Text to Video AI: Writing the Brief
Copy this brief: A one-minute short: at 11:58 on New Year’s Eve a courier delivers a letter to the wrong flat and stays for the countdown; one stairwell, one balcony, fireworks off screen.
Four links, from one sentence to a shot
| Link | What it is | Size | Who writes it |
|---|---|---|---|
| Brief | Who, where, and what goes wrong | One line | You |
| Script | The story in scenes and lines | 1,000 to 20,000 characters on AI Create; or your own | AI Create, or you |
| Storyboard | Shot by shot: framing, who is in frame, what happens, seconds | One per episode | The site, then you edit |
| Shot | One framing, one action, a line if there is one | 3 to 5 seconds | The site, then you edit |
Everything above the shot is text, and editing text costs nothing. Only the last step, turning a segment into video, is charged, per attempt. Fixing one shot without redoing everything.
The brief: person, place, what goes wrong
The one-line brief matters more than the length you pick. On the AI Create tab the options are Mini (about a thousand characters), Brief (2,500), Short (5,000), Medium (10,000) and Long (20,000), each with its own price on the button. Mini is a test, Short is the usual first choice, and a short script gives a few episodes while a long one gives many. Choosing script length and cost.
If you paste your own text
There is a floor: about a paragraph, roughly 330 characters of English. Below it there is nothing to break into episodes. There is also a ceiling, and it is a hard stop rather than a silent cut, because a script that comes back looking complete when it was truncated is worse than a refusal. Long books are read in two passes. What you can feed in · Long scripts, read in two passes.
The shot: what the camera sees
A shot is a marker with its seconds, then a few sentences about what a viewer would see and hear, in the order they would notice it: the framing, whether the camera moves, who is in the picture, the one thing that happens, and the line. Named characters are tagged with @. Keep to one action per shot, and give it enough seconds to happen.
| Weak | Strong |
|---|---|
| Mara arrives and is upset and talks to Teo. | [Shot 1] 4s: Wide shot, a stairwell at night. @Mara, in a wet jacket, checks a door number and knocks. |
| There is a big emotional moment. | [Shot 3] 3s: Close-up on @Mara. She looks at the letter, then at the paper hat. @Mara: Thirty seconds. Can I watch from your balcony? |
| They watch fireworks together. | [Shot 4] 4s: Two-shot on a small balcony, city lights below. Both count the last three numbers aloud. The sky flashes off screen. |
How to write one shot · Write dialogue in the storyboard.
Six mistakes that waste credits
- Two actions in one shot. Split it.
- No seconds, or seconds that do not add up. The seconds are real; a line cut without changing them leaves dead air.
- Feelings instead of things you can see. “A big emotional moment” gives the model nothing to draw.
- Dialogue not marked as spoken. A line has to be written as a line to be picked up. Write dialogue so it is picked up.
- Expecting readable text in the picture. Signs and prices are not drawn reliably. Put them in the spoken line. Text on screen: fixes.
- Regenerating before editing. Fix the text first, look at it, and only then spend. Spend fewer credits.
From text to video, step by step
The read costs 4 credits per 1,000 characters. Video is 10 credits a second on the default tier, and the first 15 setup images and the first storyboard are free. Generate one segment before the rest. Text to movie · Script to video.
Make it in SceneMixer, start to finish
- Script: paste your own on Upload Script, or write the brief above in AI Create and let it draft the episodes.
- Step 1: Type Short Drama; Scene Style Neon Nights, Urban Opulence; Costume Style Smart Casual, Streetwear; Visual Style Digital HD, Soft Glow; Target Region; dialogue language; 9:16 Portrait. Then read the script.
- Step 2: the two leads with one mark each and a voice; the stairwell and the balcony as locations; the letter as a prop. The first 15 setup images on an account are free.
- Storyboard: generate episode 1 first (free on your first drama) and read it before you render anything. One place per segment.
- Video: render on the default tier at 10 credits a second, fix single shots instead of whole episodes, then put the episode together and post.
| Text you write | What the site does with it | What it costs |
|---|---|---|
| A brief | Writes a script on the AI Create tab | Priced by the length you pick, shown on the button |
| A script | Reads it into a cast, locations, props and episodes | 4 credits per 1,000 characters |
| A storyboard edit | Re-reads only the shots you touched | Nothing |
| A segment | Generates the video for that segment | 10 credits a second on the default tier |
FAQ
What is text to video AI?
A model that turns words into moving pictures. One sentence makes one clip; a watchable video needs a chain of text from brief to shot.
How do I write a good text to video prompt?
Describe what the camera sees: the framing, who is in the picture, the one thing that happens, how many seconds, and the line if there is one.
How long can the text be?
At least about a paragraph, roughly 330 characters of English. There is a hard ceiling, and long books are read in two passes.
Can I write the script myself?
Yes. Paste it or upload a .txt, .docx or .md file. Or write a brief and let AI Create draft it.
Why does my video not match my text?
Usually the shot text asks for more than one action, has no seconds, or describes a feeling. Edit the text first; editing is free.
Does it make speech from my text?
Yes. Lines you write as dialogue are spoken in the video in any of 15 languages.
How much does text to video cost?
Reading is 4 credits per 1,000 characters, video is 10 credits a second on the default tier, and the first storyboard is free.
Can I try it free?
A new account has signup credits, 15 free setup images, a free first storyboard and one free short preview.
Write the brief, read the storyboard
The shot text is free to edit. Video only when it reads right.
Updated November 3, 2026