StudioVideo Types
The video type chooses the engine that makes your reel. It changes the result more than any other setting, so it is worth understanding before you spend a generation.
ai-video| Generative film with a cast — the most cinematic option |
📚 tutorial | Instructional shape, now merged into ai-video as story shapes |
The workhorse. Art direction over your captured screens with motion, titles and narration, and no characters. Fastest to generate, cheapest, and least likely to surprise you.
Because there is no character animation it largely skips the expensive phases, which also means it usually does not stop at the review gate. If you want a reel in the least time for the least money, this is it.
standard accepts an image style — an art-direction preset such as a 3D cartoon treatment — so it can look considerably less plain than “screenshots with captions” suggests.
An illustrated, animated treatment. Suited to explaining something that is not purely a click path: a concept, a before-and-after, a workflow with parts that never appear on screen.
A presenter avatar delivers the narration to camera. The most conventional shape — it reads as a person explaining something — and the most predictable.
This is the engine that renders 4:5 natively. If you are producing for a feed that wants feed-portrait and you care about true framing rather than composition, avatar is the one that gives it to you.
The generative engine, and the most capable. A cast performs a premise with your recorded screens cut in. Two model tiers:
💡 There is no reason to pay cinematic rates for a draft you are going to reject. Iterate on flash, finish on veo.
ai-video has a sub-option that changes its whole structure. This is where the old tutorial type went — it merged into ai-video as these shapes rather than staying a separate engine.
If omitted, the shape is cast.
A multi-character story. A cast acts out a premise and the guide’s beats are cut into it. This is the marketing-shaped option: it has narrative, it holds attention, and it goes through the review gate because there is a story worth approving before you animate it.
Best for launches, feature announcements, and anything whose job is to make someone want the thing rather than operate it.
One presenter. An intro, then the whole recording as recorded, then an outro.
The important detail: in walkthrough the steps keep their own guide narration. The presenter sets up and closes out, but the body is your guide, narrated as your guide — not rewritten into a story.
This is the shape for documentation. It is honest about what it is, it stays accurate because it is structurally tied to the recording, and nobody has to sit through a premise to reach the instructions.
Walkthrough plus a presenter lead-in before each meaningful phase. The AI groups the steps into phases — at most about six chapters — and the presenter introduces each one.
The best shape for a long flow. A fourteen-step setup as one unbroken walkthrough is exhausting; the same fourteen steps as four chapters with a sentence of orientation before each is followable. Chapter breaks also give viewers somewhere to scrub to.
💡 Rough guide: under about six steps, walkthrough. More than that, chapters. Not instructional at all, cast.
No engine is the best one. Each buys something with something else.
⚠️ The two cons that bite most often are cost on ai-video + cast and runtime fatigue on walkthrough. Both are avoided by choosing against what the viewer must do next, rather than against what demos best.
The question that resolves most of this: what does the viewer do next?
ai-video