#65 · Primary category: Image Generation

mmagic

aigc computer-vision deep-learning diffusion diffusion-models generative-adversarial-network generative-ai image-editing image-generation image-processing image-synthesis inpainting matting pytorch super-resolution text2image video-frame-interpolation video-interpolation video-super-resolution

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.

Project last updated:08/06/24

GitHub Stars

7.5K

Forks

1.1K

Contributors

128

License

Apache-2.0

Why we included this project

MMagic is OpenMMLab's single PyTorch toolbox for generative media, covering text-to-image work, restoration, and editing in one place instead of a patchwork of separate codebases. Teams can run Stable Diffusion or ControlNet pipelines and fine-tune them with DreamBooth or LoRA, and the bundled model zoo ships pre-trained weights for most of the common tasks. The same framework handles the cleanup side: super-resolution, inpainting, matting, colorization, and video interpolation, which helps when a production pipeline needs both synthesis and repair. Because the OpenMMLab conventions keep evaluation loops and metrics consistent across tasks, with FID for generation and SSIM/PSNR for reconstruction, researchers and applied engineers get a common set of habits instead of relearning the details per project. The real cost is learning the MMEngine and MMCV idioms, and deciding whether the toolbox's breadth beats a leaner, task-specific library.

Articles for this project

No articles for this project yet.

To suggest a topic or contribute an article, contact us.

Related projects in this category