A 7-minute hand-drawn watercolor animation made without drawing skills: GPT writes the script, Claude builds it with a skill, and Doubao TTS does the voice
Hailey polishes a script and storyboard with GPT in three rounds, has Opus 5.5 build a self-drawing watercolor animation in the browser using her open-source hand-drawn-storybook-animation skill, then adds narration and character voices with Volcengine's Doubao TTS and synced bilingual subtitles. She says it took one day.
Visit site Claude (Opus 5.5) + 豆包语音 ↗- Pricing
- Paid
- Region
- Global
- Where it's from
- Chinese community
- Added
- Oct 18, 2026
- Tested on
- Oct 18, 2026
- Version we used
- hand-drawn-storybook-animation-skill main (pushed 2026-10-16); author used Opus 5.5 + Doubao Speech Synthesis 2.0
- Original author
- Hailey @HaileyAILabs ↗
- Our test
- Partly tested
- Cost / time
- A paid Claude plan with Opus 5.5 (or Claude Code), ChatGPT for the script, and Volcengine Doubao TTS (pay-as-you-go with a free allowance). The result is a web page you can host for free.
The task
One person with no drawing or editing skills wants to turn a story into an animated picture book with a consistent art style, voices and subtitles.
Steps
- Round 1, story: ask GPT (her prompt, translated) "Design a script for a creative hand-drawn piece: healing, engaging, and shareable," covering story, scenes, character actions and lines.
- Round 2, self-critique: "As a picky viewer and an animation director, criticize this script: where it's flat, where it doesn't make sense, where it isn't worth sharing, then give me the revised version." She says not to skip this.
- Round 3: have GPT turn the revised script into one storyboard sheet, one panel per shot, labeled wide/medium/close-up, with the action and line, in pencil sketch plus light watercolor.
- Install the skill: in Claude Code copy the repo's hand-drawn-storybook-animation folder into ~/.claude/skills/; on claude.ai zip the folder and upload it as a skill.
- Send the script and storyboard to Opus 5.5 with the prompt below for a first cut, then give blunt notes ("too slow", "brighter colors", "transitions look bad", "music too loud") and let it revise.
- Voices: enable speech synthesis in the Volcengine console (she used Speech Synthesis 2.0, which has a free tier), create an API key, put it in the project's .env, then tell Opus 5.5 to voice it with Doubao TTS, giving the narrator and each character a tone and asking for subtitles synced to the audio.
- Watch the whole thing before publishing, checking that lines, subtitles and visuals line up.
Prompt
Use the hand-drawn-storybook-animation skill to make a hand-drawn watercolor animation from this script and storyboard.
What the author says
The author (2026-10-17) says she made the 7-minute "The Little Fox Who Chased the Stars" this way, plus pieces on the Qingming scroll and Halloween, with a Nüwa myth in progress, each taking about a day from script to finish. The skill documents a layered Canvas/SVG self-drawing engine, page-turn/wet-brush/paper-tear transitions, a gentle generated score, and the Doubao Speech Synthesis 2.0 API, voices and a per-line caching script.
Our test
We did not rebuild the animation with Opus 5.5 or call Doubao TTS. We only checked the open-source skill repo and its preview clips.
What we found when we tried it:Checked 2026-10-18: the GitHub repo lhlGitHub/hand-drawn-storybook-animation-skill is live (about 40 stars, no license declared) with a SKILL.md (~9.8 KB) and five preview GIF/MP4 clips. SKILL.md does document the layered Canvas/SVG engine, three transition types, the Doubao Speech Synthesis 2.0 endpoint and request format, the per-line caching doubao.py script, and Playwright screenshot verification. The preview 01-self-drawing.mp4 we downloaded is 1280×720, about 8.1 s.
How to get in
The original is a Chinese X article; the skill README is bilingual. Doubao TTS is on ByteDance's Volcengine (volcengine.com); the international BytePlus site may differ, and English voices can be swapped for any TTS you have.
Who it suits
Solo creators making bedtime stories, kids' explainers, animated picture books or short brand stories, and anyone who wants a portfolio piece fast.
Pitfalls
- Don't rush into production: she says the script decides the result, so always have GPT critique it first.
- Don't fill every second with music; quiet stretches with only ambient sound work better.
- "Done" from the AI isn't ready to publish; watch pacing, transitions and voices yourself.
- Never paste the Volcengine key from .env into the chat or commit it.
- The skill notes that cloud sessions may be blocked from the Doubao TTS endpoint by network allowlists; running the voice script on your own machine is most reliable.
- Use original characters; someone else's popular IP isn't yours to build a series on.
Images

More in this category
- A demo video per docs page that flags itself when the UI changesVideoScreenplayClaude CodeTool free; uses your agent's quota◐ Partly testedOverseas
- Code-drawn stick-figure explainers for about $0.54 eachVideoClaude CodeFFmpegChromeNeeds Claude plan or API; renders locally○ Not tested by usChina
- One topic in, a narrated tech explainer video out, locallyVideoClaude CodeChatterboxffmpegSkill free; needs a Claude Code plan◐ Partly testedOverseas
- Render restrained motion graphics from code, frame by frameVideoClaude CodePlaywrightffmpegNeeds a Claude plan; other tools free◐ Partly testedOverseas
- One reference image → a take-apart machine simVideoteardownClaude Codethree.jsSkill free; needs Claude Code/Opus◐ Partly testedOverseas