finding flow

machine learning | branding

Year: 2024
Client: Augmented Research
Role: Art Direction & Motion Design
Musik: Edict, Sukkube, Nebuchadnezzar, Raär

Tools: StyleGan3, Yolo V8, After Effects

For the Multi Artist EP release 'Intelliance', the underlying theme of flow was developed, referring to the flow we feel when dancing to music. To translate this theme into a visual concept, we looked to nature. Things like evolving clouds, growing plants, breaking waves, blooming flowers all gave this sense of flow. From all these things and events, we created a dataset of over ten thousand images, which were used to train a machine learning model (Stylgan3). First of all, to bring it all back into a technical context, and also to be able to morph, or rather flow, between these different visuals that we found in nature. 
To bring back the idea of searching, an object detection model was used on the generated footage, which also served as a visual contrast to the flowing, smooth images. As well as a pose detection model to find movement in the footage, which is then represented by a sort of dancing stick figure. All of this was combined to provide visual support for the music.