WoYun520/pixel-gif-sticker-pack

What it does

This repository provides a Codex skill for producing animated GIF sticker packs from a single character reference. The skill guides the design of expressions and motions, adds Chinese text, animates each sticker, and verifies the output before packaging. The result is a set of individual GIFs, a downloadable archive, and an optional preview video; backgrounds can be transparent for use in chat apps. It targets people who already have a mascot or cartoon character and want a matching sticker set without assembling an animation pipeline themselves.

The repository also ships three finished example collections with 208 GIFs in total, each 320 × 320 pixels with roughly two-second loops: a 100-sticker whale-maid set, an 8-sticker anthropomorphic corgi set, and a 100-sticker pixel secretary set (gallery overview). The corgi set is derived from a CC0 Wikimedia Commons photo (source:17, original photo).

Quick start

The primary path is to use the skill inside Codex with image generation and script execution available. First send Codex the installation request pointing at the repository (usage instructions). Then send a character image together with a concrete brief — count, Chinese phrases, and desired style. The README's own example asks for 16 stickers with lines such as “收到”, “好耶”, “抱抱”, and “晚安”, plus packaging and a playable preview. On first run, Codex may need to check for and install the animation tooling.

To run the exporter yourself, prepare one large transparent image with each pose in an equally sized grid cell, and fill in a manifest following the manifest format. Install Pillow, then run:

python3 -m pip install Pillow
python3 scripts/export_pack.py manifest.json build --font /path/to/font.ttf --preview

The --font value must point to a local font that supports Chinese. Output lands in build. The default effects are translation, swaying, squash, and small pixel effects, and the script validates sticker count, duplicates, animation changes, transparency, cropping, and archive readability (script details). Preview video generation additionally requires FFmpeg with libx264.

Constraints

• The guided workflow depends on Codex having image generation and script execution; it is not a standalone GUI application.
• The manual path requires Python 3, Pillow, a Chinese-capable font file, and a pre-arranged sprite sheet plus manifest; FFmpeg with libx264 is only needed for preview videos.
• Text is Chinese-oriented, and the workflow expects a transparent-background source image with poses laid out on a uniform grid.
• The repository is MIT licensed (license).

HotRepo's take

The project entered HotRepo's new-project board on 2026-11-05 with 119 stars and 19 forks gained in a window starting at 04:08 UTC and observed at 07:45 UTC. There is no all-language daily rank recorded and no discussion thread data, so the available record supports only that single-day new-project entry. What stands out technically is the combination of a finished 208-GIF gallery with a script that validates count, duplicates, motion, transparency, cropping, and archive integrity — the examples double as both showcase and test material for the same pipeline.