On the DOTV script card, click Edit at the top (or double-click the card) to go full screen. Step one, Confirm shots, is this table: one row per shot, and every cell — duration, description, shot type, lighting, dialogue, sound and camera move — can be edited directly. This step is worth taking slowly — a word off in this table is a dozen wasted images and wasted credits later on; confirming the storyboard here, before any images or video are generated, is what keeps AI video creation cheap.
| Column | What goes in it | Example |
|---|---|---|
| Shot | Which shot it is, numbered by the current order | 2 |
| Duration | How long this shot runs | 8s |
| Description | What this shot shows. @ pulls in a character / scene / prop from the asset list | @Lin An ties his laces and sets off, through the tunnel under the @Overpass |
| Shot type | Twelve in the dropdown: Extreme wide / Wide / Full / Medium wide / Medium / Medium close / Close / Close-up / Extreme close-up / Head and shoulders / Waist up / Full body | Medium |
| Lighting | What the light is doing in this shot | Overhead lamps sweeping past one by one |
| Dialogue | You can have several lines, each of them either Dialogue or Narration | Lin An: One more kilometer and I head back. |
| SFX | Ambience and effects | Footsteps, breathing |
| Camera move | Six presets, one click each: Push in / Pull out / Pan / Follow / High angle / Low angle — or write your own | Follow |
| Final prompt | The paragraph step three composes (see 6.4) |
Every cell is edited directly: click into it and you get an input or a dropdown, never a dialog. In the dialogue cell you can also bind each line to a character — once bound, batch dubbing renders that line in that character's voice (see 6.5).
The step bar at the top reads N shots to review or N shots ready. There is only one bar for ready: the description is filled in. The other cells are for you — shot type, lighting and camera move can all be blank and the shot still counts as ready, but then the model can only guess, and its guesses are mostly waste.
So the order is: say it clearly in this table first, then batch-generate. Changing a word in the table takes a second; a botched image and a redo takes tens of seconds and a charge of credits.