Surfacing... just a nightmare

A phone camera cannot possible get enough light into the lens because it is so small… yet:
google maps can predict traffic in front of me. Garmins maps cant.
I can search and phone a store as I am driving down that street in a city I have never visited, in country whos language I cannot comprehend.

The point of this thread is not to suggest that class A surfaces ready for a major auto manufacture should be easy, but that producing a rendering for a group to discuss in a meeting should not require 3yrs training after what, a decade of advanced CAD software.

This is NOT a criticism of Rhino, more a wish for the future

Pros use pro equipment and pro techniques. Consumers use toys. Yes, some day you’ll be able to talk to an AI that will design your car for you, but even then, at least in the first stage, you’ll need to understand 3d curves and geometry to get good results. Will my kid maybe one day drive a car that was designed by an AI that just took directions from a barely informed human, like an architect does from a rich client? Maybe. Maybe that AI will even fit in a phone. But it seems like you came here to learn how to draw a car in the near future. Kurzweilian hypothesizing is maybe more of a distraction from that than an aid to it, no?

Google searches suggest Google employs around 40,000 software engineers; Samsung employs around 40,000 software engineers. In 2017 there were two billion active Android users. Samsung sell around 75 million smartphones per annum. McNeel should be so lucky.

The market for CAD software is minute by comparison; the resources available are tiny. It’s going to be a while before you can say “Alexa, draw me a Tesla pickup truck”…

it’s a trend!
Newby expect to have an iPhone giving them only pre made choices and Rhino3D keeps to be the most freedom modeling software ever.
Pro users are less and less and young guys are not ready or just used to spend time learning sw.

rhino should ask to give a tour at the first start (like video games). :rofl::rofl::rofl:

1 Like

The surfacing is the most challenging and the most difficult task in modeling regardless of used CAD software. Maybe in near future the AI will propose the suitable way of surfacing but I believe the final decision will always be up the user.
According to my personal experience I try several ways how to create surface.

Chuck Norris recently trained an a"i" smart enough to satify a designers intention.

I do not think there is any question about that, what I was hoping to hear was a discourse from the forum members about their general approaches and the concepts they use

There is no general approch and so there will be no magical tool solving the puzzle for you. The approach is iterative. Usually rough to detailed. Start with big, less curved surfaces, do the transitions and continue on fillets etc. . Keep controlpoint count low and work clean from beginning on. No cheating, ever. Precise modelling is the key. Prevent using weighted Nurbs, try using different approches, never be fully satified . Get an “How can I improve attitude”. Learn to actually analyse your shape correctly.


You can start from curves, or use 3x3 surfaces and increase the cp count. Before blending main surfaces, create a good and watertight “theory”. That means main surfaces should positional match (works in most cases), blending -> use G2 (curvature matching) for most blendings, sometime G1 is better sometimes even G3. Find a good surface/patch layout. This is iterative. So it often requires to do the same steps over and over. But only like this, you’ll be able to improve the visual quality. The more you do this, the better are your first guesses.


You may also use single or low-span surfaces, and equally distribute cps for each surface. Check the surface-overall curvature. Is it smooth, wavy or even jagged? Use Curvature Graph, Zebra, Scale1d and other tools to extend your eye. If your eye stops, there is usually a problem:

Read this as a starting point :slight_smile:


As soon as you build your own workflow (the 10 golden rules from the Alias website that Tom posted above are a great start) and find out some clever tricks, car surfacing with Rhino becomes a second nature.

Made with Rhino 4 in 2009:


nightmare | ˈnītˌmer |
• a person, thing, or situation that is very difficult to deal with: buying wine can be a nightmare if you don’t know enough about it.

perseverance | ˌpərsəˈvirəns |
• steadfastness in doing something despite difficulty or delay in achieving success

competence | ˈkämpədəns | (also competency | -tənsē | )
• the ability to do something successfully or efficiently: the players displayed varying degrees of competence.

satisfaction | ˌsadəsˈfakSH(ə)n |
• fulfillment of one’s wishes, expectations, or needs, or the pleasure derived from this

Just sayin’…

1 Like

can you post a Shaded view to see the surface layout? is always useful.

Sure. The majority of the body (before mirroring it) is shaped by a single, hand sculpted surface. This keeps it perfectly smooth. The roof, rear window and half of the trunk panel is a single surface made with “Sweep 2 rails”. The rear end has several transitional surfaces made with “Blend surface”. The transition between the roof and the windshield and door windows is “Sweep 2 rails”. After creation, every surface have been simplified with “Remove knots” and (if necessary) with “Refit surface to tolerance”. Then, “Match surface” was applied to each surface edge to improve the smoothness.


Thank you for sharing that . Very cool. If I could clarify…what is a single hand sculpted surface? I envision clay an scrapers that is then scanned into CAD with a 3D scanning tool…
How is it kept perfectly smooth?

The screenshot showing the isocurves is very insightful. Thanks very much for posting it. It looks as though the “single, hand sculpted surface” starts at the bonnet/hood centreline and was then pulled around to form the flank of the car, ignoring the glasshouse and terminating at or around the rear lights. Can you tell me the surface U & V degree and give an idea of the number of points? Why was a knot(s?) inserted over the front wheel arch?

Is that Patch that’s been used to form the wheel arch lips?

Sorry for the confusion. By “single, hand sculpted surface” I meant a single surface that was manually edited by moving the control points so that they get close to the desired shape, and later adding new rows of control points where it was needed to add further detail.

That’s right, the single sculpted surface starts from the middle of the front bumper (also Y axis = 0) and wraps around the entire side of the car, up until the middle of the rear bumper. For this particular car model, the surface consist 9 rows of control points vertically and 32 rows horizontally (along the length of the car). However, the starting surface was quite simple (something like 6 by 12 rows) and I added extra rows of control points with “Insert knot” only where I needed them most.

Usually, the process is like this:

  • From top view I draw a curve that closely follows the side shape of the car, then I extrude it vertically. I try to keep it simple at the beginning, so the curve has 12 control points that are the bare minimum to draw half of the the car’s shape from top view.

  • Then, I rebuild the extruded surface to add more horizontal rows of control points.

  • From side view, I also draw 4 guide curves (2 for the front wheel and 2 for the rear one) that will help me follow the inner and outer shape of the wheel arches.

  • Then I build a “Surface from network of curves” that was created by 4 other curves forming a rectangle. However, at the time I didn’t paid attention about reducing the control point count, hence the wheel arches lip consists so many control points. Big mistake. Maybe this is why you asked if that’s a “Patch” surface". In later years I started to use “Sweep 2 rail” instead. Then, the 2 arch curves at each wheel are used to trim the Network surface. That creates the wheel arch lip.

  • Now comes the fun time to manually move the control points of the extruded surface. Basically this is what refers as to “sculpting” the 3d shape.

  • The extra rows of control points above the front fenders were added later with “Insert knot” and serve the purpose of having a precise control over the shape of the car so that I can drag the control points to closely follow the wheel arch lip.

  • Once I’m satisfied by the resulting shape, I use the outer curves of the wheel arches to create “Pipe” and use it to split the sculpted surface and the Network surface that created the wheel arch lip. Then I connected both with a “Blend surface”.

  • Since the main sculpted surface was done, it’s was time for the rest of the body.
    “Sweep 2 rail” is used to create the surface for the roof and rear glass.
    The door windows are created by extruding an arc that was then split with 4 curves from side view.
    The front windshield is made by “Surface from network of curves” (again, it was a mistake that I leaved it stock and didn’t rebuild it to reduce the control point count).
    For each of the smooth transitions around the trunk and the B-pillar I used “Pipe” with variable diameter to create a cutting surface. Then “Blend surface” is used to fill the resulting gaps.
    The roof surface was split by a curve from to view, then “Sweep 2 rails” is used to make the transition between the roof and the door windows and front windshield.

  • The air duct of the engine cover is made by “Sweep 2 rail” followed by “Match surface”. Then the surface is trimmed.

  • The tail lights are made by a rectangular plane with 5x6 control points that was additionally shaped via manual sculpting. The excess was trimmed.