Glossary · Making it with AI
Temporal consistency
Temporal consistency is how steady a video stays from one frame to the next, with faces, objects, text and light keeping their shape instead of flickering or morphing.
Why it matters
People notice change between frames much more than detail inside one. A shot that looks fine as a still can feel wrong the moment it plays.
The typical failures in AI video have plain names: flicker, where brightness or texture shimmers; drift, where a face slowly becomes someone else's; morphing, where objects blend into each other; and popping, where things appear or vanish.
They are most common in long clips, fast motion and crowded scenes. Shorter shots, simpler movement and starting each shot from an approved picture all make them less likely.
Catching them takes two kinds of viewing: watching at normal speed, the way a viewer will, and stepping through slowly, where small changes, like a ring that disappears for a few frames, become visible.
In a shot of a printing press running, the lettering on the side of the machine quietly rewrites itself halfway through. The shot is remade as two shorter shots from the same starting picture, and the lettering holds.
Last reviewed 15 October 2026.