AI Art
AutoMovie
AutoMovie is a pre video-era AI exeriment, where ChatGPT writes a full storyboard from a super-short input (e.g. Horror in Disneyland) along with promts for pictures that represent key-frames telling the story. The key-frames are then rendered via Stable Diffusion and interpolated via latent blending. The story is told via subtitles, which are overlay the images. The LLM also choses the best fitting soundtrack, which is automatically downloaded from youtube and added as background music.
It can also be used without a script to simply generate beautiful unending visuals.
BoomGan
This page is dedicated to my AI Video-DJ project Boomgan. The idea is to generate “trippy” visualizations for electronic music using Nvidia’s Stylegan 3 that vibe with the music. For now, Boomgan only supports pre-rendering sets but a live mode is planned.
Here some non-audireactive samples showcasing different methods for latent-space traversal:
| Mode | ||
|---|---|---|
| rjump |  |
 |
| twocircle |  |
 |
| rwalk / orwalk |  |
 |
And finally an audio-reactive one, combining the above with Fourier features extracted from the song: