AI video length settings: short, standard and long

Three settings, three budgets

Short builds 3 to 6 scenes, standard 4 to 9, and long 6 to 12. Those are enforced, not suggested: the director is told to build exactly within that range, no more and no fewer. Each setting also carries a rough duration target used when a template is matched to your brief, at ten, eighteen and thirty seconds respectively, which is where a promo at that scene count naturally lands.

The same budget is a different length per type

This catches people out. A promotional scene runs 2.5 to 4 seconds and carries five to eight words of narration. A tutorial step runs 5 to 9 seconds and carries ten to twenty five words, with three to five seconds for the intro and wrap up. So twelve scenes is around thirty five seconds of promo and well over a minute of walkthrough. The creator format is different again and is hard capped under thirty seconds whatever you pick.

What the published research actually says

Vidyard's benchmark across more than 940,000 business videos found 65 percent of viewers finish a video under one minute and 20 percent finish one over twenty. Wistia's 2026 report, across 13 million videos, shows engagement falling steadily with length, from roughly 52 percent under a minute to 43 percent at three to five minutes. The exception in both datasets is educational and instructional video, which holds up far better at length because the viewer came to learn something.

Pick from the placement, not the content

A feed placement is short, almost always, because the video is competing with a scroll. A watch page, a landing page or an internal walkthrough can be longer, because the viewer chose it. If you have more to say than the placement allows, the answer is two videos rather than one long one, and the scene budget makes that split obvious rather than something you discover in the edit.

How it works, in three steps

1. Step 1

   Choose the placement first

   Feed, watch page or internal. That decides the length setting more reliably than how much you have to say does.

2. Step 2

   Count what has to be established

   Every point needs a scene. Four points plus a hook and a close is six scenes, which is short at its ceiling or standard comfortably.

3. Step 3

   Let narration retime it

   Writing a longer line grows that scene to fit the audio, and shortening it later shrinks the scene back, because the original duration is remembered.

The full walkthrough with screenshots is in the guide Make a promo video with AI voiceover.

Limits worth knowing

• The creator format only offers short and standard, both labelled max 30s, because it is capped under thirty seconds.
• Twelve scenes is the ceiling. There is no setting that builds a five minute video.
• Exports bill their render time, and one over thirty seconds, or one that needs new narration synthesised, needs at least 1 credit left to start.
• On dynamic pace a scene grows to at most one and a half times its authored length, so a very long line spills over the next cut rather than stretching one scene indefinitely.

Questions people ask

Related pages