Just started playing with this today, and have stumbled across an issue.
I have used the Sub D workflow to create some creases in a cloth material I am creating. I have now taken this to nurbs, so that I can project onto it, but the whole surface is split up into sub D squares. Is there anyway to merge / convert back into a single surface I can project on to ? Or will I have to rebuild the geo using contours ? I am unable to use extraisocurve, as this just limits it to the square. Thanks
There’s work being done to convert SubD to NURBS with larger surface patches but that isn’t in the WIP yet. In the meantime, you are correct, if you need a single surface, model it using edge loops from the SubD you don’t need to convert to NURBS first. I use DupEdge on the SubD > Join for the appropriate curve layout and then NetworkSrf for instance.
I would try the new Rhino 7 this week. We added the ability to “pack” in the ToNurbs command. Even though it is early for the technology it is already showing simplified conversions.
This is already so good.
from 600 quads to 3 Srf !
only comment, [perhaps it’s in the pipeline…?] is that it would be best to have the Nurbs seam on the axis line in a symmetrical object like the one in this example.
I realize this is another manual step, but you can move the seams by exploding the geometry and running the SrfSeam command. When you’re done, the seams will all be lined up the way you like them.
This simple ring imported to Rhino from a Zbrush dens mesh with 700000 faces > quadremesh to just under 2000 SubD faces [which seems to be the top limit for Packed NURBS conversion…?]
It did create a couple of strange Srfs [highlighted in picture] that needs to be remade.
Other then that it so great, to have the ability to quickly get [relatively simple] mesh geometry back into rhino Nurbs
Not sure what the limit is. I just did a 21,600 face subD to Nurbs Packed. Seems to work fine. Rhino does throw up a warning, but that is based on the old way to ToNurbs.
Update: just did a 48,486 face subD toNURBs packed. Took some time, but with a little patience.
Hi @scottd I tried it on a 30k open mesh and I couldn’t get the NURBs to pack. I unfortunately can’t share that model, but I’ve created a simple test model that seems to break it. I’m not sure if open edges are more likely to create issues?
When this feature has been well tested for the part of building the large NURBS patches from a face pack, we are planning on improving the packing algorithm (the part that decides where the boundary of a face pack lies in the SubD). That would include improving the symmetry of the face packs, and possibly allow for some user guidance on pack boundaries.
Sure I sent to you using apple drop [a bit too large for attachment].
it also has the original high poly mesh and a symmetrical + non symmetrical subD and Nurbs results.
I’m also wondering what for this particular sample model, packing with a subD density of ever 2000 would not work [it will convert without packing] even as @scottd has tested successfully with far higher subD count.
Can this be a Mac limitation? or something to do with the particular geometry in this ring?
*it sent the file, but gave a warning so please let me know if you got it.
The small green faces on either side of the ring’s central mount do look a little strange but they match the topology and geometry of the SubD. Here it is compared to the SubD with 2 levels of subdivision so the flow of the faces is clearer:
Even without face packing, the NURBS faces will look the same. Comparing to the original mesh, I think it’s at the SubD step that this warped detail should be removed. Maybe something like one of these?
Yes, it make sense to do a touch of manual cleaning, to the SubD Quadremesher constructed out of the input dens mesh.
I was trying to understand the possibilities for a fast [and automated idealy] way to get from a Zbrush part to a rhino Nurbs.
The above example was for very simple geometry.
Next question [in a back and forth modeling between Rhino and Zbrush] is about the level of [sculpted] details that can be retained in the conversion…?
There is no hard limit to how many faces can be packed together, in that command. If you can make a SubD with that many faces, and a Nurbs surface with enough CVs to match it, it should be packable. So even complex models with a high level of detail can work.
I’m not super knowledgeable on the (Zbrush >) Quadremesh > SubD workflow. MAybe @scottd or @theoutside have more ideas on automating it, and how to choose the right level of detail with it?
Could someone post the cv hull of an advanced packed result?
Also, SubD converted surface transitions aren’t always the best… with the larger patches in this new technique, overall it seems as there will be an improvement that is very interesting, but will edges become even more visible now?
(I long for the day you can go from SubD to tooling for injection molded plastic products…)
