Bini@Museum
AI-generated artworks from the Biniôke track archive. Every background is deterministically bound to one track — produced via our own stem-splitting + AI-imagery pipeline. Not random illustrations: the image IS the song in a different coordinate system. First public curation: 39 works, hand-selected from the wider archive. More at every release.
About this museum
Why a museum? AI-generated work has a reputation problem: infinite production, little curation. Bini@Museum chooses the opposite — what stays gets exhibited, the rest of the tool output stays in the archive. Every included image is hand-selected by the maker as "art-worthy". Some tracks have one piece; some have multiple variants that each stand on their own. The curation is per-image, not per-song.
.MitD originals vs karaoke renders. The .MitD works are own compositions (Sören's artist alias). Karaoke renders are AI interpretations of known songs, tested as demo material — not for distribution, but as an AI-versus-original-song-meaning study.
Technique. Every background comes from Biniôke's own pipeline — local stem-splitting, lyric transcription, and a proprietary image generator bound deterministically to each song. Same song = same image, always. No mass-spam, no drift.