AI script generator

AI Script Generator

Updated September 20, 2026

Quick answer: if you have a premise but not a script, the built-in writer drafts one inside the project: you give it an idea, a length and a tone, and it returns a full serial script in the language your interface is in — not only English. The draft lands in the project as the source text, so the next click is script analysis rather than a copy-paste into a different tool.

Still from “The Veil at Dawn”, an AI short drama generated end-to-end by SceneMixer (6 characters, 1 episode) — Sample: “The Veil at Dawn” — Realistic HD, 1 episode, produced end-to-end by SceneMixer from the script · watch the episodes

It writes in your language, not just English

Two separate settings decide what language things come out in, and conflating them is the mistake most pipelines make.

Writing language — the language of the script itself. It follows your interface, across all sixteen supported languages. A Korean interface gets a Korean script.
Production language — the language of the working documents the pipeline generates from it: the episode outlines, the shot descriptions, the asset notes. That stays English (or Chinese on the mainland edition), because those documents are read by the next model in the chain.

The result is that a Korean script behaves exactly like a Korean script you pasted in yourself: English production documents, translated back into Korean for you to read on screen, and dialogue spoken in whichever language you choose in Step 1. Until September 2026 the writer answered in English regardless of the interface, which made it useless outside one market; that is fixed.

Pick a length that the pipeline can actually carry

What you set before it writes
Premise	One line to a few paragraphs. The more specific the conflict, the less generic the draft.
Length	Chosen from length tiers rather than typed as a number, and any tier that would not fit the analysis ceiling is hidden rather than offered and then rejected.
Episodes	Budgeted at roughly 720 words per episode for English, so a length tier is also a rough episode count.
Tone and genre	Carried into the draft; the visual style axes are set later, in Step 1, and do not constrain the writing.

English length is counted in words rather than characters, deliberately. Counting an English script by characters — the unit that works for Chinese — cuts the draft to about a quarter of a usable length, which is what our own international testing showed before the unit was changed.

Then it keeps going

The draft is written into the project as its source text. From there the run is identical to one that started from an uploaded script: analysis extracts the cast, locations, props and the episode list; reference art is generated; each episode gets a storyboard; segments render; episodes composite. Nothing is exported and re-imported, and you can edit the draft at any point before analysis — or throw it away and paste your own. See script to video for the path from a finished screenplay, or the generator overview for the whole run.

What it costs

Writing is billed by the length you asked for and by the language you are writing in. Alphabetic scripts — Latin, Cyrillic, Greek, Arabic, Hebrew and the rest — are billed at 30 per cent of their raw character count, because the same story takes far more characters in English than in Chinese; Chinese, Japanese and Korean are billed at full count. Everything after the draft is priced as normal: analysis at 4 credits per 3,333 characters of English text, storyboards at 1 credit per 333 characters per episode, video from 10 credits per second of output. Credit packages start at $1.49; see the pricing page.

What it is not

Not a chat you co-write in. You brief it and get a draft. Revision happens by editing the draft, or by asking for another one.
Not a requirement. The writer exists for people who arrive with an idea rather than a manuscript. Uploading your own script skips it entirely and costs nothing.
Not a rights launderer. A draft written from someone else's copyrighted premise is still someone else's premise; our compliance checklist covers what that means before you publish.

Every episode, in fifteen languages

The dialogue language is set once in Step 1 and the whole run is performed in it, with the target-region setting moving names, faces and streets to that market at the same time. Each sample below was produced that way, one per language.

Languages

Native dialogue in 15 languages

Every series below was produced by the same pipeline with the cast speaking that language natively: the interface, the working documents and the spoken lines are all in one language, and nothing is dubbed. Open one to hear it.

See all 15 languages

Frequently asked questions

What languages can it write in?

All sixteen interface languages, including simplified and traditional Chinese. The script follows your interface language; the pipeline's internal working documents stay in English, and what you read on screen is translated back.

Can I edit the script it writes?

Yes, freely, before analysis — it is just the project's source text at that point. You can also replace it entirely with your own.

How long a script can it write?

Length is picked from tiers rather than typed in, and tiers that would exceed what the analysis stage can take are not offered. English drafts are budgeted at roughly 720 words per episode.

Is the writer free?

No. It is a model call, billed by target length and by the language you write in, with alphabetic scripts billed at 30 per cent of their character count. Uploading your own script instead costs nothing.

Does the dialogue language have to match the script language?

No. Dialogue language is set separately in Step 1, so a script written in German can be performed in Spanish. The cast, names and settings adapt to the market you target.

Start from an idea, not a blank page

A draft in your language, and the pipeline waiting on the next click.

Try SceneMixer free

Credit packages from $1.49 · Pro $7.99/mo · No credit card to start