T.Map > patched surfaces/mesh into even diagonal grid-lines/mesh ?

hi there Hoppers

i have bunch of surfaces(parabolic. designed with tspline) which i am struggling to re Mesh and retopo the mesh

so far i used the mesh machine and got random Tri-mesh result.

would love some help here

i know i am trying to do something like the Tmap, but the reason i wish to explore this issue in GH has to do with the flexibility of the triangle size. Tmap is very powerful but manual

thank you for help

Replies to This Discussion

Permalink Reply by AJ on May 20, 2016 at 4:47am

thnx for answering . the density was actually default of the T.map

i actually look for a nice mesh. and from the triangles to generate hex-grid. since its bunch of surfaces the lunch box components do not work. but dual mesh does but not clean since each surface has its own UV or domain. so the grids do not match in continuity of the grids.

Permalink Reply by AJ on May 20, 2016 at 5:11am

Permalink Reply by AJ on May 20, 2016 at 5:47am

i tried to remesh

here is what ive been up too

Attachments:

my question.gh, 68 KB

Permalink Reply by AJ on May 20, 2016 at 6:41am

i tried my approach to different objects. sometimes even manually rebuild the surfaces to correct the uv( in the first image. since i thought its too complicated, i dropped a easier 3d brep)

the bunny is amazing but it dosent give a nice clean grid. which is the main goal. to reach something "like" the Tmap.

im looking into the surface mapping u suggested. can you direct me a little regarding to brep and multiple surfaces..

cheers

ps while looking around i just came up with this.

http://www.grasshopper3d.com/photo/conformal-map-via-circle-packing

the image on top left could be it, if it could work on few surface "as if" they are one

Permalink Reply by David Stasiuk on May 22, 2016 at 3:48pm

Since hexes area requirement, it's easiest to start with them as the base topology, but of course you're running into trouble because you're starting from polysurfaces that don't map well to each other. However, because your basic form is mostly convex and has one naked boundary, it's possible to create a rough-approximation patch from it, and apply a hex-grid using Lunchbox:

The first thing then is to cut off the hexes around the boundary trim, and then to create a clean mesh from the results. The edge conditions are quite tricky, but they can be managed:

Finally, you want to refine the geometry...the patch itself really doesn't get all of the curvature of your underlying geometry properly captured. So the last part takes the rough mesh and uses Kangaroo to pull all of the vertices to the main geometry while smoothing it all out. It takes a little while to run, but ends up with pretty good results. You will always end up with some irregular triangulated moments along the edges, but they are very limited, and all of the internal vertices are valence 6:

The definition is using a bunch of extras...Kangaroo2, Weaverbird, MeshEdit, and Lunchbox.

Attachments:

hex_mesh.gh, 86 KB

Permalink Reply by AJ on May 23, 2016 at 2:31am

hi there David. thanx for answering and helping. the result looks great. i have 1. Solution exception:Method not found: 'Int32 KangarooSolver.PhysicalSystem.GetIterations()'. ERROR. tried rolling back the NET from 461 to 451 but not working still same error.

another question. since i had the same idea as for the patch. but this generates the Lunchbox hex grid as a deformed pattern. is there a way to control the grid as tight as possible on the surfaces that are the collective brep?

Permalink Reply by David Stasiuk on May 23, 2016 at 4:58am

Which version of Kangaroo2 are you using? If you replace the components in the def with whatever working version you have, it should be okay.

Also, I'm not quite sure what you mean about a "deformed pattern?" The problem with trying to map specific patterns evenly over polysurfaces is that you are working with multiple, discrete UV grids. If what you want are hexagons of more exactly even size, then you'll have to tinker with your x/y hex counts.

Alternatively, I also just remembered that Starling has a really nice mesh-mapping capability, where you can map curves from 2D space onto a Genus 0 disc (like your target brep). This could be a good way for you to have (somewhat more) direct control over the size and orientation of the hexes you map, but you'll still almost certainly have to run some smoothing passes to achieve it. Check out Starling on food4rhino...there's an example there for curve mapping...

Permalink Reply by AJ on May 23, 2016 at 5:34am

the kangaroo is update 2.0, unfortunately the solver give me the error time after time.

since it didn't work, that's why i referred to the grid as not even.

i wish to have a clean grid, valence 6 type. regardless to the borders at this point,. is there a way to do this without the lunchbox component? i use MeshMachine+ WbDual. the point distribution over the new mesh is not in order (not valence 6)

ps. from some reason i cant upload the image

Permalink Reply by AJ on May 23, 2016 at 5:47am

actually . i had to roll back kangaroo 2.14 back to 2.12 since there were bunch of errors and in the 2.14 version i could not use the mesh machine

Permalink Reply by AJ on May 23, 2016 at 6:09am

your def works.

my bad

RSS