Video models generate short clips, and everything longer is assembled from them. That constraint reaches all the way into how the finished piece is cut.
Why the limit exists
Memory consumption grows with the number of frames held in context, and video frames are expensive to hold compared with text.
Maintaining coherence also degrades with duration, so a longer generation would spend its extra seconds drifting rather than developing.
Providers cap clip length where cost and quality both stop justifying the extra frames, which is a shorter duration than most scenes require.
How editors compensate
The most direct response is to cut faster, so that no single shot outlives what the model can sustain.
Fast cutting reads as energetic in advertising and music promos, which is why generated video appeared there first and looks native in those forms.
It reads as evasive in drama, where a sustained shot is how tension is built, and audiences register the absence even without knowing its cause.
Which shots hide their own brevity
Some shot types are naturally short and carry no expectation of duration.
Inserts, reaction cutaways, establishing views, abstract texture and slow drifting moves over a static subject all sit comfortably within a few seconds.
Dialogue is the opposite case, since a spoken exchange demands continuous attention on faces over a stretch the model cannot deliver in one piece.
Why extending a clip is not the same as generating longer
Continuation features generate a new clip conditioned on the final frames of the previous one, which produces a longer sequence from several passes.
Quality decays across each extension, since each pass conditions on output that already contained small errors, and those errors accumulate.
The practical ceiling is a few extensions before colour, detail and motion have visibly deteriorated, which is why the feature is used for a beat or two rather than for a scene.
What the constraint has done to style
A recognisable aesthetic has formed around these limits: short shots, drifting camera moves, minimal dialogue, heavy transitions and music carrying the continuity that the picture cannot.
Some of it is genuine creative adaptation and some is limitation presented as choice, and audiences have begun to distinguish between the two.
The style will date quickly for that reason, in the same way that early digital effects now read as a period marker rather than as a look anyone chose.