From archive to export.
| WORKSPACE | WHAT YOU CAN DO |
|---|---|
| Collection | Organize folders and collections, rate photographs, filter an archive and work with virtual copies. |
| Develop | Process RAW files, adjust tone and color, work with curves and scopes, correct geometry and retouch. |
| Masks | Combine automatic selections with brushes, gradients, color ranges and depth-based selections. |
| Train & styles | Learn from previous edits, build style packs and apply them to new photographs. |
| Export | Render finished files, choose output settings and apply watermarks. JPEG, PNG, WebP and TIFF encoders are included; AVIF availability depends on the build. |
Bring your Lightroom work.
RedLnx reads Lightroom Classic catalogs and XMP sidecars. Catalog organization and supported editing settings can become the starting point for your work in RedLnx, and a source for learning your editing style.
This is a translation between different rendering engines. Supported settings can be carried across, but identical slider values do not guarantee identical pixels. Proprietary profiles, masks and unsupported features may need review.
Lightroom’s cloud library is a separate migration case. Don’t assume it has the same import path as a Classic .lrcat file.
Train on your work.
Training can use a Lightroom catalog, XMP folders or selected sources in the RedLnx library, including your native RedLnx edits.
Finished JPEGs provide another entry point: RedLnx extracts a look from them. That is different from learning the original per-photo adjustments. You can also edit normally without training anything.
Processing stays local.
The desktop application uses Tauri and a Rust workspace, with Svelte for its interface. RAW processing, rendering, style training and model inference run on your hardware.
GPU support depends on the operating system, hardware, runtime and operation. CPU processing is also available. The release’s hardware requirements will accompany its download.
There is no photo upload service or telemetry. Updates and runtime downloads are separate from the editing workflow and can use an internet connection.
Built on open-source work.
darktable’s algorithms underpin several Rust ports and adaptations in the render engine. Immich was an important reference for the library and local machine learning, particularly its use of CLIP. CLIP has its own upstream authors; it is not an Immich-created model.
The project’s NOTICE file records the third-party code, models, data and algorithmic credits. RedLnx is licensed under AGPL-3.0.