Closest points from sets of curves

I hope this image is self-explanatory. I can't Flatten because I need to keep points in their indexes so I can use these indexes later on. But I am having a hard time making Crv CP search for all of the curve in each main branch for each point. Help?

Replies to This Discussion

Permalink Reply by Pedro de Azambuja Varela on June 12, 2013 at 2:57am

I hope I am seeing something wrong. Please direct me.

Check the two groups I created and their respective vector vis, so you can understand what I am talking about.

Attachments:

curveCP04_ds_va.gh, 158 KB

Permalink Reply by David Stasiuk on June 12, 2013 at 5:00am

I think I begin to see what you want to do, and unfortunately I think you might be going about it the wrong way. What happens, for example, when one of your sub-curves straddles two lower curves? Or you have an open curve descending to a closed curve?

If you look at the attached, even when the data matching is done correctly and your original points are sorted according to the t values along the target curves (which is what orders the shattered results), the results aren't quite what you're looking for. The start/end points of any closed target curves have a lot to do with this, as does the issue of the points along a single curve jumping to a level where there are multiple curves. It may be possible to get the results you're looking for this way, but it seems more problematic than it may be worth...

Permalink Reply by David Stasiuk on June 12, 2013 at 5:06am

not sure if I had the most recent edit before:

Attachments:

curveCP04_ds_va2.gh, 95 KB

Permalink Reply by Pedro de Azambuja Varela on June 12, 2013 at 6:03am

David, Abbondio, all,

I think that at this point, I might as well show you what this is all about. (I'm afraid that I should've did this from the very beginning but I thought it would be a problem easier to tackle).

My algorithm goes as follows:

1-Create level curves (GeneCurves)

2-Shatter these (GeneShatter)

3-Find GeneShatter counterparts (2Shatter) in GeneCurves a level below

3.a-Get endpoints of shatters, get closest points in immediately lower GeneCurves, and shatter these.

4-Create a surface between GeneShatter and 2Shatter.

and then I will also vake it volumetric, and number them and so on, but I think that's not needed here.

Being this said: I feel like I am almost there, but as you see, there are still mistakes.

Attachments:

ShatterDown.gh, 216 KB

Permalink Reply by David Stasiuk on June 12, 2013 at 6:26am

This should be doable, I think, with one notable one issue: how do you want to manage curve segments whose start and end points don't project to the same curve below?

There are a few of these segments that you can see will do that here:

Permalink Reply by David Stasiuk on June 12, 2013 at 7:33am

okay, here you go: hopefully will simplify things for you a bit. I've rather changed the approach, so rather than using shatter, you use sub-curves. It's much, much easier to keep the data structures hitched together this way. Basically, you test for both the start and end points of your initial segments, and if they both are closest to the same curve, you pass them through. Then you take evaluate the target curve by the start t and end t, and get a sub curve of these. there's a trick to managing when the seam of the new target curve is between the start and end points, which is that you take two sub curves there and join them. then you can build your surfaces however you like...here I've done surface from edges.

All that said, I don't know what approach you'd want to take when your initial segments want to come down to two different target curves...which is why I just eliminated those instances from the results.

Attachments:

ShatterDown_ds.gh, 85 KB

Permalink Reply by Pedro de Azambuja Varela on June 12, 2013 at 8:13am

I can't seem to reply directly to the post by David Stasiuk ( http://www.grasshopper3d.com/xn/detail/2985220:Comment:871031 ) so I'll do it here.

David: your definition seems to work very well, thank you a lot. You've rewritten so much that I'll have to "study" its flow. Your tree components seem very empowering.

Regarding the "holes", those are a defect on the primary thinking, I'm aware.

If possible, I'll post final results afterwards.

Permalink Reply by Danny Boyes on June 12, 2013 at 8:15am

That's because there has been too many indentations. Most people tend to click on the previous reply in the level above.... but this tends to get tedious. I think you've done the right thing and started over :)

Permalink Reply by Pedro de Azambuja Varela on July 7, 2013 at 10:36am

As I promised, here is a picture of the final thing. It is not a complete algorithm, as some partes of the dome are missing, but it is rather complete as in what relates to parametric design.