Grasshopper

algorithmic modeling for Rhino

Hi everyone,

i got a problem with Kangaroo that I`m not able to solve.

Using too much subdivisions the program always crashes on different computers.

I also tried to refine the mesh with weaverbirds "split quad subdivisions" command, but I dont know how to get sharp tips on the Anchorpoints to my mesh.

My favorite solution would be a fine mesh from Kangaroo.

Help would be wonderfull!

Views: 721

Attachments:

Replies to This Discussion

Hi NiStu

it's good that you try to solve this in Kangaroo as Weaverbird does not have iterative force finding methods and the mesh-grid is made of rectangles, so it fits this type of logic well.

An idea could be to directly shape the connection that you have (bridges) before subdividing, and then applying the chosen iterative force method. It seems you are trying that. If you make the bridge in the mesh, maybe less fixed points would help speed this definition up, but only profiling Kangaroo's code would help figuring what is happening, and I did that too long ago.

So, while not sure about the performance hit that is causing the higher-refinement mesh to slow down considerably, on Weaverbird's side I can suggest how to leave the boundary unchanged: toggle NakedEdges to Fixed in any subdivision method. That keeps the naked boundary of the mesh exactly the same, and you can approximate the interior of the mesh via subdivision.

I hope this helps, on the side I know something more about :)
Thanks,

Giulio

--

Giulio Piacentino
Weaverbird development

Hi Giulio,

 

thank you for the tip :)


i tried to model the shape in t-splines because it is the only form of mesh modeling I know. Then I applied the iterative force method.

Because Weaverbird`s "Split Quad Subdivision" command doesn´t works for T-Spline Meshes I have to do the whole process in T-Splines.

Compares to the original Mesh the result has bigger holes and smaller bridges. The tips of the Mesh are optimized not by modeling but by the "Weight Points" command.

 

To get a more constant result I copied the element in the middle, as you see in the second row.

Maybe someone else has an idea how to get bigger meshes from Kangaroo for more constance or controled variations in the holesize?

 

You helped me a lot!

 

NiStu

Attachments:

Hi NiStu,

I was away before and missed this post, but here are a few thoughts in case it is still relevant:

-You could use one level of subdivision for the relaxation, and then subdivide the output of Kangaroo further, after running the relaxation.

That said - How big were you trying to make your meshes ? I tried directly relaxing your file at a higher level of subdivision (so around 25000 faces), and while it slows down a bit it still works. Are you switching off the timer when resetting the simulation - if not it may be trying to run the reset operation multiple times causing a crash.

-Stiff cables around the hole boundaries could let you control the hole sizes (make them as springs with a non-zero rest length and higher stiffness than the rest of the mesh

Hi Daniel,

 

You are right, to stop the timer was one of my problems. My favorite facecount is 40000 (bigger square). That is too much I think. But with 10000 it works now.

 

When subdividing after relaxation , I got Problems with the shape of the mesh, because the interpolation changes the lower tips.
When I use T-Splines i get another shape and I have more control over the tips by weighting them.That works also with less subdivisions.

 

Stiff cables around the hole boundaries is a good idea to control the holesize! Probably better then increasing the holesize before relaxation.

 

Thank you for helping and for Kangaroo :)

 

NiStu

Attachments:

RSS

About

Translate

Search

© 2024   Created by Scott Davidson.   Powered by

Badges  |  Report an Issue  |  Terms of Service