- Home
- Image Upscaling Software
- Real-ESRGAN
Real-ESRGAN
What it is
Real-ESRGAN is a free, open-source project for practical image and video restoration. It provides models for general images, anime images, and animation video, with JPG, PNG, and WEBP inputs and matching image outputs. Portable inference supports 2, 3, or 4 scale ratios, while the Python implementation offers arbitrary output scaling. Images with alpha channels, grayscale, and 16-bit data are supported, and GFPGAN integration adds face enhancement. Portable NCNN executables are available for Windows, Linux, and macOS, including required binaries and models without requiring CUDA or a PyTorch environment. Those portable builds do not expose every Python feature, including outscale. Python use requires Python 3.7 or newer and PyTorch 1.7 or newer. Online inference is available through a Tencent ARC demo and Colab demos. Training uses pure synthetic data, and released code supports fine-tuning on personal or paired data. The project uses the BSD-3-Clause license.
For and against
For
- Free and released under BSD-3-Clause
- Portable builds include binaries and models
- Supports alpha, grayscale, and 16-bit images
- Models cover general and anime content
- Training code supports custom fine-tuning
Against
- Portable builds omit some Python functions
- Python requires Python 3.7+ and PyTorch 1.7+
- Portable scaling is limited to 2, 3, or 4
The verdict
Real-ESRGAN offers broad restoration models with both portable and Python routes. The portable option is easier to deploy, while Python provides more controls but adds environment requirements.
Who it suits
Real-ESRGAN suits developers and creators seeking free restoration models, portable executables, or Python workflows. It also fits users working with anime imagery, video models, or custom fine-tuning data.
On the record
- Batch processing
- Yesgithub.com · read 20 Sep 2026
- Custom training
- The project supports finetuning on users' own data or paired data.github.com · read 27 Sep 2026
- Dependencies
- The Python installation instructions specify Python 3.7 or later and PyTorch 1.7 or later.github.com · read 2 Oct 2026
- Desktop app
- Yesgithub.com · read 20 Sep 2026
- Face enhancement
- Its inference code can use GFPGAN to enhance faces.github.com · read 2 Oct 2026
- Image enhancement
- The project provides pretrained models for general images and anime images, including a smaller anime model.github.com · read 2 Oct 2026
- Image options
- The inference code supports tiling, alpha-channel images, grayscale images, and 16-bit images.github.com · read 2 Oct 2026
- Image support
- The inference code supports tile options, alpha-channel images, grayscale images, and 16-bit images.github.com · read 1 Oct 2026
- Input formats
- The executable accepts JPG, PNG, and WebP images.github.com · read 27 Sep 2026
- Integrations
- The README says Real-ESRGAN was integrated into Hugging Face Spaces with Gradio and links to a web demo.github.com · read 2 Oct 2026
- License
- The repository includes a BSD 3-Clause license that permits redistribution and use in source and binary forms subject to its conditions.github.com · read 2 Oct 2026
- Licensing
- The repository is distributed under the BSD 3-Clause License, which permits redistribution and use in source and binary forms with conditions.github.com · read 1 Oct 2026
- Limitations
- The README says the portable NCNN executable lacks some Python-script features, such as arbitrary outscale, and tile processing can cause block inconsistencies.github.com · read 2 Oct 2026
- Maximum scale
- customgithub.com · read 20 Sep 2026
- Model options
- Provided models include general, anime-image, anime-video, and Real-ESRNet variants.github.com · read 27 Sep 2026
- Models
- The project provides models for general images, anime images, and animation video.github.com · read 1 Oct 2026
- Online demos
- Online inference is available through the Tencent ARC Demo and two Colab demos.github.com · read 1 Oct 2026
- Online inference
- Online inference is available through an ARC Demo and Colab demos.github.com · read 27 Sep 2026
