Uneven surface subdivision

Replies to This Discussion

Permalink Reply by Suryansh Chandra on January 6, 2010 at 11:52am

If your divisions are UV based, you could divide your main surface into sub surfaces (use the Isotrim component). This gives you smaller pieces of the surface, which you can divide further based on your attractors. The divisions in your reference image are perfectly achievable using UV divisions alone.. loft between multiple curves, make sure the ones at the ends are straight, and all the ones in between are warped. See this:

If you want non-UV divisions, your task gets a bit more complex. You need to make your points on the surface in an organic manner (somehow!), then make the 4 edges using InterpCrvOnSurface, and then script the EdgeSrf function yourself (because its not natively available yet, but this is not the toughest part) to get the little pieces, and then use an attractor to subdivide them again.

Permalink Reply by Luis Fraguada on January 6, 2010 at 2:06pm

This kind of subdivision could have some recursive logic. In any case, there are not more than 2-3 levels if further subdivision. The solution I post is basic (no frames, just the subdivided surface patches) but should be easy to follow. There is a group...each time you want to subdivide further, copy the group, and move it to the end. Connect the previous group´s output to the new group´s input.

To control the levels of subdivision more dynamically you could script it. As seen here:
http://www.livearchitecture.net/archives/3419

For some further subdivision strategies, try these:
http://www.livearchitecture.net/archives/4883

This last one takes an initial surface curved in 3D and meshes with custom mesh settings. This algorithm refines areas which need more definition. With weaverbird you can turn that mesh into a catmull-clark quad mesh... not the most balanced quads (compared to your initial image), and more a meshing strategy than a 1:1 scale subdivision.

Attachments:

• unSubDiv.zip, 48 KB

Permalink Reply by James Low on January 7, 2010 at 10:26am

hi Suryansh and Luis, thanks alot for your explanation and help. I'm still trying to understand the definition...

i've some more queries, how is it possible to define or reduce the size of the biggest subdivided cell? such that the difference between the different levels of subdivision are not too great.it's because i was hoping to achieve sth like the image posted earlier.

and also how do i trim the lofted surface to a desire boundary curve?

act i intent to create extruded boxes from these organically divided surfaces.
also want to ask how do i "delete" away boxes away from a series of extrude boxes, using an attractor system, such that i can achieve the effect as the image. like an "eroded" box.

man, sorry for these whole lot of newbie qns.. i'ill be very glad to hear from you guys.


best,
James

Attachments:

• fragmented grid.jpg, 644 KB

Permalink Reply by Patrick Bedarf on January 7, 2010 at 12:03pm

As for the culling part of it you could go for point or curve attractors and cull out breps by distance value. Some of my recent studies included subdiv logics and attractor driven solidification. Let me know if you need something maybe that helps.

patrick

Permalink Reply by James Low on January 7, 2010 at 1:10pm

hi a-ngine, wow your studies on urban massing distribution is really interesting to me.
i'm act trying to test out means for more variations and differentiations on urban form, density and parcellations. would it be possible to share your definition on that?
i'ld really appreciate it.

i'm still trying hard to pick up grasshopper, it's v difficult! i'll try out the culling part. Thanks for your advice!