Skip to content

Computational Couture

Thoughts and inspiration

Future Story

Coming soon

Thoughts

There are many exciting figures and projects in the field of computational fashion, such as Anouk Wipprecht, Danit Peleg, Brigitte from Variable Seams, Julia Koerner, who gave us a lecture and shared her insights, and Studio Nervous System, which has also been active in this field for a long time. As a regional figure in this field, I’d also like to mention Yvonne Dicketmüller once again.

I've been following the field of 3D-printed clothing for several years now. While the garments were initially very rigid artsy objects, they have increasingly evolved into wearable clothing as flexible 3D printing materials such as TPU have improved and 3D printing has become faster and more cost-effective.

Doing research on parametric design, I learned, that its roots lie in architecture, but found its way into Design, engineering and also Science. On the Side of Architecture and Design, you can see a lot of organic, naturally grown-looking structures. I was a little bit bored by all those similar looking structures and forms, to be honest. But I guess, there are many points, that led to this. First: There's a reason why things are growing Like that in nature. It's efficient and the structures are stable, when we're thinking of structural integrity. Another Point, I can imagine, is the difference between classical 3D-modelling and parametric design. 3D-modeling is often very rigid, limited in the ability to shape freely and create precisely those kinds of organic forms. Digital sculpting which (as far as I know) didn’t develop until later—does offer the ability to design freely, but at the same time, it’s harder to control what you’re actually doing. Parametric design combines these two aspects, and that’s why I completely understand why software like Grasshopper has sparked a major movement in parametric design.

Inspiration

Zellen, Pilze (der mit dem Rock), Flechten, Schleimpilze, Seifenblasen

I, too, wanted to draw inspiration from natural forms. However, rather than working with random shapes and structures—which might, for example, resemble cells — I preferred to deliberately attempt to replicate existing forms and structures. The other source of inspiration came from the field of fabric manipulation. The images we were shown earlier in the introduction and the ones I found during my research—strongly reminded me of technical “origami,” which is used, for example, in satellites (yeah, we’re back in space again), and, on the other hand, of shibori. That’s why I was curious to try a few experiments in that direction.

Another small source of inspiration is the alternative scene—specifically punk, metal, and goth—where jewelry with spikes is a common theme.

INSPIRATIONSCOLLAGE IN CSS

Documentation

Parametric Modelling

I was surprised how much fun I had with parametric design. I was thrilled by how it works and what you can create with it during Ricos tutorial. I liked his approach of thinking of the nodes as a recipe. I think it’s important to remember that the comparison isn’t so much like a cooking recipe that you usually just follow step-by-step. Many of us learn to cook and bake as children, so even without specific knowledge of the chemical and physical processes, we develop a sense of what the individual ingredients and steps do. I see parametric recipes more as scientific formulas, where it’s important to know which ingredient has which effect, what interactions occur, and what must be avoided so that everything doesn’t blow up in your face (or the computer crashes).

Naturally inspired

My first idea was to create a slimemold. On saturday, I gave a workshop for children (Giant mushroom and superslime) and met Emma van der Leest on Dutch Design Week, who explored slime molds as a design concept in her thesis some years ago.

Since I didn’t feel skilled enough yet to create the structure myself, I looked for tutorials on how to create slime molds using Blender. However, these were all designed to create an animation, and either that was too complex for me or something else distracted me (I can’t say for sure right now), so I decided to generate a lichen instead. Since it seemed like no one had done this yet—or at least hadn’t shared their knowledge about it, I tried using ChatGPT to help me.

Tutorials I tried: Slime Mold Simulation Nodes Tutorial (Blender 3.6a)
Simulating Slime Mold in Blender

A screenshot of a slimemold trial

First and last slimemold steps

I used ChatGPT to create the braided look. I usually avoid working with AI, but I had a specific vision for my design and didn’t yet have enough knowledge to create it. So I explained to ChatGPT what I wanted to do, so it could tell me how to do it. I was surprised that ChatGPT asked me about different types of lichen. It took 1–2 rounds of clarification to make it clear that I didn’t want a true three-dimensional object (such as a sphere), nor did I want a simple textile weave. Once that was clarified, I was able to select the style I wanted in ChatGPT (crustose lichen or foliose lichen) and whether the pattern should be a solid surface or have gaps. I opted for the leaf lichen with gaps. ChatGPT then explained the three levels of lichens to me: colonies — “Multiple growth centers are irregularly distributed across a surface. Each colony has its own size and orientation,” lobes — “Lobes of varying lengths, widths, and curvatures extend from each center. Their edges are wavy, not circular,” and tendrils — “Some lobes branch into narrow side arms. The tendrils end in different ways, so the pattern doesn’t look like a regular rosette.”

ChatGPT created a guide, but it was still a little too complex for me. So I asked ChatGPT to explain everything to me step by step. And that's exactly what it did. It even asked me to upload my screenshots along the way to check if everything was correct.

I worked with the free version of ChatGPT. The downsides were that uploading screenshots was very limited, and after a while, ChatGPT kept forgetting what it had already explained to me. That was time-consuming because I had to write down which steps had already been completed and ended up repeating some things unintentionally. In the end, though, that’s exactly what helped me gain a better understanding of what I was actually doing. I ended up creating a kind of lichen that I really liked. I would have liked to make the edges more voluminous, but that was too much for my computer to handle, so I decided to leave the braid as it was. I was also able to arrange it on a grid. I would have liked to adjust the size of the individual braids and 3D-print them as a pattern on fabric. But that, too, was too much for my computer. My guess was that there were too many calculations involved. So I considered exporting the braid on its own and then duplicating it. But even that was too much. I thought about checking the model in Meshmixer to see if I could simplify it somehow. However, Meshmixer kept crashing whenever I tried to have the model repaired. Also meshrefinery wasn't able to repair the file.

VIDEO LICHEN BLENDER

A picture of the generated lichen pattern repeated in a grid

The generated lichen pattern repeated in a grid

The generated lichen in Meshmixer

Trying to repair in meshmixer

So I decided to use just the outline of the lichen. I exported it as an SVG file. In Inkscape, I nested the outlines inside each other to make the structure a little more interesting. I wanted to print it as a pattern on fabric, and I also wanted to test Filafabric. At first, I had some problems because my object seemed to be too large. Once I scaled it down, I was able to export it.

A screenshot of Filafabric with my lichen model

Trying to work in Filafabric

I generated the pattern so that the outer layers are filled with the background pattern and the inner layers are completely solid. I wanted to print it with TPU, but unfortunately the print didn't turn out well. The layers for the background pattern weren't printed. It seemed to me that the G-code was missing the command to move the filament.

A screenshot of Filafabric with my lichen model

Trying to work in Filafabric

A picture of the failed 3D-print

Printing fail

The next attempt: I created the scaled pattern in Blender as curves and then applied a thickness using the xyz command. I could already tell in Blender that it didn’t look quite right. There are probably too many curve points—which might also be the reason why that 3D model caused my laptop to crash. In Bambu Slicer, it became clear that something was wrong. I got a warning that there were too many points, and only a few areas were even marked as printable.

A screenshot of a lichen pattern in blender

Lichen: the next try

A screenshot of a lichen pattern in blender

Lichen: the next try

I was a little frustrated that I couldn't print the braiding pattern, but maybe I'll still find a way to make it printable after all.

Fabric Manipulation

Inspired by this book and various images of folds that I found online, I came up with different shapes that I wanted to create and try out on stretchy fabric. I decided to use the crosses and the spirals. The idea was, that the 3D-printed parts would bring the fabric to form spike-like forms, when being relaxed.

A screenshot of crosses positioned in a grid pattern in blender

Cross-grid-pattern

A screenshot of spirales positioned in a grid pattern in blender

Spirale-grid-pattern

One (important) thing, I didn't take care of, was the right sizing of my models. I modeled them, having metres as units. Of course, that was too big for the 3D-printer, so I had to transform it into the right size in bambu. I will take care of that in the future.

A screenshot of the dialogue window, warning me, that the file is designed in the wrong unit

Wrong unit

To create the fabric manipulation effect, I stretched the fabric onto the printing plate. The fabric was only stretchy in one direction, so I didn’t expect the distortion to be uniform from the start. Stretching it and attach it to the printplate turned out to be difficult. A combination of painter’s tape and magnets produced the best results. With the magnets, we had to be careful that they didn’t interfere with the print head. Otherwise, the print head would shift the magnets, causing the piece of fabric to warp as well.

A picture of the moved magnets in the 3D-printer

Moved magnets

Unfortunately, I didn't get the results I expected. The material didn't warp. I then tried using TPU instead of PLA for the crosses and the larger spirals. With the larger spirals, there was at least a hint of warping.

Another problem was that the printers didn’t always reliably resume printing after the pause. Together with a colleague, I tried resending the G-code—which was paused after the break—to the printer for the cross sections. However, it seemed that this part of the code didn’t contain the command to extrude the filament. As a result, the printer dutifully ran through all the steps correctly but didn’t actually print anything.

I think that to achieve a good effect here, I should use a fabric that stretches in the beige direction and also rethink the geometry of my objects. Instead of crosses with four arms, I probably would have been better off using only three arms. I wonder how complicated it would be to test the effects beforehand in a Blender animation.

Punk spikes

Inspired by this book and various images of folds that I found online, I came up with different shapes that I wanted to create and try out on stretchy fabric. I decided to use the crosses and the spirals. The idea was, that the 3D-printed parts would bring the fabric to form spike-like forms, when being relaxed.

I created the rivets in Blender. I wanted them to be randomly varying in size. I also didn't want round rivets, so I reduced the number of sides. In the final print, however, this was barely noticeable, and they looked more like circles.

I printed the spikes using PETG on a non-stretchable mesh fabric. Some spikes didn’t print correctly even in the first layer. This then led to further errors. In these spots, the filament got stuck on the nozzle and then on the next spikes. At first, I removed these defective spikes and cut them out. However, this caused more filament to accumulate on the nozzle at these missing spots, which in turn got caught on the next pins. So I decided to leave the misprinted spikes in the printer and use them, so to speak, as scrapers for the nozzle. As a result, the remaining spikes actually turned out well.

Tools and sources I used

  • Squishtool (for PNG and JPG): https://squish.tools/de/compress-png
  • Blender 5.2.2
  • ChatGPT free version
  • Inkscape
  • Bambu Slicer
  • Filafabric
  • Meshmixer
  • TPU Filament
  • PETG Filament
  • PLA Filament
  • Mesh fabric one directional stretch
  • Mesh fabric non-stretch

Reflection

There were a few frustrating moments along the way, and of course I would have liked to show more successful samples after this week. But I’m focusing on everything I’ve learned this week about parametric design and 3D printing. Until now, I’ve used 3D printers only very rarely, so I hadn’t had much practice. This intensive week has definitely helped me develop more proficiency.