Cast a votive into the cosmos. Find matching seats and earn tokens.
Upload an image. CLIP embedding + cosine similarity against seat concept embeddings. Anonymous votive created, tokens earned.
Requires: CLIP model on hub + concept embeddings populated per seat.