Sort Items for Data Matching

I'm having a lot of issues with data matching. Basically I took a icosahedron and played with it. However, the stuff on each face of the icosahedron wants to match itself differently. So as you can see from my bloated definition, I had to explode and manipulate each branch/face.

From my geometry you can see there are pentagons and there are triangles. The yellow group is connecting the vertices to make likes, and the mint green group is making surfaces from those lines.

I had to explode the tree because the points and lines were all ordered differently for each face, and that made weird lofts, or the lines crisscross, so I had to go in and change numbers in the list item, or change the direction of lines in an attempt to normalize each face.

Please tell me there is a way to sort items in a list so that they are all the same, so that I don't have to explode it!

Attachments:

buildz thingy2.gh, 169 KB

Replies to This Discussion

Permalink Reply by martyn hogg on January 13, 2014 at 12:24pm

I am a long way off expert, but I think you can do this sort of thing in a few different ways.

Straigtforward lists of data can be manipulated using ShiftList, SortList etc.

If you have data trees, there are ways to remap the data (PathMapper etc) and there's an excellent tutorial by David Rutten about path mapper on this forum somewhere.

And always look at whether you simply need to flatten your data to ba able to work with it.

For point lists I often use the PointNumber component to help visualise the data and the good old Panel component helps too!

When you see some of the elegant, compact definitions on here, there often seems to be some mystical foresight needed right from the first component but hopefully this jedi skill comes with practice!

Permalink Reply by JS on January 14, 2014 at 6:30am

Hi Martyn,

I do think the sort list would be the most elegant approach. I started to work towards a solution here.

What I have is a list of 6 points on 20 branch, I don't want to flatten it. In order to sort them, I figured I would take the 0 item on each branch and find the distance from it to its neighbors.

I got 5 distances, then I added 0 to the list to get 6 distances.

I sorted the list of points and then tried to find the 0, 2, and 4 indices and connect them to the 1,3, and 5 indices respectively.

If everything work out, which it hasn't, then you would see complete pentagons with no overlapping lines.

Would anyone be willing to take a look at my definition (which I cleaned up! It's not big anymore!)

and let me know what's wrong?

Thank you!

Attachments:

unnamed.gh, 16 KB

Permalink Reply by Danny Boyes on January 14, 2014 at 7:05am

I'm not sure what your end result should be but this is how I would collect the points into Pentagon Groups

Attachments:

GroupingPentagons.gh, 18 KB

Permalink Reply by JS on January 14, 2014 at 12:48pm

Hi Danny!

Thank you for giving it a shot. That is not exactly what I wanted. I'm sorry for not being descriptive enough. It's not so much grouping them into pentagons, as much as making lines that relate to triangles. You can see the triangles up in the first image of the post.

I think I figured it out though. When I sorted the points by distance, there were duplicate distances (ex. 0,1,2,3,3,5). So grasshopper picked among the duplicate distances which led to variation in my sorting.

Permalink Reply by martyn hogg on January 14, 2014 at 11:57am

I'm on Rhino4 so I got some errors when i opened your file. My first thoughts were that the way that the point sets are numbered didn't seem to have a nice pattern to them, and I was going to suggest trying to start with more uniformly patterned point numbers to make sorting through them easier... i.e. numbered 0 to 4 clockwise around the pentagons.

But then Mr Boyes just went and solved it anyway!

Permalink Reply by Matt Gaydon on January 15, 2014 at 7:26am

Tried to solve you question over lunch as the bigger code interested me but could not open that file so looked at your second simpler question and in the process seem to develop a way to maybe create your picture you first posted.