Sweep 2 Rails creates crease on cross sections (boat hull)

Hello community!

I am dealing with quite a challenge with smoothly surfacing a set of curves for a boat hull (half). There are noticeable creases close to each cross section, like you can see in the picture.


My work around: First of all, after all experiments, Sweep2 seems the only command to respect the outlines of the hull (curve network and loft might solve better, but create slightly out of boundary shapes, especially at the forefoot/stem of the boat, loosing continuity). Anyway, the sweep rails are top and below (yellow - selected) and the cross sections are the rest of the lines. I also tried the Smooth command afterwards, but it destroys the hull. So questions are: what suggestions do you have for creating smooth lofts or sweeps? Am I approaching hull creation the wrong way?

Thanks!
work-decembrie5.3dm (576.5 KB)

i am not a hull surface specialist, there are quite a few here maybe the chime in at some point. what i could figure out so far is that the corresponding rail and section curves have different CP´s maybe you could rebuild that non uniform to match each other, that usually helps the flow. but at the same time you might have to check the fairness of your section curves which and again i am not a specialist but look somehow pretty messy. sorry if that is not much of a help for now.

@stevebaer i tried to check this file on rhino 7 for mac, but i had severe issues with rhino quitting non stop. it has 2 extra floating viewports open, after closing them rhino quit, i sent a report. also the viewport was completely unresponsive. keeping this file and opening a new empty file rhino worked, returning in the same session to this file it did not respond. i only got it working after rescuing the data in the end.

also, Rescue3dmFile should be working without having to open a new file at first imho.

1 Like

Thanks, it did help a little to improve curve fairness and having same amount of control points on rails. Otherwise, still some crease, especially on the second one from the bow. Curve Network seems to work much better now though!

Thanks, I can repeat this and have added it to our bugtracker at
https://mcneel.myjetbrains.com/youtrack/issue/RH-62052

@george.kudor Rounded forefoots can be problematic. This thread may be of interest: Boat hulls with smoothly curve stems - how to model

What is the source of the curves?

How close to the input curves does the surface need to be? What will the surface be used for?

1 Like

Hello - if you re-arrange the points on the section curves to have more even progression from one to the next, you can Loft them, get nice progressive curvature and come very close to the rails. Th

-Pascal

1 Like

Thanks david.

Well, indeed rounded forefoots seem a challenge. I tried splitting the forefoot before the rounded part, after, in the middle :stuck_out_tongue: Sweeped all directions, lofted up and down, tried going into negative space and trimming the hull half… but yeah, I will further study the topic you forwarded and hopefully not get into deeper problems.

As for your questions, the curves are just out of my imagination, as an exercise for boat hull designs (most probable for further fiberglass mold production). At some point I will need to respect the station sections with high accuracy. But for now, I just want to design something that feels realistic. I managed to have the best results with curve network (as in the photo attached).

But then I run into new philosophical questions… like these ideal shapes that are a nightmare to fillet and smoothen: for example the sharp mid-curve (rabbet) between the two hull pieces. I got a few white hairs unsuccessfully trying to fillet (with variable handles of course) such apparently simple surfaces. The boat with deadflat also doesn’t seem that realistic.

work-decembrie - hull attempts.3dm (848.4 KB)

Thanks Pascal! That seems quite neat, although I am scared of the results at the forefoot. I’ll try more!

Good practices when designing surfaces and curves:

  • Start with the minimum number of control points needed to capture the shape.
  • Work to the accuracy required.
  • Add control points only when required.
  • Try simplifying generated surfaces to see if the desired shape can be retained with ferw control points. Useful methods for simplifying surfaces include RemoveKnot, Rebuild, RebuildUV
3 Likes

Hi George,

You might want to take a look at the ‘Better Living Through CNC’ tutorials on Primary Surfacing on YouTube, in particular Episode 4 https://youtu.be/zX3Jd6s5Ls4

I watched this recently and, following the concepts espoused in that series, I made my first attempt at a hull, using your starting curves. I am quite pleased with the partially finished result:

My approach was to simplify your lines so that I could make edge-curve-based surfaces with minimal control points. I had one surface each for the aftward and forward halves of the hull, with the forward one based on bow and keel lines extended to meet (ie initially ignoring the curved forefoot). I matched the surfaces for curvature, then adjusted the control points to get the surface closer to your intermediate stations. Once I had that sorted I split off the bow with an isocurve through the start of the curve and made a new four-sided surface from the edges and curve, which I curvature matched back to the existing surface. This is kind of a “to get to your destination you might want to start from somewhere else” piece of guidance, but worth considering as it results in very clean surfaces.

Regards
Jeremy

1 Like

Thanks for the tutorial tip! That guy is a surface perfection guru! Definitely better shapes now.

1 Like

This bug was reported a few days ago as https://mcneel.myjetbrains.com/youtrack/issue/RH-61979 and was fixed in 7.3.

1 Like

A fix for this crash will come sooner than 7.3. @JohnM back-ported his fix to 7.2, which is currently the Release Candidate. Hopefully, if you are testing Release Candidates, the fix for this should :crossed_fingers: show up quite soon (but, Holiday season being what it is…)

RH-61979 is fixed in the latest Rhino 7 Service Release Candidate