AI Video, Explained
What “Reference-to-Video” Means
Reference-to-video means the model is given both a still image of the subject and a separate reference clip showing the motion to copy, then it combines them.
Image supplies identity
Your uploaded photo tells the model what the person looks like — face, hair color, clothing, room.
Reference clip supplies motion
A short example video of the haircut action gives the model a template for how hands, scissors, and hair should move, frame by frame.
Why combine them
Describing a precise physical action in words alone is unreliable. Giving the model a motion reference produces a much more consistent, realistic result than a text prompt by itself.
Fake the bob before you commit to it
Bob Haircut Filter turns one photo into a viral haircut-reveal video in a few minutes. Free on the App Store.