Explore different visual styles by generating images with Style Jukebox, a tool that allows users to blend between styles and create new hybrid styles.
Style Jukebox is a creative generative art tool developed by researchers at MIT-IBM Watson AI Lab. It utilizes AI and machine learning techniques to enable new forms of human-AI collaboration and creativity.
The key feature of Style Jukebox is the ability to interpolate and extrapolate between different visual styles. It contains a library of over 75,000 style images spanning a wide range of aesthetic styles including artworks, architecture, fashion photography, graphic designs, etc.
Users can seamlessly navigate in the style space by dragging a slider to generate new images that blend attributes of different styles. This allows for creating smooth style transitions and hybrid styles that combine aspects of multiple styles. The interface also allows sampling random styles and locking to particular styles while blending others.
In addition to blending, Style Jukebox provides advanced controls over image generation. Users can specify constraints like image size, aspect ratio, color palette and content class to steer the output. There are also options to generate style mosaics, collages and videos.
By empowering intuitive style exploration and hybridization, Style Jukebox aims to support new creative workflows for artists, designers, photographers and causal creators. The technology and learnings also inform further AI research into style representation and manipulation.
Here are some alternatives to Style Jukebox:
Suggest an alternative ❐