I hear that the BF in general may end up reworking on how they approach GSoC projects (like making it a bit more focused in terms of what Blender needs and not fancy marquee features that do not work so well in the end).
The issue is as mentioned before, the student might complete the project, but my no means does that equate to code of a high enough quality to get in master. The BF has approved hacky code for certain things in the distant past (<2.49b) and it only created a mess that would have to be cleaned up later (some of which is still in master).
Unfortunately for NURBS aficionados, it’s possible that better support for it will always be nothing more than a pipe-dream, but then again it seemed the way for proper GI for nearly 10 years until Brecht popped out of nowhere with Cycles.
I remember talking to Ton 2008 in LA at the SIGGRAPH and he complained about GSoC resulting more unfinished projects because of the nature of the short term contract.
If Jonathan has no time anymore - I can understand. The GSoC is not like a full time paid job.
But then again what’s the point to start something and the not to finish it. In this regards Ton mentioned he rather would wish not to do something and rather focus dev power on concentrated projects.
NURBS still is pretty important / useful but I feel the focus on too much SubD made it less attractive. Modo mesh fusion brought some subd booleans but the result is way to tessellated for any serious use besides concept modeling. People even pay an insane 500$ for getting NURBS into modo for rendering …
I don’t think that his work is pointless. His branch seemed to work well and, although it will lag behind master, it will not suddenly break all on its own. And it may even be in such a good shape that someone knowledgeable could peek at his code a revive the branch more easily than before. NURBS are advanced maths*, that’s the main issue I guess, so finding someone who can develop + maintain it is not easy.
Daniel
You can do a naïve BSpline implementation if a few lines of Python, but pushing that to BSpline Surfaces, NURBS, analytical normals, trimming, tesselation, animation, raytracing, etc… is another beast.
I think Sergey had some ambitions to work on the Nurbsside of the moon in the past, but i think there are other stuff he is doing now.
Nurbs must be the King discipline of coding :eek:
For me the Booelan Modifier does the job, alltough some filleting or chamfering on the cut/union/intersect regions would be damn nice.
But hey you dont get everything you wish.
I hope that this will not be abandoned once again. I recently sent a private message to jjoonathan (Jonathan deWerd), to send a document, by his answer I understood that nothing has been abandoned, or rather is thinking of supplementing with subdivision surfaces. These are his words:
“Maybe I should reconsider the possibility of integration opensubdiv.”
For years I had hope, though now I’ve learned to do many things with subdivision surfaces, there is no comparison with nurbs modeling of solid objects design.
Nothing is lost yet, before I was pessimistic but having seen at least of the changes this time tend to have hope.
I think it may be time to forget about Nurbs ever making it in (whether in 1 year or 10 years), as there’s few people here in the community who want to see full-fledged Nurbs support and even fewer in the development scene. With OpenSubDiv and the promise of adaptive resolution, interest might wane even further, and then there’s the promise of new boolean tech. like Mesh Fusion which can even allow good filleting with polygons and subdivision surfaces.
The more that Blender’s geometry tools advance, the less appeal that Nurbs have outside of more specific applications, it’s pretty much a non-starter at this point barring a surprise patch from a new developer.
Is it any wonder anymore why Ton and co. were looking at moving away from the idea of large GSoC projects altogether? The program in general has shown a number of limitations in terms of the final payout for Open Source projects, and it’s possible that replacing it with an effort to help on-board new developers (without any specific project assignments) will yield a far greater benefit.