Zebra has a problem with Blend Surface

Hello, when using the Blend Surface tool sometimes there is a problem with the analysis of Zebra when I add a new form. It when the new form strongly changes an initial surface… What in such cases to do?
2018-01-18_212530
2018-01-18_212447

I can’t tell if you’re seeing tangent matching at the surface edges, or a coarse analysis mesh.

Does the image change if you crank up the mesh to make it more dense?

No, the problem of a surface can be seen even without zebra.

Upload that polysurface please.

There is a simple example:
2018-01-18_221223
Untitled.3dm (628.8 KB)

I think what you’re seeing is a combination of a loose analysis mesh, and non-uniform isocurve distribution.

That said, I’m a tech support monkey. I do not model professionally.

One of the more experienced surface modeling people will need to jump in here I think

You do not indicate where you are adding the shape. Here, with no shapes added and a decently accurate analysis mesh, the blend looks pretty good. Without more details it’s hard to diagnose…

-Pascal

@pascal, @RichardZ, here it will be more clear…
Untitled.3dm (879.5 KB)
Red surface - where I have added a new form in the specified point.


Here I didn’t begin to add anything:

Additional forms spoil smoothness of a surface in difficult transitions… Sometimes it is very necessary to add an additional form for editing a surface, but such piece turns out - and what to do?

Here the difficult grid, any additional forms lead to problems.
1.3dm (88.0 KB)
2018-01-19_005822
2018-01-19_005842
2018-01-19_005924

I have no idea why you’re modeling these forms this way. There’s no need to do this with trimmed surfaces. You’ve also created surface conditions that make it impossible to create smooth pleasing forms.

3 Likes

Ok, means it is necessary to model so that Blend Surface could create then the correct surface quietly.
All thanks!

Looking at your highlights, you can see that you already started wrong.

Usually you start modelling with parts with less curvature involved. You can call these surfaces “base”-surfaces. If you don’t ensure they are perfectly smooth, you can never expect other parts of your model to be cleaner than them. There are dozens of strategies for clean modelling, but personally I start with 3x3 single-span surfaces (degree = cpcount -1 = Bezier), increasing degree up if needed, Making them join up g0. Such surfaces are perfectly smooth, since they are parabolic, or if of higher degree, they are at least controllable.
Parts representing higher curvature are blended in between your base-surfaces. Now the difficulty is to layout surfaces in a way that a shape can be represented with that. But this is practice only, so try again… :slight_smile:

2 Likes

Basic (the main, basic) surfaces have to be such that then it was possible to smooth, create additional geometry in addition to basic and her quality will be strong depends on basic geometry. It is necessary to think beforehand…

It would be desirable to learn your opinion…

I do difficult polysurface of 5 surfaces easier and that there was an opportunity to edit - it is impossible to add a set of spans, but then at Join there can be a shortage of spans and connection will have shortcomings - but it can be corrected through Render mesh quality.

If I correctly understand, then at increase in degree (after 3) - the quantity of control points without change of Isocurves is shirked. It somehow influences quality of a surface?

If you increase the degree of a single span/bezier surface you get another bezier surface with more controlpoints. You can also rebuild, however rebuild changes the shape, whereas increasing degree keeps the initial shape.

The reason single span is “better”, is that you control the curvature global, not just for parts.This keeps a surface much smoother when modified.The moment you modify a multispan/nurbs surface, it gets much harder to maintain a good curvature.

Well, in Rhino its kind of difficult to do pure Bezier modelling, however that is not important since you can do multispan, but your main surfaces should st least be single span or of low span count, because the smoother a surface is, the easier it actually gets later on. I work in automotive, and I can tell you that all visible surfaces of cars are Bezier only. So if someone can model cars in single-span, you can definitely model a ship hull in Bezier,(… although its difficult in Rhino).

Rendermesh: If you load in a manufactures final car model in Rhino, you don’t even need to render it, because even in simple opengl this modell will look much better as a model made by an amateur being rendered in an advanced renderengine. Why?Because the reflections/highlights are exactly where they should be and someone having an eye for this will always see if its made by someone professional or not.
So as a conclusion, model good and you won’t have pain in further processing, no matter if building in more detail, rendering, manufacturing or analysis. If its clean, its also simple to work with.

2 Likes

Autodesk made a good summary of golden rules:

Its cad platform independent…

1 Like

It is clear, but still it is possible to use InsertKnot - the form doesn’t change. :wink:

Interesting feature, I didn’t know… :grinning:

On the other hand the disadvantage of a higher degree single span surface compared to a lower degree multispan surface with the same number of control points is that moving any control point in a single span surface affects the entire surface, so local modifications to the surface are more difficult to achieve. The desirability of single span versus multispan depends in part on the importance of “smoothness” of the surface and the experiences of the user.

1 Like