Hj714110-oss

Idea To Film — Development skill for Claude Code

Development community

One line of idea + one music track → a finished, beat-synced cinematic film.

How to install Idea To Film

This entry records only its repository, not the path inside it, so there is no exact command to give. Open Hj714110-oss/idea-to-film and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What Idea To Film does

One line of idea + one music track → a finished, beat-synced cinematic film. Agent skill for Claude with a deterministic canvas engine. 一句话想法 + 一首音乐 → 卡点成片。

Alternatives in Development

Better Statusline — A status line for Claude Code: the session's figures as dot-matrix rings, and the music you are playing beside 119 ★
Add Music — Generate background music and add it to your Remotion project 113 ★
Generate Assets — Generate real game assets — images, textures, skyboxes, 3D models, SFX, music, voice, and intro/cinematic vide 76 ★

README

idea-to-film

正在输入… TYPING… — title frame

**One line of idea + one music track → a finished, beat-synced cinematic film (MP4).** An agent skill for Claude (works with any agent that reads `SKILL.md`) plus a small deterministic canvas engine.

一句话想法 + 一首音乐 → 一部卡点、有故事、有氛围的成片。知识视频、品牌片、情绪/故事短片都能做。

What it does

Give it an idea — *"how lighthouses work"*, *"a launch film for our goggles"*, *"a love story between a person and an AI"* — and a track. The skill:

Sets the dials itself — length, pacing (TikTok-fast ↔ cinematic), palette, typography, lens and grain — and states them in one line instead of asking.
Writes a 12-beat story with a 0.2 s hook, a recurring motif, two voices, a quiet chapter, a mirror ending and single-frame smash cuts.
Locks every cut to the music (scripts/audio.py fits a beat grid and finds the big entries, the quiet stretch and the dips).
Picks a visual mode
- A — pure code: particles, morphs, type, UI, all drawn in canvas.
- B — footage + code: your clips under a code overlay.
- C — AI plates + image-to-video ("OPUS method"): it writes the image prompts and motion prompts, you generate them with your models, then it crops, upscales, slows, blends and composites the overlay on top.
Renders deterministically with Playwright → frames → ffmpeg, then reviews stills and fixes what it sees.

Motion vocabulary comes from a 178-term film-relevant subset of the **motion-lexicon** (MIT), including 46 compound recipes with code — see *Appendix A* in `SKILL.md`.

Made with it: 《正在输入… TYPING…》

A 3:18 love story between a person and an AI, told through the *typing…* indicator. Every chapter is painted in a different classic style; the code layer adds the chat UI, subtitles, particles and beat sync. 一部 3 分 18 秒的人与 AI 的爱情短片：每一章换一种经典画风，代码层负责聊天界面、字幕、粒子与卡点。

The opening hook</div> </section> </div> <section class=