# 6 edges blend

#21

#22

The solution proposed by MOI will not be perfect, but it’s always a good try with just a few mouse clicks. Rhino, meanwhile, sleeps …

#23

Method is just to use the multiblend tool.
Continuity is not perfect, but it’s a good starting point.

@ec2638 , you right, sound interresting to see Xnurbs on this kind of shape.

Moi-MultiBlend.3dm (1.1 MB)

#24

Thanks, nice tool, but NEED G2!!! Above I see it’s mathematically impossible)))
XNurbs looks interesting but I have’t SW(((
Anyone know similar solution?
May be is possible to create these curves and use the Network?

#25

VSR tries to make Degree 7 bezier surfaces to patch in the opening. It’s pretty lumpy with naked edges along the outer perimeter. Moi does a smoother job with degree 3 non bezier but with bigger continuity problems.

#26

I tried to make the surface of the curves. I used Adjustable Curve Blend G2. I think the solution is somewhere near. The yellow curve should be the base one.

The file:
6_edges_blend-1.3dm (152.1 KB)

#27

I don’t understand why you’re bothering to try this? You’re not going to get an acceptable result. You’ve been shown a much better option, that will take less time and will produce a good result.

#28

If I’m not mistaken, the VSR “multiblend” is an acceptable solution. Rhino should look in this direction, not only with this tool, but also for the blend surface, etc.

(Currently, Rhino is the only modeler that does not have tools to generate complex surfaces (multisweep, multiblend, Y branch surface, SubD …). Even parametric modelers do everything and even better!).
In my opinion it has remained a little behind, stopped at the first versions! This is a fact, undeniable!

#30

I found an acceptable solution for n-sided multi blend surface:

1. Build blue curve between two corners.

2.Blend two edges G2.

1. Cut the surface blue curve

2. Extract isocurve green curves and blend G1 red curves

3. Create network G2. Red curves does not intersected, but the surface created betwen them.

#31

How did you decide on the first curve?

#32

AKA…hoop jump! I can crawl on my knees to the market too.

‘A’ for educational effort however! Not your fault. You are teaching yourself means to ends, with available tools.

#33

Approximately. Snap to corners and move points in planar view.

#34

A good way to make that curve would be to use BlendSrf to make a surface from edges A to C (Zoom in and use the sliders to match the arc at the bottom) Then extract the top edge (that will be the blue curve) and delete the BlendSrf,

#35

#36

VSR multiblend is problematic. It struggles to deal with trimmed surfaces. Even then the shape tends to flatten out, even with control over the intersection. I gave up on it once I realized it’s limitations. Fusion 360 does a much better job with theirs.

(Tom) #37

In my opinion “multiblends” are no acceptable solution at all. Although I have the highest respect on VSR Tools, I don’t believe this particular functionality is good at all. It can’t work well, because at its centerpoint, it is nearly impossible to match g2. Since at this point there are no edges aligned, you basically deal with different principal curvatures to get matched together. This practically only works for situations with only very few curvature involved or by strong symmetry involved. You could also increase matching if you pull cvs at the centerpoint using multispan. But then you get bad overall curvature. I think the t-spline to Nurbs conversion does that, resulting in bad overall curvature models. Only solution is to use multiple surface with trimming involved. The last solution is one workaround, but I still don’t get why not doing it like this:

#38

It would be most useful to see more of the model, meaning how the user has arrived at these oddball surfaces that then are the cause for a 6-sided situation. Even on Magis’ Easy Chair, which was fairly intricate at the time, there is no such situation present that would make the tooling surface model that involved.

#39

Hello,

@Nail, After untrimming your surfaces to get a better idea of what it seems you’re attempting to model I think you may get further ahead by simplifying your input surfaces and trying not to use those trimmed surfaces as inputs. The sphere is no too problematic but that swept profile looks packed with a ton of control points that probably aren’t needed.

The problem with multi blend in my opinion is that you need a perfect storm of conditions in order to get a passable result, it’s not really a great path to go down especially if you need to modify those surfaces at a later point.

(Przemysław Doliwa) #40

This chair blowed my mind - is it your model i’d like to see how it is made in rhino

#41

That’s not a rendering from a Rhino model. It’s a photograph of the original, designed by Jerzy Seymour for Magis in 2004. It’s a good archetype for G2 surface modelling - even though it may look very difficult at first, the patch layout is fairly straightforward after analysing what is going on and knowing where to part the injection moulding tools, which is why it is surprising to see the OP’s 6-sided situation (this chair having none of these). With more context from the OP, things could probably be done differently. It is also a good example that, even when the original CAD model is “perfect”, the injection moulding process/cooling introduces quirks in certain places.

Very often I find it really comes down to finding a suitable patch layout/shape decomposition before storming head-first into CAD.