@sciensman:
The first biggest advantage of NURBS over Subd is that polygons in subd should meet on vertices and they share full-length edges. NURBS patches may connect in the middle of the edge and are connected together if they are close enough (what may look as “messy” wires because it`s not so visual tidy as in subd). That is an advantage in building patches but may be a disadvantage during moving NURBS from one soft to another (what is a solid in one soft maybe not be solid in the other with different settings of tolerance). Closed subd into solid will be always watertight in every software.
Below I showed subd items which I converted to NURBS only to do NURBS blend over it. So converting few separate items from subd to NURBS may be also used to use the blend tool (which works only with surfaces - it doesn’t work with subd). You can`t do that type of bridge/blend in Subd because as I said before polygons should be connected sharing their edges (so polygon bridge should have same count of edges over both side). NURBS can smoothly connect them basing on a “close enough” position. So this is an example of why mixed workflow is nice IMHO
This was only an example. In that specific situation, you are right (when both sides have the same count of edges but you change your topology of both sides. In more complex situations you have to rebuild topology to blend two parts together (to meet the same count of edges on both blended parts which is not easy without disturbing present topology, which can change lighting drastically. My way do not change the lighting of current objects only adds blend to them. Their topology is untouched (so shading is also untouched). This allows you to blend also with a much bigger radius. When you want to come back to subd you can do always manual retopo over it.
not for dispute, I just showed the method of mixing!
Here, the torus has no edge at all, it takes them from the cube, the same method can be used in other cases. There is no need to adjust the topology for this reason, and the topology is not distorted in any way.
SubD allows you to mix with any radius, any arbitrary radius in real time.
Everything is doable in pure subd manually. I’m only saying that after working few years making subd meshes for 3d print when I`ve met mixed workflow, it was for me a holy grail because I am not limited to the same polygon count (edges count) on both sides of the blend. So both topologies may only be described by themselves without disturbing polygon flow (I think about more complex shapes than torus and cube of course).
I understand you, this is the method of mixing so that the number of edges coincides, one of the elements takes the right number of edges from another element. Allows you to bypass your holy grail, without going to NURBS.
I use it from time to time when I have to blend shapes and I don’t want to change their existing topologies. In 3d print, you won’t see the difference. It`s always better to rethink topology to be optimal but not always you have enough time for it.
That’s right, when a complex SubD object starts from one simple part, and as a living tree according to the scheme - root, trunk, branches and leaves, the optimal topology of the part is obtained.
You can go the other way, everything can be done in NURBS (so even faster), and feed the QuadRemesh grid, you will also get a topology, but not optimal.
SubD surfaces always have zero curvature normal to creases and edges, which can be a significant limitation for some models. Refining the mesh close to the edge can reduce the area of low curvature but not eliminated it.
In contrast a NURBS surface can have any curvature normal to a crease or edge.
My understanding is this is inherent in how the SubD math works. I recall previously posting about this but cannot find the thread.
somewhere around 1:59… you do an operation … align to get it close to the shape of the circle… as you show there is still a bit of gap or mismatch around the curve… then not long after that… you run a second command… or do something… that gets it like dead on the curve… what command or operation did you run?.. did you run align to curve a second time?.. mine always has a mismatch… how did you get it spot on?..
awesome… no prob… guess i was just wishing… i typ do those operations after exporting to nurbs… but guess i keep running across this periodically … hoping for the best… when a perfect circle becomes a reality…
It’s mathematically impossible to create perfect circles in SubD. We do a lot of SubD work for production/tooling, and in those cases we convert the SubD models to burns and we add in. It a all the perfect holes as trims/booleans.
I wish it was a way to to this with live booleans where you can edit the main SubD and see the booleaned result with the Nurbs holes applied. We can do this with Grasshopper but it gets extremely slow really fast.
This is another reason why I hope Rhino would implement a high quality voxel engine to preview this stuff fast, and then bake that as is for meshing/rendering/3D printing, or to a slow bake of the proper SubD+Nurbs operands to obtain a valid/clean/mechanically accurate polysurface.
