Many questions about designing 3D curves for conceptual design

As a means to get comfortable working with curves in Rhino, I’m hoping to start “sketching” daily sneaker concepts using only curves. I know most sketch artists work start by drawing their concepts using main profiles in 2D sketches (pen/paper) and then airbrushing/coloring.

I’m trying to do something similar, but in 3D from the very beginning, by designing the sneakers using curve. This way, I can rotate at multiple angles, render out stroked lines, and then texture/colorize in another tool (like Photoshop/Illustrator).

Here are 2 reference images that are really close to what I’m hoping to start doing. Note, I have no idea what tool this was made in, but I know/believe this is possible to create in Rhino! :slight_smile:



The rendered part is the easy part for me. If I were given the strokes of the shoe, creating a textured 2D mockup is straight-forward.

Designing 3D curves like that in Rhino, though, is very difficult :slight_smile:

To get started, I’m trying to see if I can first create a Rhino reproduction of the curves in the first image above.

I’ve attached a file I started…

Sneaker.3dm (3.0 MB)

…and I’ve hit a dead-end wall right about here :slight_smile: I’ve got quite a few open questions. I’d be so grateful for any knowledge sharing on any of them. I’ve annotated the images in relation to my questions…

  1. This interesting ankle collar/flap wasn’t so hard to make for me. It’s the first shape I made. However, I instantly start getting anxious — my shape is clearly flat, but really it should curve somehow along the Y axis. Should I take all of my joined curves and bend/rotate them somehow? Would I do that with a cage edit, or with some other tool? Is cage edit overkill here? How else could I make the collar dip down in the middle as if gravity is acting on it? Should I have built this from the get-go using vertically bent curves?

  2. The bottom-right corner of this really kills me. I’m very used to using the Illustrator pen tool and manipulating handlebars until I’m really happy with shapes. The handlebar tool in Rhino doesn’t really work the same way. I made this corner shape by running BlendCrv on the back-right curve and the sole curve…. but fiddling with the settings feels imprecise and lackluster. I wonder if I should even be building this entire back profile yet? Should I instead be duplicating the bottom sole (offset upwards) and focusing more on the horizontal profiles of the sneaker…and working about the vertical curves later on?

  3. The centerline curve is also killing me. I basically drew a crappy control point curve connecting the sole to the ankle with a bunch of throwaway points in the middle…and then hand butchered them into something vaguely resembling a profile. Is this even remotely a good idea at this stage? Similar to the back profile, should I be focusing on these curves sooner, or later on in the process? Should I even have a single centerline like this, or should I have 2 parallel lines that are offset from the dead-center instead?

  4. This question is all about symmetry, or asymmetry. First off… is there even a “standard” foot profile to work with? Mine looks like a complete joke. It’s hard to tell how the curve should be built at all from the top view. And what about the front view? My front view looks very symmetrical, which seems like it would make it easy to work with….but that has to be unrealistic, right? Should the front view actually be asymmetrical as well?

  5. I’m basically wondering if I’m even building these curves in the right order. Just to see what would happen, I tried running Sweep2 on the centerlines, sole, and first ankle….and the resulting shoe surface looked like a complete joke. I know that I stated I’m really only concerned with building curves, so should I not even care what a resulting surface could look like? Is that just a distraction? Should I just keep hammering away making more and more curves?

  6. The bright white curve that spans the whole top surface of the shoe totally baffles me. I feel like if there were already a good shoe surface to work with, I could simply draw this shape in 2D, and then “project” it downwards onto the shoe and it might magically turn into this 3D profile. But I don’t have a surface to work with at this point… so should I be hand building this profile? Is it foolish to try and torture one long control point curve into this? It feels incredibly difficult to wrangle those into contours like this, with a mix of long/smooth and narrow/tight corners.

  7. Similar to the previous question, this “fan” shape on the lace area of the shoe also seems easy to handle if it were a projection down onto a surface… but if you’re doing it by hand, what’s the right way? Should I design this as a flat 2D profile first, and then somehow have it “flow” along the centerline curve of the shoe? Is there a command that’s ideal for this kind of transformation?

  8. This isn’t related to a specific part of the reference, just an open question. Am I thinking about this completely wrong? Is there some really intuitive better workflow for producing the critical contours/profiles of a shape that is both organic but also somewhat technically precise like this? Could I be relying on more tools/actions to ease this, or is it really all about expert manipulation of really long control point curves… in other words… using Rhino like a 3D version of Illustrator’s pen tool?

Thank you so much for any advice you’re willing to spare.

1 Like

It definitely is a sort of thing that people do, but you’re trying to over-do this. The stuff you’re finding hard to draw in 3D–DON’T DRAW IT IN 3D! What you’d be generating in 3D is an underlay for your hand render, which might also include simple surfaces. There’s no point doing anything more, since making a detailed wireframe that’s meant to be shaded into a sketch has virtually nothing to with actually building it in 3D(the curves would be garbage) and only really work from one angle.

Your goal images are a Rhino technical display mode of a 3D model, not a 3D wireframe, and a render of that 3D model not a “hand” drawing, not extensively tweaked anyway.

1 Like


I‘m not sure if you ever build a shape like this in 3d but what you experience is just the difference between a pseudo 3d sketch and challenges of building a 3d model out of it. Even if its just about the curves…

Multiple perspectives, means curves might look good from perspective A, but very bad from perspective B. Therefore you need to construct them, not sketching them.

Many curves, even „feature curves“ are not the result of a direct drawing in 3d, but because of surface intersections or curve to surface projections. I‘ve learned 3d surface modeling from people doing this for decades, and guess what. They not as often use 3d curves to define their surface outline as you might think.

You build theoretical shapes (main surfaces intersecting positional, which then getting blended in between) . If you do it right, many curves simply appear. I have no clue how you draw them correctly if you don‘t have the support of surface for this.

So yes, you are overcomplicating it like this. On the other hand your curves are really not as bad as you believe…

Thank you @TomTom and @JimCarruthers. If I understand you correctly, are you basically saying:

  1. My reference image was created with a different modeling technique (could be all surface modeling, or even sub-d from another tool) and therefore the feature curves I’m looking at are merely the byproduct of that model, but the model was originally not created FROM those curves.

  2. Designing complex curves like that from the ground up in 3D is a complicated and difficult task, and not really the ideal way to go about this, because managing the precise flow of curves so that they look good at every angle is very painful/difficult.

  3. Designing curves is probably best suited in simple 2D planes (top/front/side), and projecting them later on in the workflow onto established surfaces in order to create complex surface features…

Is that the gist of it?

If so, what WOULD be the ideal way to start approaching daily rapid-fire sneaker conceptual design in Rhino, where I’m focusing mostly on interesting shapes and “big idea” features / surface details?

Should I instead be trying to build and manipulate lots of surfaces, and then slowly connecting/blending those surfaces, and then finally designing and projecting 2D feature curves onto them, then performing trims/etc?

Yes. This hits the nail.

I‘m no product designer and I never created sketches in such a quality. However my experience with product designers (and their sketches) is, that the quality of curves and technical correctness is totally irrelevant and even often not 1to1 feasible.

A sketch is about understanding the design intent and the form itself, but nothing more. Its not a blueprint. It really doesn‘t matter if this sketch is drawn on paper, if its photograph of something similar or a 3d sub-d model. Even if you provide a Nurbs Surface Model I would remodel anything from scratch.

With that in mind I wouldn‘t care about perfect curves at all. I believe you could even draw it nice if you just provide very rough discontinuous 3d curves as a reference for your hand drawings.

I mean the whole point of sketching is to try different ideas before even starting to construct.

1 Like

This is a common approach if you want to model a quality shape. I‘m sure its an overkill to do this for sketching. The challenge is to place the surfaces in right order and reduce the amount of them to a bare minimum. Feature lines can directly draw or being projected or even created by surface intersections. It really requires experience and time to find a good solution when to take what, and which constraints do change what. Thats why you want a sketch before you do something like this before actually . You see the dilemma :wink: