Skip to main content
To turn a script into video with AI in Automat Studio, you map the script onto the app’s structure rather than pasting it into a prompt box: each character in your script becomes a reusable character asset, each slugline becomes a location and a scene, and each beat within a scene becomes a shot you direct individually. Then you generate the shots and assemble them into the film. That mapping is the whole technique. A screenplay is already broken into the units a production uses, so most of the structuring work is done before you start.

Map the script onto the app

Step 1: Pull the characters out first

Go through the script and create every speaking character before you generate anything. Use the description from your action lines and expand it — hair, build, age, distinctive features, wardrobe, and the way they carry themselves. Assign each one a voice; that is the voice used later when you generate their dialogue. Doing this first is what makes the rest consistent. The character is generated once and then reused, so the person in scene 12 is the person from scene 1.

Step 2: Turn sluglines into locations and scenes

Every distinct slugline becomes a location, described the way a production designer would describe the set. Then create a scene for each one: pick the location, mark it interior or exterior, set the time of day, and add the characters and props that appear. You need at least one location before you can create a scene. If your script returns to the same location at a different time of day, reuse the location — the time of day lives on the scene, not the location.

Step 3: Break each scene into shots

This is where you direct. A page of screenplay is rarely one shot: it is a wide, a couple of singles, maybe an insert. For each shot, choose the lens and shot type, describe the action in that specific framing, and @-mention the characters and props involved so the generation uses the assets you already built. Then generate — either a first frame that you turn into video, or video directly when the shot is simple. Use the Draft quality setting to block out the scene, then regenerate the keepers at Final.

Step 4: Add the dialogue

With the shot’s video in place, generate the character’s line in their assigned voice and apply lip sync. You can also add sound effects and visual effects per shot. If you have a filmed performance you want to drive the shot with — your own take of the action — you can upload it and use performance to video instead, which restyles your real performance rather than generating the motion from scratch.

Step 5: Assemble and export

Watch the cut in the app and share it with a link. Independent plans and above can export the image, video and audio files, and the Hollywood plan exports an XML project for Final Cut Pro or Premiere — which is usually how a script-derived piece gets finished, since you will want real editorial control over timing.

Practical notes

  • Work scene by scene, not front to back. Getting one scene fully right teaches you how to describe the rest.
  • Costs are per generation, in credits. New accounts get 500 free credits with no credit card; after that, packs start at 10andplansat10 and plans at 10/month. Failed generations are not charged.
  • Shot attributes steer rather than dictate. Lens and shot type influence the generation but are not always reflected precisely — expect to iterate on the description for a specific framing.

Next steps