Hi,

I'm trying to add all the volumes in the attached obj to 1 solid.


Each of the parts is solid. I use an assemble SOP to nicely separate them by connectivity. Then I feed the volumes into the boolean SOP, but the resulting mesh still has interior faces.

I'm sure the coplanarity of some faces might cause some issues, but as the volumes are well-defined solids, I assume this should work…

[[
Using a VDB to add up the volumes works well enough, but I really need a resulting topology as close to the original as possible.
Maybe SideFx could implement a guide-geometry input (e.g. derived via VDBs) to the boolean SOP that allows the user to give a rough estimate on how the result should look like ..
]]

Any inputs on how I can make this work?

Thanks!
m.

You do have groups inside your OBJ file. In this image I used the Delete SOP to isolate each group then recombine them via a series of boolean operation.

first of all your poly normals are flipped, just put Reverse SOP before boolean (no need for Assemble SOP) and you may want to treat it as solid or whatever works for your needs as I see you were using Assemble SOP to create isolated surfaces (Cusp Edges On) not sure that is an intention

hi guys!

many thanks for the inputs, especially the normals!!

I'll play with this.

hi again ..

I have tracked the issue down and it seems the rotation of ‘touching’ volumes causes an issue where the boolean operation sometimes works, sometimes sadly not. The result should ‘add’ the two volumes to a single solid.

Please see the attached simple example. edit the y rotation in the red transformation node.

by ‘definition’ of the node, the 2 boxes should touch each other perfectly, but at some rotations, it seems the boolean SOP has a precision issue with the coplanarity of 2 primitives.

Any inputs that could help with this? (Scaling the objects so they ‘actually intersect’ is not feasible for my workflow, as it introduces other issues.)