select curve by vectors

Replies to This Discussion

Permalink Reply by Ethan Gross on July 27, 2016 at 12:08pm

Hi Max, not quite sure if this is what you had in mind.

Attachments:

• Choose lines by direction.gh, 10 KB

Permalink Reply by Joseph Oster on July 27, 2016 at 1:49pm

The "Test vector" with "angle tolerance" is an interesting wrinkle to filter the lines.  I added some visual aids to help me see it better, including the blue "Test vector" in the corner.  Looks to me like you miss lines/vectors "aiming toward you", so to speak, within the same angle tolerance.

Chooselinesbydirection_2016July27a.gh

I applied your data to my code, posted earlier.  As vectors (preview disabled), all the blue lines point down, the yellow up - selectbyvector_d2_2016July27b.gh:

Attachments:

• Chooselinesbydirection_2016July27a.gh, 27 KB
• selectbyvector_d2_2016July27b.gh, 27 KB

Permalink Reply by Ethan Gross on July 27, 2016 at 2:51pm

Yes, you're right about my only looking in one direction. Here is my amended code.

Attachments:

• Choose lines by (both) directions.gh, 13 KB

Permalink Reply by Joseph Oster on July 27, 2016 at 12:37pm

You could start by providing some internalized data in a GH file.

The one you posted has no internalized data, not even a single point.  Better would be to provide a collection of lines to study for their directional vector properties, which can only be determined by assuming a vector between start and end points?

Given that (the 'Random Lines' param in attached code), I would answer your question this way:

Attached selectbyvector_2016July27b.gh shows lines or vectors:

Finally, selectbyvector_2016July27a.gh, the code I wrote to create the 'Random Lines' internalized data:

Attachments:

• selectbyvector_2016July27b.gh, 17 KB
• selectbyvector_2016July27a.gh, 16 KB

Permalink Reply by Max Rind on July 28, 2016 at 1:16am

Both are really great approaches, I still can't believe that you can solve such complex issues so fast.

Many thanks I got it done perfectly :)

Permalink Reply by Andrew Heumann on July 28, 2016 at 5:44am

I think something even simpler is possible! Whenever you want to compare vectors, Dot Product is your friend - with unitized inputs it will return a 0 for perpendicular vectors, a 1 for parallel vectors, -1 for anti-parallel vectors, and values between 0-1 for "pointing generally the same way." The nice thing is this works if the vector you want to test against ISN'T conveniently one of the cardinal axes. (And Joseph - I think you can generate random vectors a bit more easily as well :) )

Attachments:

• Vector Direction Test.gh, 6 KB