A stickman video director that lives inside Codex
It transforms raw text into six-scene director proposals and locked-down Gemini Omni Flash prompts so you can produce consistent one-minute explainers without writing a shot list or calling an API.

What it does
This Codex Skill ingests your notes, articles, or half-baked ideas and exports a complete one-minute video production package. It writes a ~130-word English voice-over, breaks the narrative into six timed scenes, and bakes character design, palette, camera moves, and transitions into every prompt so Gemini Omni Flash can render clips that actually cut together. Everything happens inside a Codex chat session—no API wrangling, no MCP plumbing.
The interesting bit
Most AI video workflows cross their fingers and prompt. This one behaves like a line producer with a sign-off sheet: it will not generate final prompts until you explicitly approve the six-scene director proposal, and if you change aspect ratio it rethinks the entire composition rather than swapping a tag. It also knows each 10-second segment is generated independently, so it repeats consistency locks—line thickness, character details, audio cues—in every single prompt to fight drift.
Key highlights
- Handles
9:16,16:9, and1:1with scene-specific framing and text placement instead of simple ratio labels - Ships with two finished visual families: high-contrast minimalist and a “Modern Beanie Zeke” style the author says avoids limb-jank and eyeball-melting artifacts
- Produces ~130–150 words of English VO and six self-contained prompts with timing beats, visual metaphors, and negative constraints
- Built for Google Flow / Gemini Omni Flash; the README notes you can generate roughly sixty 10-second clips per month on the included Pro tier credits
- Roadmap includes nine additional styles, from sitcom dialogue to cave-punk aesthetics and RPG card-battle explainers
Caveats
- Audio consistency across independently generated segments is explicitly flagged as a weak point; the author recommends replacing generated sound with a continuous external track for best results
- This is a curated prompt framework and workflow, not a rendering engine—you still need access to Gemini Omni Flash to make pixels move
- Several promised styles and a dedicated website are on the roadmap but not yet shipped
Verdict
Short-form educators and explainer-channel operators who want structured storyboards without learning After Effects should look here. If you are hoping for a one-click video renderer, this is just the clipboard.
Frequently asked
- What is kaomei/stickman-video-director?
- It transforms raw text into six-scene director proposals and locked-down Gemini Omni Flash prompts so you can produce consistent one-minute explainers without writing a shot list or calling an API.
- Is stickman-video-director open source?
- Yes — kaomei/stickman-video-director is open source, released under the MIT license.
- What language is stickman-video-director written in?
- kaomei/stickman-video-director is primarily written in Shell.
- How popular is stickman-video-director?
- kaomei/stickman-video-director has 1.1k stars on GitHub.
- Where can I find stickman-video-director?
- kaomei/stickman-video-director is on GitHub at https://github.com/kaomei/stickman-video-director.