"region union" and "section brep-plane" problem

hi guys,

i have a set of parallel planes on top of which are curve sets that i wish to combine into one closed curve on each plane. "region union" would be my weapon of choice here. unfortunately, it doesn't do what i want. i shrinked the problem into a lean definition. pls find attached. solid union the brep before sectioning is not an option for performance reasons.
feeding all sections in the "region union" will result in one region that encloses all curves on all plane.

----------> OK, while explaining i found a partial solution. all curve on one plane have to be in one path. will work then.

which brings me to the next problem:
i am sectioning a couple of boxes in e.g 3 levels with 3 plane. the "section brep-plane" component will not maintain the path structure. or is it me?

could it be done like this?
brep(pathONE)tested against plane(pathONE)
->output result(pathONE)
...
resulting in 3 paths.

instead i get one path for each found intersection which renders the result useless for the problem described above.

any help highly appreciated,
best,
frank

Attachments:

RegionUnionTest_01.ghx, 116 KB
SectionPath_problem.jpg, 65 KB

Replies to This Discussion

Permalink Reply by FrankS on October 15, 2009 at 11:25am

just found out the path structure in the second example wasn't matching. still, correcting this didn't improve the situation.

is there a way to "section brep-plane" work on path-by-path basis?
is this maybe a vb.net task?

thanks,
frank

Attachments:

SectionPath_problem_02.jpg, 44 KB

Permalink Reply by Giulio Piacentino on October 19, 2009 at 5:18am

Hi Frank,

I understand your need to construct a "waffle structure". I am sending two continuations of your design, which I believe should help you continue with this task. The first one will show an alternative approach which should be a bit simpler (boolean union before constructing the intersection lines), the second one simply uses a couple of tree operations to carry on.

I hope this helps,

- Giulio
______________
Giulio Piacentino - giulio@mcneel.com
McNeel Europe - Barcelona

Attachments:

RegionUnionTest_01-001.ghx, 93 KB
RegionUnionTest_01-002.ghx, 182 KB

Permalink Reply by FrankS on October 19, 2009 at 7:21am

hey giulio,
thanks a lot for your answer. i attached a screenshot to better explain the problem.

green: the input brep
red: cubes created on a 3d-point grid, filtered by "inside the brep" condition
blue: section planes

i will have to section up to thousands cubes with a changing number of section planes.

solid union is not going to be much fun, if working at all due to the sheer amount of cubes (which are not intersecting, but only share co-planar faces...).

do you have an idea how to adapt the path-based approach you provided to changing and potentially big numbers of cubes/planes?

the aim is to obtain one closed outline section curve for each section plane.

thanks again,
frank

Permalink Reply by Giulio Piacentino on October 19, 2009 at 8:40am

Hey Frank, I think it is possible to do this in the way you ask, but I will be able of testing your case only if you post it on the forum. For now, I am attaching a definition showing a way to re-construct the tree: I believe it might be useful in your case.

I hope this helps,
- Giulio
______________
Giulio Piacentino - giulio@mcneel.com
McNeel Europe - Barcelona

Attachments:

planar-test.3dm, 363 KB
planar-test.ghx, 103 KB

Permalink Reply by FrankS on October 21, 2009 at 3:16am

hey giulio,

thanks a lot for your help. the replace component definitely does the job. never uses that one before.
i have got the definition working producing sections on the yz-plane now. dealing with the second orientation (xz-plane) in an elegant fashion will be the next challenge. i need to provide a constistent layer structure for the second "section" operation. the way the grid is set up provides a correct path structure for one section plane orientation only.

anyway: my major problem is solved, for now... pls find the definition attached.
i didn't post it before cos it was a bit messy. i tried to boil the core problem down into a smaller chunk and uploaded more explanatory files.

BWT: is there a more elegant way to create a 3d-point grid? may this a good job for a script or a new item on the wishlist ;-)

thanks again,
frank

Attachments: