea!

Comment by David Stasiuk on May 27, 2015 at 3:22pm: Exactly. For each strut, check it against each of its neighbors, and test the angle. From this - and the radius of both struts - you can calculate a minimum offset required. Then you set them at the "worst" case value.

Comment by martyn hogg on May 27, 2015 at 1:25pm: I'm trying to do it with components.

It's the "angular proximity to it's neighbours" that I'm struggling with.

I guess I need to work out the angle between the struts and then just look at any angles that are acute and then work out the required offset?

Comment by David Stasiuk on May 27, 2015 at 1:06pm: Hi Martyn- the way it works is that for each node, the convex hull is calculated for either the minimum offset to ensure that it will correctly capture all of the vertices or the offset defined by the user...whichever is bigger. Then, after it solves the convex hull topology, it goes through each strut and "shrinks" it to the minimum depth, based again on the user input, and the strut's radius and its angular proximity to its neighbors. There's a fair amount of trig involved, but nothing really fancy. Are you writing a script, or are you doing it with components?

Comment by martyn hogg on May 27, 2015 at 12:40pm: Hi David, I'm revisiting my exoskeleton definition in order to create one that has square section struts and I need to work out a way of keeping the node depth to a minimum without the struts overlapping. i.e. calculate a node depth for each strut.

I notice your exoskeleton component does this - if you set the node depth too small it holds some node arms away from the node centre point to avoid overlapping.

Can you give me a clue on the strategy for doing this please?

I was going to try and sort the struts of each node in order of angle, then calculate a distance for each strut that would avoid strut overlap but wondered if there is a simpler, smarter way of doing this!

Thanks!