DeJPEG

An app for removing noise and compression artifacts from photos.

Features:

- Batch processing - Supports most image formats - Before/after view - Fully offline

This is not a "super resolution AI upscaler," but a simple, non-destructive method for cleaning up and restoring images.

Models (required):

- FBCNN (JPEG compression) - SCUNet (Grain/Noise)

More info (also in the app’s FAQ): https://github.com/jeeneo/dejpeg/wiki/model-types

You can also run other experimental models. More info here: https://github.com/jeeneo/dejpeg-experimental

Limitations:

- Processed locally; minimum 4 GB RAM and 4 threads recommended - Limited to images below 8000 pixels; images above 4000 px may affect performance

See the wiki for more information: https://github.com/jeeneo/dejpeg/wiki

Note:

Please see this issue before downloading: https://github.com/jeeneo/dejpeg/issues/26

Disclaimer:

DeJPEG is not affiliated with or related to Topaz DEJPEG or any other similarly named software/project.

This app originated from personal use and was modified for public release.

License

This is a GUI for a select amount of 1x ONNX processing models, used under their respective licenses (Apache 2.0).

Versions prior to 4.0 were licensed under GPLv3. Starting with version 4.0 onward, this project is licensed under AGPLv3-or-later.