Just over four years ago, in March 2013, Dita’s Gown, a fully articulated 3D printed dress, was unveiled at the Ace Hotel in New York. The collaboration between Francis Bitonti, Michael Schmidt, and Shapeways was a watershed moment for 3D printed fashion. The project pushed the boundaries and limitations of existing 3D printing technology to produce a beautiful, functional, and wearable garment. The result of Dita’s gown, and its widely covered unveiling, was a significant leap forward for 3D printed fashion. The project demonstrated that 3D printed clothing is no longer a speculative possibility, but an emerging reality. We recently spoke with Francis Bitonti about the groundbreaking project and its unique challenges.
Could you give some background on how the Dita Von Teese dress project came to be?
The project was initiated by Shapeways. They were looking to do something for Fashion Week. People in the field were trying to figure out how to make it relevant to different industries. Shapeways approached the Ace Hotel to see what they were doing for Fashion Week. They were already involved with Michael Schmidt. Ace Hotel proposed this [Shapeways collaboration] to him and he said that it sounded cool. Dita got involved after that. I got involved once all those partners were in place.
How did that collaboration end up working?
The workflow was interesting. A lot of the way the dress looked and was made was because we were trying to figure out a way to work really quickly with Michael’s team. Constructing and iterating a 3D model is usually pretty slow. His team was used to sketching, so I had to figure out the fastest way to set things up to do a screen share with him and proceed to make revisions. Generally, that was how we worked. We had a low-resolution version of the dress that he and I were able to kick ideas off of. From there, there was the slow process of modeling the dress.
“All of these tools are not set up for models as complex as the dress; no one thought we would model 3,000 articulated connections”
What modeling software did you use and what challenges, if any, arose with it?
We were using Maya, Rhino, Grasshopper, and a bit of Python for different operations – a cobbled-together workflow that I still use to this day from software that’s not really intended for industrial design. All of these tools are not set up for models as complex as the dress; no one thought we would model 3,000 articulated connections. We’ve noticed a similar problem while pursuing more engineering-type work that requires particular kinds of meshes and specificity, so we’ve started using remote computers that we could scale up. We can work much faster and more efficiently now.
Were you using computer farms for the Dita project?
No, not at that time. It was just painful [laughter].
After modeling, were there any physical iterations?
The samples were really small. They were three-by-three mockups, so we didn’t know if the plan was going to work because the tile was so big that I couldn’t render it. I only saw the wireframe. The deadline was tight, and we didn’t have the computing power to visualize large samples at that time.
What other elements were added to the dress?
After it was printed, it went to Michael’s studio in LA, where they polished, dyed, and applied a coat of lacquer onto the nylon (it was rough) – and glued 12,000 Swarovski crystals onto the dress before it was sewn into a nude silk corset.
What was the timeline like?
Two weeks … [laughter]. We said two months, but the actual working time was two weeks.
So the dress was done in separate pieces?
Yes. There were 12 pieces because that was all we could fit in the machine. We didn’t try to orient the parts in any particular direction and we weren’t aware that the materials would behave differently in different axes, so the dress was really thin –about a half of a millimeter thick. Nonetheless, Dita got in and out of it like a regular dress once everything was assembled.
Most of your projects are algorithmically generated; how come this project isn’t?
It’s hard when you’re working with a designer because we make such complicated patterns and structures. If we have an algorithm that makes a patterning texture or material properties, we tend to talk about it as a material independent of design. Even with Michael, I would say “we designed that material,” because it was designed procedurally. You could say it was made as an algorithm since that’s how we deal with all the different shapes. That’s why we do tests to figure out how everything behaves.