Nike Sole Pattern_Noise generator/ field charges

it’s my first post to the forum.

I want to recreate a similar form as seen on this nike outsole shoe.
First thought was to use magnetic forces/ field charges to manipulate points on a mesh. But didn’t get a grip on how to use these vector forces.
Then I stumbled upon a tutorial on Entagma for Houdini FX ( where I saw a similar form appearing and managed to translate that principal into GH.

The pattern that I am chasing here is basically dictated with noise algorithm (I have used plugin from Gulio “4d noise”). But I am struggling to understand what is actually dictating this form.

My question is: How can I have more control over form like that, as to say, how to manually set these focal points of the pattern as seen on nike? What I think is that the noise can only get me so far. For full control over form I think I have to get back to field charges. Any other ideas on how to produce this displacement?

I am attaching very simple .gh file with noise displacement and image of the result.
Final mesh was optimized using Zbrush.

intricate pattern_noise (9.0 KB)


Michael Pryor works at Nike

You have a surface and points on it and for each point a number of elevation (not the final result, but a function, which can be the distance to some points (as in the outsole shoe) or perlin noise or value generator depending on the position). Normally this height map is fluid and not wavy, but this effect is achieved, among other ways, passing its values by a sinusoidal function.

Then, try changing the component that generates perlin noise to the Pull Point (or other) component, and use other points (or other geometry) to play with the distances. Scale those distances (or deforms them using powers or logarithms or whatever) and then pass those values through the sine function. Then scale those values back to an amplitude suitable for your model.

1 Like

Can’t say anything specific about that for obvious reasons, but generally factories in most mass production fields need clean and editable nurbs geometry, a super dense mesh displacement is nice looking but pretty useless :smiley: (even film and video game industry can’t really use them)


This different example, that does not use noise but intentional point charges, uses a tensor field to generate the displacement. Be warned about what @Michael_Pryor said, obviously…

Getting the mesh to a good-looking state is a matter of tweaking the charges… (72.0 KB)



Giulio Piacentino
for Robert McNeel & Associates


@Michael_Pryor I Understand that completely. I was exploring visually what form of pattern works on my design. It was more an exercise for me, a starting point to reach something different.
(Film and game uses nurbs? Didn’t know that) Well, to do something like this in nurbs I think I can’t just sample any grid of points, the points have to follow the geometry? But different topic.
@piac Great starting point for me to understand tensor field, thank you and @Dani_Abalde.
The sine functions is what I was missing in my efforts.

Are there not some production facilities who would machine moulds from STL files or similar, especially for certain parts where the goal isn’t very high precision?

Film and game do not use nurbs, but they don’t use that high density displacement mesh.

the goal isn’t very high precision


These “peaks” are a no go.

Somethings of course can only be mesh, but this design in question is quite easy to make via curves and sweeps while maintaining clean and crisp geometry.

Yes, its funny too see so many people doing amazing mesh geometry here. But its almost always pointless if there is no way in making this a clean Nurbs or Bezier geometry.

1 Like

You’re getting into muddy waters here, TomTom. This might be even a philosophical question, since there are people that simply like to do geometrical experiments, art pour art so to speak, and others that have for instance a physical project in mind. I guess most of us are a mix of both.
Things that are pointless for one person might be motivational for another, it’s all just a matter of perspective! I’m sure that there are lots of people that find even such sneaker designs pointless, since they are too ingrained in what they know, and thus resistant to “innovation”.
Furthermore, I’ve dealt with a couple of manufacturers, mainly small shops that do additive or subtractive manufacturing (i.e. 3d printing, milling, etc.), and that gladly accept meshes. None of these ever asked me for N.u.r.b.s. geometry.
I also doubt that the children in Nike’s Vietnamese sweatshops that stitch, glue, and sow the sneakers together care for geometry types.


If you see it like this I totally agree. Pointless in a way, that for a real production process usually a mesh representation is simply not sufficent. You can ignore that and thats okay, you can also dislike and question this.But that doesn’t change anything, currently the real challenge about parametric design is to reach the same quality standard as the traditional industrial approch would do. Anything below that level is no real innovation or improvement and rather makes it a “waste of time” .


I have never heard of that before. How would the geometry being a mesh in any way have a significant influence on the time the print or milling takes. I don`t even know a CAM software that computes internally on nurbs, let alone have read any paper on 3d cnc path generation not using mesh as a base.
Can you share some infos on what specific process you are talking about? What software and in which case you notified a difference?

Another question, what do you mean with repeat ability? Once you have a gcode, the repeat ability is up to the machine, isn`t it?

I cannot get into more specifics than I have. But in that scale of production there are issues and ways to circumvent those issues, believe me if I could just shoot off displacement meshes all day I would have a much easier job to do, I wish it were the case but it is not. :smiley:


Are you talking about displacement meshes or meshes in general? I understood the later, if the former then I 100% agree.

Its also kind of problematic that students(esp. in the architectural field) often just go from hastly meshed up GH definition to fdm printing, completly forgoing all the parts where digital to physical gets “interesting”.

1 Like

Are you talking about displacement meshes or meshes in general?

The former, mostly related to the mesh in question of the topic. I even said before some things can only be mesh but they are usually well thought out and not as Brute force as a displacement, edge flow should at least always be thought out.


If you have mesh geometry you can end up with a faceted surface if your process includes machining a mould for example. On a concave surface the faceting all but disappears if it is machined with a ball nose cutter with a radius similar to the mesh face size for example and with a very high density mesh the faceting will all but disappear on convex surfaces.

Often, the CAM software is integrated with whatever CAD software the production facility use so for example if they use Solid Works and you give them a mesh to create a mould from, Solid works imports a huge amount of useless little surfaces and your mould design process is frustrated from that point onwards. Perhaps there are better mesh import plugins for SolidWorks now that can create intelligent surfaces? So I wouldn’t say meshes make things slower but they do if you use a mesh with CAM software that is based on surfaces.

My cheap CAM software does 3d surfacing from STL meshes and 2.5d machining from 2d dxf entities so I might as well just design meshes, lines and arcs but design with process in mind.

For 3d textured surfaces the more important thing I have to remember is that if the curvature is tighter than the radius of the cutter then the cutter cannot follow the surface properly and I get a bad surface finish.

1 Like

Regarding Mesh vs Nurbs. Its not only about what is better to produce. Much more important is how to guarantee that everyone participating can do his/her job. Therefore many companies have a metric on how data has to be.

As Michael pointed out for Nike, very often production and development are done in different companies. This is not so different in Automotive. A manufacturer has much more freedom when having a mathematical description of a shape. Thats whats Nurbs are about and what Meshes don’t do.


I made some experiments and I can say that quadremesher in rhino 7 is doing good job in following the hills and valleys. As there are lots of parameters to take into account I am not yet to a (perfect) mesh but it seems doable.


Definitely a no go! If you can’t set your watch to something from displacement that is producible then, what’s the point. As you say further on, you would just be shooting off displacement meshes here and everywhere without a care.

I work with direct moulded footwear, so features and so on adhering to minimum requirements becomes even more crucial, not to mention making sure you’re actually going to fill a mould.