Director Flow Storyboard Drama and Series · publiczny skill Open Skill

Director Flow Storyboard Drama and Series

Director Flow Storyboard Drama and Series to publiczny Open Skill w Digen. Otwórz go, by zobaczyć, co tworzy ten workflow.

@jeneek3337

Film

30 4 26

https://d7qlwi7tl2u27.cloudfront.net/skill-covers/prod/2173956/2026-09-25/6sznvebw.png

# Director Flow Storyboard Drama and Series

## When to use
- The user wants a complete prestige-TV-drama storyboard divided into ~15-second scenes, each designed to be one Seedance / short video clip, from a one-line idea or a set of inputs (genre, setting, tone, goal, ending style, style reference, platform, signature prop).
- The user wants the skill to auto-plan how many episodes the story needs (no limit) so the story flows well, and to produce a cut per episode.
- The user wants to revise one scene, regenerate one clip, or stitch the generated clips into episode cuts.

## Reference files
- `references/storyboard-rules.md`: The full SEEDANCE 2.0 prestige-drama master prompt (15-second scenes, continuous story, voice actor lock, seamless transitions, anti-AI-slop, spatial/180-degree, match-cut, camera, lighting, audio, transition quality test, no-teleportation, final directives). **Read it before producing any storyboard.**
- `references/character-sheet-template.md`: Fill out one sheet per character to lock appearance and voice. **Read it when establishing characters.**

## Language lock (spoken dialogue)
Apply these rules whenever a scene contains spoken dialogue, and follow them when writing every clip prompt in step 6. The video model must produce the spoken language directly from the prompt — no TTS/voiceover is used.
1. **LANGUAGE LOCK**: Always state the spoken language explicitly and prominently at the START of the video prompt, e.g. "The characters speak in BAHASA MALAYSIA (Malay) throughout." Never rely on the storyboard header alone.
2. **EMBED THE LINE**: Put the exact spoken line in double quotes inside the prompt, right where the character speaks, e.g. 'ASP Faridah (female police officer, speaking in BAHASA MALAYSIA) says: "Semua unit, serbu." Her lips move as she delivers this line.'
3. **CHARACTER + LANGUAGE TOGETHER**: For every speaking character, pair the character name, the language, and the quoted line in the same sentence so the model ties them together.
4. **DRIVE THE LANGUAGE IN THE VIDEO**: Write the prompt so the video model renders the dialogue in the requested language, making the language instruction as explicit and prominent as possible. Do not rely on any external voiceover.
5. **NEVER GUARANTEE**: Never tell the user the video "will" speak the language. State it as "the prompt requests X language; the model may still improvise." If the output language is wrong, do NOT fall back to TTS — regenerate the clip with a clearer, more explicit language prompt instead.

## Voice actor lock
- Every recurring character must sound like the same actor throughout the entire series (voice identity, pitch, texture, accent, dialect, rhythm, speed, articulation, emotional baseline). Emotion changes the performance but must NOT change the actor's fundamental voice identity. See `references/storyboard-rules.md` sections 15–17. Fill the voice profile in the character sheet for every important character.

## Dialogue-to-character lock
- Every spoken line MUST be explicitly attributed to the exact named character in the prompt. Never leave a line ambiguous, floating, or unattributed.
- For each speaking character, write: "<Character Name> (role, appearance, speaking in <LANGUAGE>) says: "<exact line>". His/her lips move as he/she delivers this line."
- Only the character who owns the line speaks it. All other characters present are LISTENING — their lips stay still (or react silently) while the speaker talks. Never let a non-speaking character's mouth move or deliver another character's line.
- Pair the speaker's name, the language, and the quoted line in the same sentence (see Language lock rule 3) so the model cannot confuse who is talking.
- If two characters speak in the same scene, sequence them clearly and attribute each line to its owner; never merge, swap, or reassign speakers.
- In the storyboard scene format, always label every line of dialogue with the speaker's name before the line (e.g. "FARIDAH: Semua