how would you nurbs - model this Handle. It s the handle of a (swiss) Barmix Hand-Blender.
I am interested in the approach you use - and hopefully learn some new approaches.
What’s your Strategy and patch layout?
Initial surfaces, secondary transitions, final transitions ? a single surface ?
The original / produced surfaces - at least of the older one i own - are quite bad. Most online-photos have photoshop-reflections.
The reference mesh is from a SubD Model (but i want a nurbs–strategy) - and i somehow did not hit the initial, intended bottom profile.
Interesting for patch layout teaching. It depends a lot on if one must strictly follow the mesh and imperfections of the existing design, or if one would design it from scratch.
From scratch, I’d probably do it so that one can run a chordal fillet where the red line is.
that s my intend
But i don t want to teach a mess.
the mesh-data is already somehow an interpretation - of how i think the initial design was meant. So the idea is to be close but not 100% match the mesh.
i think this will raise a “how to find the correct trimming positions / curves” questions. I tried a version like this and it failed - but i might miss something.
here is my first approach - tangent only - and still needing some polishing work… i tried to integrate the ruled / cone like surfaces:
did anybody else look at this challenge ?
i would like to have a nice, convincing solution to tell to my students - not only what i personally like the most)
my 10 min crack at it using a very minimal approach- I opted for one blend surface and some point editing. This could be refined further with more points, and some more intensive point editing.
This is the classic approach in Rhino 1, 2 etc. Offers the best possible smoothness (if you avoid adding extra knots) and unlimited possibilities for editing. I would do it either this way or using History-enabled “Loft” (Loose type) with editable curve profiles.
It’s a Swiss product, so I’m almost certain it was originally constructed. However, reverse engineering is always tricky and time consuming, particularly if one doesn’t have the object in one’s hands, to draw possible patch layouts on it; only a wobbly scan is available, so quite some guesswork. But, projecting possible patch curves from the side onto the mesh could be a way forward. From that, one can try to find the sections and theoreticals.
i think one of the challenge here is, that the basic design-concept uses tangent lines, arcs, ruled surfaces, cones… with some model-makers transitions. And to get it alive as Nurbs you somehow have to juggle a G2 that visually feels more like G1-fillets.
i think @Lagom 's approach is chaptering this very nicely.
CV-juggling
this is what i hoped at one of my first trials.
I followed 2 approaches:
set up a good old “simple sweep2”, extracting isocurves at Knots…edit curve and surface - refine a sweep2 surface - repeat
set up an initial loft and find some magic increasing the amount of CV’s
The hassle here is: a degree 3 Curve will need 4 CVs on one line to have a straight section. so i end up with 8x12 or 8x13 points - to mimic above “feels like fillet but hidden G2-Style”
but - my guess the straight / ruled (cone like) sections make it hard - as soon as they deviate from “nice and straight” - the result looks imprecise. Its much easier to have a surface that is either, obviously convex or concave (in each direction) - than being ruled (in one direction). (do you agree ?)
cone-like, ruled, straight
there are those cone like / ruled surfaces hidden (violett) and the green section is straight / parallel to x-Axis ;-(
not 100% sure on the small cone-like to the left.
As we neither have the original data, nor knowledge of how it was actually constructed, it could well be that everything was done only to G1, and then wangled and botched by the toolmaker. I am trying to find a compromise between the scan and a modern G2 approach. Right now, I am convinced that sculpting the object from a large surface by noodling CPs was not how it was constructed, but let’s see where this goes. I’m almost tempted to buy one to draw possible patch layouts on with a sharpie ; )
Sometimes, you might receive a pile o’ shite from a client, so working it out from this mesh, whether scan or noodled manually, is more challenging : )
what really helped me for this approach was the section in the center with a nice “arc-like” that fades out to a line. No real arcs: _rebuild 6 CVs Degree 5
i only used a few spots with g2 - many transitions are g1 only - especially where it “feels like radius / feels like fillet”
Very busy right now, but the start of a more “controlled/constructed” approach, as if it were to be designed, not reverse engineered. Projecting feature lines on the wobbly mesh helps a bit. bamix grip.3dm (4.5 MB)
