Creating a surface

Hi Thanks for your post. I think this is addressing the problem I’ve had for a long time.
Could you help me with some of the points you mentioned (or maybe recommend tutorial if you know about something on this topic?)

Is this to say that you project a grid in one of the views? For example 5x5 and in the complex transition areas you split these squares into more segments…


I suppose you were refering to something like this instead(below) but I dont understand the process how to get there when you are working on a random shape not a simple sphere.

Why would I need to make a new cplane? Is it because the idea is that everytime you project a new square with new c plane?

Shortly, I dont understand how to approach laying out single span surfaces on the finished nurbs concept model. I understand that you need to project it and then rebuild it but what determines where to put them and what shape they are?

1 Like

Hi @Radovan_Horsky,

I’ll try to give you some inputs based on what I’ve understood from the julz method and also based on my own experience:

To project the curve network propperly you have to be used to the CPlane Toolbar commands, also to some of the Curve from object Toolbar like _Pull, _ExtractIsoCurve, _ShortPath (Geodesic curve) or _ExtendCrvOnSrf that is placed into the Extend Toolbar. Those tools should help you to make projections like the sphere one upon complex shapes.

Shortening, try to find the best point of view, create a CPlane (choosing the best tool for it) and project (choosing also the best tool). In the case of the sphere, if I would have to copy your curve network I would choose the top view to project a squared grid instead of choosing the front view in order to project a more complex arched network to get the same result. Once you’ve made the projection, as you said, you have to rebuild and maybe split your new lines in order to get the “clean” network.

Keep in mind that a 90º network does NOT mean a squared network, not even parallel. Just take a look at the polysurface on the @julz last post and notice how the intersections describe 90º corners while there is not even a single straight line (neither in any theorical projection view).

Finally, about how to know where to put them or what shape should the network have, I think that the answer is not that easy as we would like. I guess there are infinite possibilities to build a network, but there are several options that will behave really good until we finish our model, maybe just one option in some cases if we want to achieve class-A surfaces, but now that is escaping my understanding… So the answer to your final question is that we need to spend hours practicing and researching in order to train our building logic and our eyes. Try to find well-done networks around internet and understand why they are good, then try to apply that logic to your models.

Here some tutorials/webinars that could help. They are not exactly about this projecting method, but it’s shown a good modeling method for transition surfaces with G2 continuity.
· Creating and Controlling a Transitional Surface in Rhino3d
· Novedge Webinar #27: Advanced Rhino 3D
· Rhino3d Tutorial - Engine Cover - Modelling the Overall Form and Complex Corner Blend (1 of 3)

Maybe you can also find something useful in this thread.

Hope it helps!

2 Likes

Thanks Jordi,
I feel like you can keep practicing but with CAD if you run into an issue you just get stuck and I don’t really see how keep getting stuck will help you in any way. Perhaps as Tom said in your post the best way to learn is to be constantly asking your colleagues for help and learn from their advice. That is if you work somewhere of course :smiley: …and in my case I don’t. The only resource that I know of at the moment is this https://learningalias.co.uk (it is for alias but again its still nurbs and bezier).

I didn’t know about this one I will definitely check that out. Thanks…
“Creating and Controlling a Transitional Surface in Rhino3d”

1 Like

Hi Radovan,

Glad it could help. I understand you… I am alone at the office (in terms of CAD) and sometimes it’s hard to find solutions just by Google, it takes time to understand a new problem in order to find the right words and concepts to point a solution research.

Luckily we can also come here to share our doubts and issues… at the end they always use to be solved one way or another :muscle::muscle:

Have a good modeling!

Hi Radovan,
it’s clear that your curve network need a flow a kind of architecture if you prefer … from that simply lean on the overall shape of your rough NURBS.
-1 The silhouette (command _Silhouette ^^ may help for understanding)
once you get that flat silhouette on your cplane , cross the volume from one side to the other by projecting cross sections perpendicular to the left–>right flow him self and project all of that on your nurbs , rebuild and surface with the technic of your choice sweep, blend, loft, curve net work .
-2 the plugin “mesh2surface” mimic perfectly well the philosophy of 3D reverse engineering but replace the mesh by your nurbs instead,
but mainly it all about mapping your rough by small “clean” nurbs patch which have to get one degree in less than the control point …
—> in that way you get a 1 span surface (best quality possible in nurbs for instance: {4 ctls points, 3 degrees}, {6ctrl points 5degree} ect …
-3 cplane are just there for helping you to create strait curve (as much as possible) in perpendicular fashion to the main flow of your architecture .
but it’s a real job so you need to anticipate what you are going to do…

1 Like

Thank you very much… That cleared things up. I will try the trick with silhouette

Here’s my attempt. Sweep 1 simple. Boolean 2 objects. Pipe. Boolean spit. Surface blend positional both ends. Then curvature match both ends with preserve isocurve direction.

6 Likes

Wow!

As always. :slight_smile:

// Rolf