Lip-Sync Story Studio · compétence Open Skill
Lip-Sync Story Studio
Lip-Sync Story Studio turns one photo (or two) plus a one-line idea into a 45-second to 2-minute vertical animated episode, and your characters really talk. Highlights: - Real English lip sync: characters speak their lines on screen with mouth movement that follows the actual words, not gibberish. - Locked voices: each character's voice is designed once, approved by you, and reused for every line in every episode. - Four art styles: high-contrast vector, Afrocentric watercolor-painterly, torn-paper collage, or classic 2D cartoon. - Story your way: Start from any idea, or pick a two-character quick-start: Love Story, Rivals, or Team-Up. - Narrator or not: choose a narrated story with lip-synced dialogue at key moments, or let the characters tell it all. - Locked looks: every character is transcribed into a Visual DNA card and checked for drift in every shot. - Every spoken line is checked word for word; any clip with missing or garbled words is redone before you see it. - Continue the series anytime, characters, style, and voices stay locked. Why your credits aren't wasted: Nothing moves forward until you approve it. Characters, voices, script, keyframes, and clips each get their own checkpoint. Finished episodes come with captions for TikTok, Instagram, Facebook, and YouTube.
ShortDrama
7 1 5
https://d7qlwi7tl2u27.cloudfront.net/skill-covers/prod/925535/2026-09-25/8bv7dxxm.png
---
name: lipsync_story_studio
display_name: Lip-Sync Story Studio
description: Turn a photo (or two) and a one-line idea into a short vertical animated episode where characters actually speak their lines in English, with mouths moving to the words. Choose high-contrast vector, watercolor-painterly, torn-paper collage, or classic 2D cartoon. Locked character looks, locked voices, optional narrator. Default cast is African-American.
allowed-tools:
- generate_image
- generate_video
- generate_tts
- generate_music
- mix_clips
- recognize_image
- recognize_video
- transcribe_audio
- split_image
- create_document
- get_document
- update_document
- memory_query
- memory_store
- memory_forget
- list_models
- download_video
---
# Lip-Sync Story Studio
## When to use
- User wants a short vertical (9:16) animated episode, 45 seconds to 2 minutes, where characters speak English dialogue on screen with mouth movement matched to the words.
- User wants to continue an existing series, redo a shot, or change a character.
## Tool facts (confirmed with Digen, do not deviate)
- `generate_video` and `generate_image` take **channel IDs only** (`t2i`, `i2i`, `i2i.hd`, `i2v`, `ref2v`, `ai2v`...). Never pass concrete model names (e.g. `rm3.5-ref`, `nano_banana2`), `width`, `height`, or `gen_mode` — they are rejected. Size is set with `orientation`, `aspect_ratio`, and `resolution`.
- `generate_tts` takes `text`, `model` (`tts.design` or `tts.clone`), `voice_instruction` (design only, English), `reference_audio_url` (clone only), `name`, `entity_type`. There is no `mode` parameter. It does **not** return a duration — estimate speech length from word count.
- `mix_clips` clip object: `{video:{url, volume}, audios:[{url, delay, volume}]}`. Times are seconds; `delay` is relative to the clip's start; `volume` is a 0.0–2.0 multiplier. `video.volume: 0` mutes the clip's baked-in audio.
- There is no last-frame extraction tool and no video upscaler.
## Fixed preferences
- **Orientation:** always 9:16 vertical (`orientation='portrait'`, `aspect_ratio='9:16'`) on every image and video call.
- **Resolution:** always request the highest tier (`resolution='1080P'` for images and non-talking video; the talking-shot route below states its own). Never hard-code a low resolution.
- **Episode length:** 45 s – 2 min, set by how many shots the story needs, not a fixed count. Shots are 4–10 s.
- **Default cast:** contemporary African-American characters unless the user specifies otherwise — contemporary American names, a US setting by default, contemporary Black American hair and fashion. Describe skin tone with layered descriptive terms (deep ebony, rich dark brown, deep mahogany, warm brown, warm cocoa, caramel, honey brown, golden brown), rotated across the cast, never racial labels.
- **Language:** all user-facing text, dialogue, and captions in English.
- Family-friendly by default; decline adult/NSFW requests.
## Art styles (user picks one per series; lock it for every call)
Write the chosen style block, word for word, into every image and video prompt.
- **High-contrast vector:** "high-contrast vector illustration, bold clean outlines, flat graphic color blocks, crisp shapes, dramatic lighting with hard-edged shadows, limited rich palette, smooth vector animation"
- **Watercolor-painterly:** "Afrocentric watercolor-painterly illustration, soft wet-edge washes, visible paper texture, luminous layered color, gentle ink linework, hand-painted animated feel"
- **Torn-paper collage:** "torn-paper collage animation, layered cut and torn paper shapes with visible fibrous edges, textured paper grain, mixed patterned papers, subtle drop shadows between layers, stop-motion-like paper movement"
- **Classic 2D cartoon:** "classic 2D animated feature style, clean expressive linework, vibrant saturated color, soft cel shading, fluid expressive character animation, cinematic camera"
## Story modes (ask once per series)
- **Open idea (defa