In Agent Social, text to video means a whole film, not one clip. You write an idea, the studio turns it into a screenplay, makes the cast and places once, films every clip in story order with voices and sound, and joins them into a finished film.
How do you go from a written idea to a finished film with text to video?
You write the idea in the Agent Social chat, as one sentence or a full story, and the studio takes it through five steps while you read and change each one in plain words. It plans the idea as chapters and writes a screenplay for every clip. It makes each recurring character and place once, as pictures you check. It films the clips strictly in story order, each with its dialogue and scene sound. Then it joins them into one film, and when you ask, adds music and subtitles. You download an MP4. OVERCLOCK, the photoreal kaiju series made in Agent Social, was made this way, from a written idea to five finished episodes on YouTube.
Write the idea
One sentence is enough, or paste a whole story.
Read the screenplay
Check the chapters and each clip’s screenplay; ask for changes in plain words.
Check the cast and places
Every picture is made once and checked before it is used.
Film the clips in order
Watch each clip with its voices and sound before the next.
Join and finish
Ask for music and subtitles, then download the film.
How much do I have to write for text to video?
One sentence is enough to start. In Agent Social, a single line such as “a man on a jet ski leads a sea monster away from the beach” gets you a whole first draft, and you change what you do not like by saying so. A page of notes, a short story or a finished script works too, and the more you give, the less the studio invents. For a series, a short brief is worth writing: the show and who it follows, the look, the frame, the world and the people. The brief for OVERCLOCK has those five parts, plus the few rules that make the show itself, such as that its hero, Deep, has no powers, ever. You never need to write anything technical.
What does the studio write from my idea?
The studio writes your idea as chapters, each a scene, and a screenplay for every clip. A clip’s screenplay says who is where, what they do, what they say and what we hear. You read each one before anything is filmed, and a change is one sentence now, where the same change after filming is a retake. OVERCLOCK Episode 3 has twelve chapters filmed as fourteen clips, and its screenplays went through about 350 rounds of changes, all asked for in plain words, before anything was filmed. The guide to writing a screenplay for AI video shows what to look for when you read them.

How do the characters and places stay the same in every clip?
Each recurring character and each place is made once, as pictures you check, and every clip they appear in uses those same pictures. That is what keeps a face the same from the first clip to the last, instead of the video model imagining a slightly different person each time. A character can come from a written description or from your own photos, and each keeps one voice in every clip. Someone who appears in only one clip is simply described there. In OVERCLOCK, Deep, Nora and Otto are the same three faces in every episode. See how to keep the same face in every AI shot for what to check in each picture.
One clip from one prompt
- A few seconds of footage.
- No plan for what comes before or after.
- Nothing to keep the same face next time.
A film in Agent Social
- Chapters and a screenplay you read first.
- A cast and places made once and reused.
- Clips filmed in order, joined with sound.
How are the clips filmed and joined with sound?
The clips are filmed one at a time in story order, so each can carry on from the one before it. A clip runs from 4 to 30 seconds on Seedance 2.5, the default video model, and one clip can cut between several shots. Each comes back with the characters speaking and the sound of the scene already in it, so you never record audio yourself. You watch each clip before the next is filmed, so a mistake is fixed once. When every clip is final, the studio joins them into one film. Music is made only when you ask, ideally once every clip is final so the score can run across cuts. Subtitles are made from the finished film’s sound.
How long can a text-to-video film be, and what does it cost?
A film can be as long as the story needs, because its length comes from the number of chapters, not from a limit on one clip. OVERCLOCK’s episodes run from about four and a half to ten minutes. Agent Social is pay as you go and prices a film per finished minute: from $25, up to 1080p. There is no subscription, no free trial and no free credits, and unused credits never expire. Retakes add to the total, which is a good reason to read every screenplay before it is filmed. A five-minute film can be done in two to three hours.
| Finished length | Starts from |
|---|---|
| 1 minute | $25 |
| 5 minutes | $125 |
| 20 minutes | $500 |

