open-mmlab/mmagic
OpenMMLab Multimodal Advanced, Generative, and Intelligent Creation Toolbox. Unlock the magic 🪄: Generative-AI (AIGC), easy-to-use APIs, awsome model zoo, diffusion models, for text-to-image generation, image/video restoration/enhancement, etc.
Built on PyTorch with the OpenMMLab 2.0 framework, MMagic unifies diffusion models, GANs, and CNN-based architectures through a standardized DataSample and DataPreprocessor interface, enabling seamless switching between generation and reconstruction tasks. It integrates xFormers optimization, DiffuserWrapper for sampling flexibility, and ControlNet for conditional generation, while supporting both single-image operations and batch video processing through refactored MultiValLoop/MultiTestLoop evaluation pipelines. The toolkit consolidates MMEditing and MMGeneration codebases with enhanced support for model composition, fine-tuning methods like DreamBooth LoRA, and multi-dataset evaluation with both generative (FID) and reconstruction (SSIM) metrics.
7,402 stars and 820 monthly downloads. No commits in the last 6 months. Available on PyPI.
Stars
7,402
Forks
1,100
Language
Jupyter Notebook
License
Apache-2.0
Category
Last pushed
Aug 06, 2024
Monthly downloads
820
Commits (30d)
0
Dependencies
17
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/generative-ai/open-mmlab/mmagic"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Related tools
jdh-algo/JoyVASA
Diffusion-based Portrait and Animal Animation
haidog-yaqub/EzAudio
High-quality Text-to-Audio Generation with Efficient Diffusion Transformer
CMLab-Korea/Awesome-Video-Frame-Interpolation
[IEEE TCSVT'26] 🂡 AceVFI: A Comprehensive Survey of Advances in Video Frame Interpolation
linzhiqiu/t2v_metrics
Evaluating text-to-image/video/3D models with VQAScore
TIGER-AI-Lab/AnyV2V
Code and data for "AnyV2V: A Tuning-Free Framework For Any Video-to-Video Editing Tasks" [TMLR 2024]