Data

Tools

Indexes

Applications

Experiments

Open Source Science

https://github.com/cyberagentailab/mg-gen

Project page for "MG-Gen: Single Image to Motion Graphics Generation with Layer Decomposition"

https://github.com/cyberagentailab/mg-gen

Science Score: 36.0%

This score indicates how likely this project is to be science-related based on various indicators:

  • ○
    CITATION.cff file
  • ✓
    codemeta.json file
    Found codemeta.json file
  • ✓
    .zenodo.json file
    Found .zenodo.json file
  • ○
    DOI references
  • ✓
    Academic publication links
    Links to: arxiv.org
  • ○
    Academic email domains
  • ○
    Institutional organization owner
  • ○
    JOSS paper metadata
  • ○
    Scientific vocabulary similarity
    Low similarity (10.7%) to scientific vocabulary
Last synced: 10 months ago · JSON representation

Repository

Project page for "MG-Gen: Single Image to Motion Graphics Generation with Layer Decomposition"

Basic Info
  • Host: GitHub
  • Owner: CyberAgentAILab
  • License: agpl-3.0
  • Language: Python
  • Default Branch: main
  • Size: 4.59 MB
Statistics
  • Stars: 3
  • Watchers: 1
  • Forks: 0
  • Open Issues: 0
  • Releases: 0
Created about 1 year ago · Last pushed about 1 year ago
Metadata Files
Readme License

README.md

License: AGPL v3

MG-Gen: Single Image to Motion Graphics Generation with Layer Decopmosition

This is an official repository for MG-Gen.
MG-Gen is a novel method to generate motion graphics from a single raster image preserving input content consistency with dynamic text motion.

  • Paper: https://arxiv.org/abs/2504.02361
  • Project Page: https://cyberagentailab.github.io/mggen/

Setup Experimental Environment

Create and activate a Python venv. (Requirements: python >= 3.10) python -m venv .venv source .venv/bin/activate

Install dependencies. pip install -r requirements.txt playwright install

Download anime.min.js mkdir src/libs/ curl -o src/libs/anime.min.js https://raw.githubusercontent.com/juliangarnier/anime/3.2.0/lib/anime.min.js

Download the weights from GCS. If you have not installed gsutil, see the installation instructions. gsutil -m cp gs://ailab-public/image-to-video/mggen/weights.zip . unzip weights.zip

Generate motion graphics from images

Set your Gemini APIKEY in .env. You can get an API key for free from Google AI Studio. (20250331) ``` echo "GEMINIAPIKEY=\"your-api-key\"" > .env ```

Start a gradio demo sever. cd src python demo.py

Run a generation batch script. cd src python inference_batch.py --testset_dir "../example_inputs"

Demo videos demo videos

License

This project is licensed under the GNU Affero General Public License v3.0 (AGPL-3.0).

Citation

bibtex @article{shirakawa2025mg, title={MG-Gen: Single Image to Motion Graphics Generation with Layer Decomposition}, author={Shirakawa, Takahiro and Suzuki, Tomoyuki and Haraguchi, Daichi}, journal={arXiv preprint arXiv:2504.02361}, year={2025} }

Owner

  • Name: CyberAgent AI Lab
  • Login: CyberAgentAILab
  • Kind: organization
  • Location: Japan

GitHub Events

Total
  • Watch event: 3
  • Member event: 1
  • Public event: 1
Last Year
  • Watch event: 3
  • Member event: 1
  • Public event: 1

Dependencies

requirements.txt pypi
  • einops *
  • google-genai >=1.5.0
  • gradio *
  • imageio-ffmpeg *
  • omegaconf *
  • opencv-python *
  • paddleocr >=2.0.1
  • paddlepaddle-gpu *
  • pillow *
  • playwright >=1.50.0
  • pycocotools ==2.0.8
  • python-dotenv *
  • scikit-image *
  • torch ==2.4.1
  • torchvision ==0.19.1
  • tqdm *
  • transformers ==4.45.2
  • ultralytics *