Why AI video still can’t reproduce a real person — and what we are changing to get closer
Five reference clips of the same person should be enough for a video model to hold her face, body and voice across shots. It is not. Here is the real reason, and the concrete fix we built and are testing.
The Video Funker team
