Problems with angle force on polyline using Kangaroo 2?

Hi Guys,

I am trying to visualise the changes in angles for the polyline shown in the image below.

However I cannot get the bend force to work in this case, the resulting polyline seem to expand and wiggle in space. Any help would be great.

Many thanks,

Arthur

Attachments:

zara.gh, 15 KB

Replies to This Discussion

Permalink Reply by Laurent DELRIEU on November 2, 2016 at 3:27pm

I made some change, add some anchor, the first segment. Change drastically the strengths, very low on angle and high on lengths. Work better but still some problems !!! I continue to search

Tried a signed version of angle but it didn't worked

Attachments:

zara_LD.gh, 18 KB

Permalink Reply by Arthur Mamou-Mani on November 3, 2016 at 5:52am

Thanks a lot Laurent! It works!

Yeah curious strengths ratio, I really wonder why. It's interesting that the curves go aligned as opposed to curled which is what I thought would happen.

Permalink Reply by Laurent DELRIEU on November 3, 2016 at 6:43am

I also thought it will curl but as angles are constant and multiplied by a constant value it stays linear. It could be possible to multiply angles with progressive factor depending of the node/point number.
It could be also possible to multiply with a factor depending on the sign of the angle using the cross product to know if positive or negative.

Permalink Reply by Daniel Piker on November 3, 2016 at 7:40am

Hi Arthur,

I changed a few things-

Firstly, for the angle between consecutive segments, the list was shifted, but without also taking the shortest list, which caused a problem with the last item.

Second - I scaled the whole thing up. To make the angle goal give numerically correct results for bending of elastic rods, there is a step in the calculation where the vectors are unitized, which can cause issues when the geometry is very small or very large. I'll try and post soon an alternative version which avoids this issue.

Finally - yes, the angle goal does not distinguish the sign of the angle, since it works in 3d. It effectively tries to keep one segment on a cone around the axis of the other segment, for which there is no positive or negative sense:

You could potentially make a custom angle goal which acted in a particular plane though, in which case it would be possible to distinguish between positive and negative.

Attachments: