Problems connecting curves with the skin/loft nodes

[SnowballWar]

I am trying to to resample the number of loops of different types of quad meshes.
My process is 1. get the loops 2. extract the curves 3. resample them 4. Get a new mesh out of it.

But i am having problems, trying to connect the curves into a surface again. The skin node produces bad results ( see picture 4 )

The polyloft node is kinda working ( i have to tweak the u/v wrap ), but because it produces triangles, when i quadrangulate them i often get messy results. So is there a way to get quads out of the loft node, or  how to avoid the problems with the skin node ?

curveSkin.hip

Edited March 3, 2021 by SnowballWar

[dleonhardt]

To avoid this happening with the Skin SOP, you need to make sure that all your curves have the proper vertex order. Reversing the last curve should fix your problem.

[SnowballWar]

Yeah ... but the problem is that i am trying to create hda for maya that should work for all quad based models i apply it to. Maybe if i manually reverse the curve in this case it would work  but in any other case a different curve will pop up. So it will always require manual input ... which doesn't work for me. I want it to be automatic.

Basically what i am trying to do is to change the number of vertical or horizontal edge loops in a quad based model and convert them to curves. The closest thing i can come up with that somehow works is by using the subdivide and the get the horizontal or vertical loops and delete the ones that i don't need. The problems with this approach are

1. It always multiplies by 2

2. I can't reduce the number of loops that are already present.

I can't think of anything better as an option. If someone has an idea how to achieve the desired outcome without the skin node please do tell.

 

[Aizatulin]

Hi,

you can try to set primgroups based on the neighboured points. If you walk over both prims and taking the sum of the lengths of a set of point pairs with different directions, one direction should have a smaller sum (which should be prefered). Based on this information you can set the primgroup. Here is an (heuristic) example.

curveSkin_mod.hipnc

[SnowballWar]

11 hours ago, Aizatulin said:

Hi,

you can try to set primgroups based on the neighboured points. If you walk over both prims and taking the sum of the lengths of a set of point pairs with different directions, one direction should have a smaller sum (which should be prefered). Based on this information you can set the primgroup. Here is an (heuristic) example.

curveSkin_mod.hipnc

Thank you for the effort you put into this. 

I am still not very experienced with Houdini so i don't understand everything you did. 

This method seem to work with most of the models but it breaks when i apply it to the horizontal curves of the second test model. ( pCube2 )

 

But either way, i will play around with it and see if i can fix it somehow. Thank you for your help.

 

[anim]

this may help also https://www.sidefx.com/forum/topic/73762/#post-311733

[Aizatulin]

Here is another update with some comments. If the primitives are sorted in a right order, you can shift the vertices based on the differences between the u-values. This will probably only make sense, if the polygons are closed (or open with same first/last point ~ which is the case here). You can also look into the file from anims link, there is less VEX Code. 

curveSkin_modA.hipnc

[SnowballWar]

I am grateful for your help guys. Will look into it so i will understand it better. Again, thank you for your time.

Edited March 4, 2021 by SnowballWar