What decides how long a generated shot runs
Three arrangements decide how long a shot runs. A fixed menu, which reaches a script. A range, which leaves the choice to a writer. Or the supplied recording, which moves the decision to a session. As of 2026-09-12.
| Model | What is documented | Detail |
|---|---|---|
| Hedra | The supplied recording | Omitting the duration follows the audio |
| MiniMax | A range, four to fifteen seconds | Chosen on the call |
| Sora 2 | Two fixed options, then extensions | Sixteen or twenty, to a total of 120 |
| Veo | A menu of three | Four, six or eight seconds |
| Wan 3.0 | A range, two to thirty seconds | Chosen on the call |
Inclusion rule. Entries whose documentation states what determines the duration of a generation. An entry publishing no duration at all does not earn a row. Order. Alphabetical by model name.
1A menu reaches the script and a range does not
Where the options are four, six or eight seconds, dialogue has to be written in those units or it will not fit. That constraint travels upstream into the script, which is inconvenient and honest: everybody knows the shape of the thing being written.
A wide range postpones the decision and lets a writer produce lines that need cutting later. The freedom is real and so is the rework, and which is preferable depends on whether the script or the schedule is the tighter constraint.
2Length as a property of the take changes who decides
Where the audio decides the duration, pacing lives in the recording session. A performer who pauses for effect is buying frames, and a production working to a fixed episode length has to budget those pauses with a microphone in front of them.
That is the cheapest place to make the decision and the least convenient place to change it. Fitting a slot afterwards means re-recording rather than adjusting a number.
3Extensions are a fourth answer disguised as a longer ceiling
Adding twenty seconds at a time to reach two minutes produces an assembly rather than a take. Every addition is a join where a voice, a level or a room tone can move, so the published total describes what can be stitched rather than what can be shot.
Read as a budget for a beat the figure is useful. Read as a promise of two continuous minutes it will disappoint, and the documentation does not say which reading it intends.
- Shot length set by the takeThe audio normally determines the video length, and omitting the duration follows the source audio
- Shot length a line has to fitGenerations run 16 or 20 seconds, and an extension adds up to 20 seconds at a time to a total of 120
- Audio sourceThe audio parameter defaults to true so the returned video carries an audio track, and false returns one withouton unless it is switched off
- Audio sourceAudio is generated with the video rather than added afterwardsgenerated with the picture
- Audio sourceNative speech with lip-sync tied to the speaker who is on screenspeaker-aware
4Sources
Every line is read from the vendor documentation linked on the speech controls page, checked 2026-09-12. What counts as documented is on how read. Nearby: Charging for audio, Keeping a voice, Lip-sync.