Hi @abarbar, thanks.
While the solver itself should now always be stable and not prone to energy explosions like the old Kangaroo, there are some specific types of geometric goals which can still cause runaway reactions, one being the isothermic goal. As I was mentioning in the linked thread - if the angles of the quads deform too far beyond some threshold this runaway effect can happen. There is then a grasshopper bug that if any point travels extremely far from the origin, the viewport stops showing grasshopper geometry altogether. If you reset your Kangaroo simulation and viewports(by clicking the 4 viewports icon) it should return though.
@TheCyclist That looks really nice, can you explain the procedure!
Hi @DanielPiker, the function developable surface using discrete geodesic nets is already available? could you attach an example? thanks
2 Likes
It’s pretty simple. I began by lofting a number of random shapes together (triangle lofted to square, lofted to hexagon, etc.) That lofted surface was then remeshed and the developable surface component was applied via Kangaroo. The script has several adjustable components as well. Sometimes, depending on the originating geometry, the result doesn’t turn out as nice. Some trial and error may be necessary.
2019.04.29_mcneel_forum_totems.gh (30.9 KB)
9 Likes
Amazing, Daniel! Thanks for making the work of this research group available to us.
I have done some splitting using angle thresholds and minimal face counts in Ivy and the results are promosing. I subsequently flattened using a Zombie Solver by using an OnPlane constraint and Hinge with rest angle at 0.
The paper has some interesting examples using meshes from topology optimisations, which is a great idea. It could be nice also to work with RemeshByColour outputs to see what details could be achieved. With the horse it could be good to have higher mesh resolutions in the tight parts, which I omitted.
9 Likes
Here is the definition
8 Likes
Hey @DanielPiker , I know this is a beta release – sorry I’m asking anyway:
When I replace the the files with the new Kangaroo 2.5 files, K2Engineering for some reason stops working (as in not recognizing components anymore). Any idea why or how I could work around this?
Hi : ) please tell me in which version of Kangaroo can the component RemeshByColour be found ?
I have Kangaroo2 (242) and Kangaroo0099, it’s not here((
@kamilagiz111 - the RemeshByColour component is in Kangaroo v2.5.2 (which is included with the latest release of Rhino 6). It can be found in Kangaroo2/Mesh.
Thank you!
Спасибо вам большое! : )
@mathias_gmachl any chance you could internalise the horse mesh you used?
I’m not certain, but I’m guessing maybe Mathias used this horse model
Here’s a version as a Rhino file remeshed to a reasonable density
horse.3dm (727.6 KB)
It is a common test model used in many geometry processing papers to demonstrate various algorithms on, along with the Stanford bunny, armadillo, buddha etc.
I’ve also used it for several examples posted here in the past.
This was the original source:
http://web.archive.org/web/20010126102500/https://www.cc.gatech.edu/projects/large_models/horse.html
1 Like
Thanks @DanielPiker but that is not the one he used in Kangaroo update anytime soon? - #25 by mathias_gmachl
His definition linked below his post (https://discourse.mcneel.com/uploads/short-url/vw76Yf1DV7ohMwx8i3SlY2gWItf.gh) does not contain any of the components in your developablize example definitions. So I was guessing he baked the output from developablize_example2.gh and used that as input for his definition. I tried the same but it’s not splitting the surfaces along the sharp angles. I guess the key missing part is this quote below?
I subsequently flattened using a Zombie Solver by using an OnPlane constraint and Hinge with rest angle at 0.