One source video.
N distinct variants.

Forge a clip into a spectrum of algorithmically-distinct copies, each engineered for real distinctness and verified against PDQ perceptual hashing and audio fingerprinting. No watermarks, no guesswork.