Guide
English
Search the docs⌘K
6.2

Generating a screenplay

Not sure how to break a story into a professional shot script? DOTV's script node does not invent from nothing. It needs a starting point: a screenplay or a character image. Give it one and it works a director's-eye shot sheet back out of it — how long each shot runs, what it shows, what the shot type is, how the camera moves. You don't need to know storyboard conventions to end up with an AI video script you can take straight into production. This section covers the three ways to start, how to work backwards from a film you already have, and what actually comes out.

Three ways to start

Create a script node and the card lists these three:

Three ways inScreenplay to shot list / Character to shot list each build the source node on the left and wire it up for you. The last one uses no model at all — you fill it in yourself.
The three options on an empty script node: two feed it a source, one you write yourself.
  • Screenplay to shot list — the usual one. Clicking it creates a Text node on the left and wires it in; write the screenplay in there (or have that text node generate it), then come back to the script node and submit.
  • Character to shot list — creates an Image node on the left for a character image. Good when the characters come first and the story second.
  • Write the shot list yourself — no model involved. It starts a list with one empty shot for you to fill in.

The first two create the source node and wire it up for you; all you do is put the content in. Once it is in, select the script node, add a line about what you want in the input bar below (how long, what style, roughly how many shots) and submit.

Starting from a film you already have

If your starting point is a film that already exists (a competitor's, a reference, something you made before), Analyze on the video node's toolbar saves a few steps:

The bar above a finished videoLeft to right: Trim, Crop, Upscale, Analyze, Remove subtitles, Audio separation, Edit frame, with download and fullscreen at the far right. Grab frame sits at the end of the Extract dropdown — the same entry as the camera on the control bar under the picture.
"Analyze" on the video toolbar, a little left of center.

It does the setup for you: it creates a script node to the right of the film, wires the film in and selects it, so the input bar comes up below. That bar is read-only (it reads "Generate a script from video" — there is nothing to add), the model can be swapped at its bottom left, and it starts working the shots out only when you hit send. What comes out is that film's shot structure: how long each shot runs, what it shows, how the camera moves. It is the quickest route when you are breaking a film down or making your own version of one (see 4.7).

What comes out

What comes out is not an essay. It is three things:

What it is Example
Global visual style One line, appended to every image generated afterwards Night city · cinematic · cool tones
Asset list The characters / scenes / props this film needs Lin An (character), the overpass (scene), running shoes (prop)
Shot list One row per shot See the table below
One shot per rowOne shot per row: Duration, Description (@ to reference an asset), Shot type, Lighting, Dialogue, SFX, Camera move. Click any cell to edit it in place.
This is what comes out: one row per shot.

The shot list is where the rest of this chapter happens. Watch for the @Lin An and @overpass in the shot descriptions — they point at entries in the asset list above, and when prompts are composed they are replaced with that character's or scene's full description (see 6.4).

Try it in DOTVReading only goes so far: open DOTV and make an AI video with the feature this section covers.