DIGITAL FABRICATION
DIGITAL FABRICATION
Cloud Table
Everything is designed through artificial intelligence and big data. An auto encoder is developed using a vast number of 3D table models obtained via ShapeNet to extract the likeness of a table. This model generates infinite new table geometries as a point cloud. The point cloud is transformed into a mesh where the edges and nodes represent metal rods and joints, respectively. Here, selection is the only process where designers engage. They just have to pick one geometry out of the numerous output geometries from the model. This is a proposal on the new way of design. In charge of machine learning model development. Collaborated with Nicolas Feihl. Photo by Julius Hatt.
ニューラルネットを利用したテーブルのデザイン。机のCADモデルを用いてオートエンコーダーを構築し、「机とは何か」を人工知能に学習される。その結果、与えた3Dモデルに派生する無数のバリエーションが得られるニューラルネットを獲得した。