open-mmlab/mmagic

Project overview

MMagic is an open-source toolkit for AIGC tasks including image and video restoration, inpainting, super-resolution, and generation. It provides modular components for researchers and engineers who build and customize generative and restoration pipelines using PyTorch.

Project type

Model Development · Model Runtime · Image & Vision · Video

Deployment

Refer to project documentation

License

Apache-2.0

Best for

• Researchers and AI engineers who need to programmatically construct customized image and video editing pipelines using modular components.

Key capabilities

• Provides restoration algorithms including NAFNet, Restormer, and SwinIR.
• Supports fine-tuning methods such as Dreambooth and DreamBooth LoRA.
• Supports video generation based on MultiFrame Render.
• Supports image inpainting and matting applications.
• Provides GAN-based super-resolution algorithms like SRCNN and ESRGAN.
• Decomposes the editing framework into modules that can be combined to create customized editors.

Limitations and risks

• Best practice operation requires Python 3.9 or higher.

Getting started

• Install PyTorch, then use MIM to install MMCV, MMEngine, and MMagic. Verify the setup by importing the library and checking its version in Python.

Evidence and sources

• README: MMagic (**M**ultimodal **A**dvanced, **G**enerative, and **I**ntelligent **C**reation) is an advanced and comprehensive AIGC toolkit that inherits from [MMEditing](https://github.…
• README: MMagic will provide more agile and flexible experimental support for researchers and AIGC enthusiasts, and help you on your AIGC exploration journey.
• README: MMagic provides state-of-the-art generative models to process, edit and synthesize images and videos.
• README: By using MMEngine and MMCV of OpenMMLab 2.0 framework, MMagic decompose the editing framework into different modules and one can easily construct a customized editor framework by…
• README: An advanced and powerful inpainting algorithm named PowerPaint is released in our repository.

Reviewed at: 2026-07-16T12:56:25Z