“Generative AI for Concept Creation in Footwear Design” by Suessmuth, Fick, van der Vossen, Aktas and Gesell
Conference:
Type(s):
Interest Area:
- Art / Design
Title:
- Generative AI for Concept Creation in Footwear Design
Session/Category Title: Tailor Made: Techniques in Computational Cloth
Presenter(s)/Author(s):
Abstract:
We present AI Archive, a footwear design tool using generative artificial intelligence (AI) that we successfully integrated into the design process at adidas. AI Archive is based on diffusion models and was trained on the entire archive of adidas sneakers, which dates back to the company’s beginning in the 1950ies. Being trained on this unique dataset enables the AI to generate new and innovative sneaker designs that draw inspiration from the archive and pay homage to the rich history of the adidas brand. AI Archive has been rolled out to our designers as a web application in 2022. The tool has since established itself as an essential ingredient in the concept-to-prototype process of many of our designers. The proposed system gives users a high level of control over the design process, enabling them to precisely guide the AI to create designs according to their direction. We believe that the use of generative AI in footwear design has the potential to transform the industry, as it allows designers to explore hundreds of different concepts in almost no time.
References:
[1] A. Radford, J. W. Kim, C. Hallacy, A. Ramesh, G. Goh, S. Agarwal, G. Sastry, A. Askell, P. Mishkin, J. Clark, G. Krueger, and I. Sutskever. 2021. Learning Transferable Visual Models From Natural Language Supervision. In Proceedings of the 38th International Conference on Machine Learning, Vol. 139. 8748–8763.
[2] R. Rombach, A. Blattmann, D. Lorenz, P. Esser, and B. Ommer. 2022. High-Resolution Image Synthesis with Latent Diffusion Models. In Proceedings of CVPR 2022.
Additional Images:
ACM Digital Library Publication:
Overview Page:
Submit a story:
If you would like to submit a story about this presentation, please contact us: historyarchives@siggraph.org
