DA Motion Sticker Skill
DA Motion Sticker Skill turns one character reference and nine reactions into nine separate transparent animated GIFs, with 36 styles, Codex key-pose and external AI-video routes, local keying, validation, and ZIP packaging.
Project overview
DA Motion Sticker Skill is for turning one character into a complete chat-sticker pack. It takes a reference image and nine reactions, builds a fixed 3×3 source sheet, uses either Codex key poses or an external AI-video route for motion, and then splits, keys, encodes, validates, and packages the results locally. The repository requires all nine GIFs to work independently, contain real transparency, and pass loop and completeness checks instead of delivering only a grid preview. This page is based on public GitHub files and CLI help output, without claiming an unperformed generation or video-processing test.
Core capabilities
Nine reactions and 36-style planning
Users can provide nine reactions, actions, or emoji directly, or give a theme and approve the Skill's nine-item expansion before selecting one of 36 presets.
Transparent sheet and real grid splitting
prepare_sheet.py creates real Alpha from a flat edge-connected background, detects 3×3 gutters, writes nine PNG cells, and selects a lower-conflict screen color from foreground pixels.
Codex key-pose route
Each cell receives genuine start, anticipation, peak, and recovery poses before deterministic sequencing. Failed identity or pose generation must not fall back to whole-layer movement, rotation, or shake.
External AI-video route
A chroma-key sheet can be sent to a user-selected image-to-video service, then processed locally to detect the grid, remove only near-screen regions connected to crop edges, and encode transparent GIFs.
Completeness checks and delivery packaging
package_delivery.py requires exactly nine GIFs and checks at least two frames, infinite looping, a transparency index, matching dimensions, and nonempty content before packaging reports, sheets, prompts, and a manifest into a ZIP.
Installation and usage
Clone the repository into the local Skills directory scanned by Codex, install the Python requirements, and refresh or restart Codex. Confirm FFmpeg and FFprobe before starting the video route.
Let an AI Agent install it
Send this prompt to Codex, Claude Code, or another AI agent that can work with your local environment.
Help me install DA Motion Sticker Skill from https://github.com/avocadotear/da-motion-sticker-skill. Read README.en.md, SKILL.md, LICENSE, requirements.txt, references/output-contract.md, references/keypose-route.md, and references/video-route.md first. Clone the repository into `$HOME/.agents/skills/da-motion-sticker-skill`. If that directory already exists or has uncommitted changes, do not overwrite it; ask me first. Explain which Python environment will be modified before installing requirements.txt. Ask before administrator access, installing FFmpeg or FFprobe, using an external image-to-video service, uploading character media, incurring cost, or replacing a delivery directory. Refresh Codex, run `python scripts/compile_prompts.py --list-styles`, and confirm that all 36 styles appear. Use only a nonsensitive test image for minimal verification and do not upload or publish media without approval.- Git
- Python 3.9 or newer
- Pillow 9.0 or newer and NumPy 1.21 or newer
- Codex with local Agent Skills support
- FFmpeg and FFprobe for the video route
- A character reference that you have the right to use and that contains no unauthorized sensitive data
git clone https://github.com/avocadotear/da-motion-sticker-skill.git "$HOME/.agents/skills/da-motion-sticker-skill"- 1Review the environment and media boundaries
Confirm Git, Python 3.9+, Pillow, and NumPy, and prepare a nonsensitive character image that you have the right to use. For the video route, also review FFmpeg, FFprobe, external-service cost, upload, and retention terms.
- 2Run the documented Codex installation
Run: git clone https://github.com/avocadotear/da-motion-sticker-skill.git "$HOME/.agents/skills/da-motion-sticker-skill" If the directory exists, inspect uncommitted changes before using the README's git pull --ff-only update path.
- 3Install Python requirements and inspect styles
Run `python -m pip install -r requirements.txt` in the repository, then `python scripts/compile_prompts.py --list-styles` and confirm that all 36 presets appear. Avoid modifying an unknown global Python environment.
- 4Confirm nine reactions and a style
Attach the character reference in Codex and provide nine reactions, or give a theme and review the Skill's proposed nine items. If no style is selected, choose from three presets recommended from the character's medium and silhouette.
- 5Choose an animation route and complete local acceptance
After the transparent and chroma-key sheets are ready, choose the Codex key-pose or external AI-video route. Create the final ZIP only after all nine GIFs pass Alpha, frame-count, transparency-index, loop, boundary, and nonempty-content checks.
Refresh or restart Codex and confirm that `$da-motion-sticker-skill` is discoverable. From the repository, run `python scripts/compile_prompts.py --list-styles` and confirm that 36 styles are listed. Then use a nonsensitive test character only to confirm the nine-reaction intake and route explanation, without uploading or generating video unless authorized.
- Use git pull --ff-only for an existing clean Skill directory; do not overwrite a development copy with uncommitted changes
- The video route requires FFmpeg and FFprobe. Palette quality on the key-pose route may be lower without FFmpeg
- External video tools may charge fees and apply their own upload, training, and retention terms
- Review local paths, character rights, and service terms before sharing reports or the final ZIP
Use cases
Personal character chat stickers
Expand a person, pet, or virtual character into nine individually sendable transparent animated GIFs for reactions such as happy, angry, confused, thumbs-up, and goodbye.
Brand-mascot style exploration
Choose a consistent direction from plush, pixel, traditional art, Riso, Y2K, retro computer UI, and the other 36 presets while preserving recognizable character traits.
Controllable key-pose animation in Codex
Generate genuine anticipation, peak, and recovery poses for each cell and sequence them locally instead of faking action by shaking a single image.
Local transparency after external video generation
Send the chroma-key grid to an image-to-video service for continuous motion, then use local scripts for grid detection, edge-connected keying, and GIF encoding.
Assessment
Based on the README, SKILL.md, Python CLI entry points, and output contract, DA Motion Sticker Skill turns character consistency, transparency, cell boundaries, and delivery completeness into explicit checks, with separate paths for in-Codex generation and external video services. Its main constraints are generation tools, FFmpeg, external services, and media-rights boundaries. This assessment uses public material and CLI help output without a real character-generation or end-to-end GIF task.
Why it may be useful
- The 36 presets span plush, pixel, comics, traditional art, and retro UI with visibly distinct directions
- The Codex route requires genuine key-pose changes and explicitly forbids whole-layer translation, rotation, or shake as fake motion
- Local keying removes only near-screen regions connected to crop edges, reducing the chance of deleting matching colors inside the character
- Final packaging requires exactly nine GIFs and rechecks frame count, transparency index, looping, dimensions, and empty frames
What to know first
- The workflow targets exactly nine separate GIFs in a 3×3 sticker pack, not ordinary image edits, single stickers, or generic GIF conversion
- The full workflow needs a callable reference-image generator, while the video route also depends on an external image-to-video service
- GIF uses binary transparency, so edge quality is usually less natural than transparent PNG or animated WebP
- Character consistency, key-pose generation, video keying, and final nine-GIF quality were not tested for this page
- FFmpeg, FFprobe, and pytest were unavailable locally, so video processing and the repository test suite were not run
README
DA Motion Sticker Skill
Overview
DA Motion Sticker Skill turns one character reference and nine reactions into nine separate transparent animated GIFs, with 36 styles, Codex key-pose and external AI-video routes, local keying, validation, and ZIP packaging. Codex skill helpers for nine-grid transparent animated sticker packs
Getting started
- Clone the repository into the local Skills directory scanned by Codex, install the Python requirements, and refresh or restart Codex. Confirm FFmpeg and FFprobe before starting the video route.
- Project focus: turn one character reference and nine reactions into nine separate, transparent, infinitely looping animated GIF stickers.
- Published features: 36 styles; Codex key-pose and external AI-video routes; optional static PNGs; local Alpha, grid, frame-count, transparency-index, and loop checks.
- Refresh or restart Codex and confirm that `$da-motion-sticker-skill` is discoverable. From the repository, run `python scripts/compile_prompts.py --list-styles` and confirm that 36 styles are listed. Then use a nonsensitive test character only to confirm the nine-reaction intake and route explanation, without uploading or generating video unless authorized.
Configuration
git clone https://github.com/avocadotear/da-motion-sticker-skill.git "$HOME/.agents/skills/da-motion-sticker-skill"