Video studio: character, scene and voice

You give the character, scene, dialogue, voice and style; a model plays the director and writes a precise shot script, the video model makes the clip with synced sound, and in parallel you get a character poster and a separate voice-over.

Advanced10 min

In this lesson

  • Several structured inputs instead of one prompt
  • A language model as the director (prompt writer)
  • Keeping the character consistent with an exact description
  • Three media outputs from one run: video, image, audio

Build it yourself

This workflow is created in your account and opened in the editor, where you can change and run it.

The nodes in this workflow

Character

character

Face, age, clothes and manner; the more precise, the more consistent the character across videos.

Scene

scene

Place, time of day, weather and what happens.

Dialogue

dialogue

What the character says (in any language).

Voice and tone

voice

Gender, age and feel of the voice, e.g. “male, calm and warm”.

Visual style

style

This studio’s fixed look; locked so every video matches.

Text template

{{character}} {{scene}} {{dialogue}} {{voice}} {{style}}

Gathers every input into one tidy brief.

Director

Claude Sonnet

The director: writes one precise visual prompt in English — shots, camera moves, light, beat-by-beat action, the dialogue in its own language and ambient sound.

Generate video

8 s · 1280×720

Makes the clip with synced sound (dialogue and ambience); takes a few minutes and shows its progress.

Video

video

Returns the result under this key in the API response and in chat.

Shot script

script

Returns the result under this key in the API response and in chat.

Text template

{{character}} {{scene}} {{style}}

Builds the poster description from the same character and style.

Generate image

poster 1024×1536

A portrait poster of the character, for covers and social media.

Poster

poster

Returns the result under this key in the API response and in chat.

Text to speech

voice: onyx

Records the same dialogue separately in the requested tone, for dubbing or editing.

Voice-over

voiceover

Returns the result under this key in the API response and in chat.