This release is the result of a one-week sprint focused on making the core of RapidRAW faster and lighter on RAM. I rewrote large parts of the image pipeline, so thumbnails, previews, exports and batch actions are now much faster and use a lot less memory. On top of that, this update brings a full Immich integration with edits synced across all your computers, GPU acceleration for AI models, smart AI masks that adapt to every photo, new sky and foreground models, AI Text and Face masks, a rewritten dehaze and noise reduction, more accurate RAW colors and a new job queue.
This release is all about speed. I went through the whole app to find out where time and memory were wasted and rewrote large parts of the image pipeline:
RapidRAW can now connect directly to your Immich server. A new Immich section in the library lets you browse your albums and timeline, and edit any photo just like a local file. Images show Immich's preview as a thumbnail and the original is only downloaded when you open or export it, so you don't need to sync your whole library to your computer.
The best part: your edits are stored together with the photo on your Immich server. This means your edits look the same on every computer and they are kept even if you clear your cache. If you edit the same photo on two computers, nothing gets lost: your local edits stay and the other version is kept as a virtual copy.
What you can do:
Immich 2.5 or newer is required, as it is the first version that lets apps store their own data with a photo. Huge thanks to @cl1x for building this whole integration! Comparing your edit with the original is now much more flexible. Next to the full original view, you can now choose split left/right, split top/bottom and side by side. Just right-click the compare button to choose a mode, or press B to toggle. In split mode, you can drag the divider to any position, and your zoom and position are kept when switching modes, so you can easily compare small details. Until now, when you copied a mask or saved it in a preset, AI masks kept the exact mask of the original photo. A pasted sky mask still showed the sky of the first image, which made AI masks pretty useless for batch editing.
Now, AI masks adapt to every photo. When you paste adjustments or apply a preset that contains AI masks, RapidRAW automatically generates them again for each image in the background: a sky mask finds the sky, a subject mask finds the subject and a face mask finds the face in every single photo. The same works for depth & normal maps and AI denoise. Lens Blur, Fog, Relight and AI Denoise are also new groups in the copy & paste dialog.
This makes it super easy to edit a whole shoot with local adjustments, without opening every image one by one.
Everything that runs in the background now shows up in one place. The new job queue replaces the old AI queue and covers exports, imports, file transfers, AI model downloads, AI tagging and all other AI work. You can follow the progress of each job, cancel running exports, and jump to your files with one click when a job is done. AI processing can be paused, resumed or retried, and pending AI work continues where it stopped after a restart. In v1.6.6, I promised GPU acceleration and non-destructive batch denoising for AI Denoise. Both are now here. AI Denoise runs on your GPU on Windows and macOS, and you can simply paste your denoise settings onto as many photos as you like. The job queue denoises them in the background, without creating a single extra file.
The old denoise modal, which saved a separate TIFF file, is now completely gone. For photographers who use AI-Free Mode, the Details panel now offers a non-destructive Classic Denoise based on BM3D instead.
The Data Management section in the settings has been redesigned into a clear Storage page. You can now see exactly how much space RapidRAW uses on your device, split into thumbnails, metadata, AI models, AI denoise results, depth & normal maps, Immich downloads and logs. Each one can be cleared on its own, and you can set size limits for thumbnails, AI denoise, depth & normal maps and Immich downloads. Once a limit is reached, the least recently used files are removed automatically. Tools for clearing tags and edits from your library are also on this page. The old U2Net models for sky and foreground masks have been replaced with EfficientSAM3 for the sky and BiRefNet for the foreground. The masks are now much more accurate, follow fine details like hair, branches and thin structures, and have cleaner edges. AI models are now also downloaded only when you first use a feature, so you don't need to download everything at once.
There are two new AI mask types:
Simply type what you want to select, for example "person", "car" or "water", and RapidRAW creates a mask for it. This makes it super fast to select objects that are hard to paint by hand. The new face mask finds the faces in your photo and lets you pick exactly which parts you want to edit on a small face illustration: face skin, body skin, hair, eyebrows, eye whites, iris & pupil, nose, lips and teeth. This is perfect for portrait retouching, like whitening teeth, brightening eyes or smoothing skin, without painting a single stroke. The built-in AI features now run on your graphics card: denoise, masks, depth & normal maps and AI tagging. On Windows this uses DirectML with your high-performance GPU, and on macOS it uses CoreML. If your GPU is not supported, RapidRAW falls back to the CPU automatically, and you can turn GPU acceleration off in the settings if you ever run into problems. Auto adjust now comes with three strength levels: Subtle, Normal and Strong. Just right-click the Auto button in the editor, or pick a level from the right-click menu in the library to apply it to many photos at once. Auto adjust also no longer touches the Brightness slider and only sets the exposure, so your results are cleaner and easier to fine-tune. Highlight recovery got another upgrade. RapidRAW now works directly on the RAW data and finds the real clipping point of each color channel in your photo. Areas where all channels are blown out are rebuilt with color from the surrounding pixels, so clipped areas keep their natural color, while bright light sources like lamps and the sun stay neutral instead of turning pink or yellow.
Comparison: RAW colors are now much more accurate, especially under artificial light:
I rewrote the manual noise reduction and sharpening in the main shader, together with the color noise reduction and sharpening that run when a RAW file is first loaded. Noise is now removed more cleanly while fine details stay crisp, and curved edges no longer show jagged artifacts. Dehaze has been completely rewritten. It now estimates the real color and brightness of the haze in your photo and removes it with an edge-aware filter, so distant mountains, cityscapes and foggy scenes get their contrast and color back without dark halos around edges. The haze is removed evenly across the frame, and bright areas roll off smoothly instead of getting crushed. Pulling the slider to the left still adds haze for a softer, dreamy look.
Negative conversion is now a normal non-destructive tool in the Details panel. The old modal is gone. You can adjust the red, green and blue balance, print exposure and print grade at any time without creating a new file, and convert many negatives at once from the right-click menu. This release improves several core parts of the image pipeline, so some of your existing edits may look slightly different after updating:
New Features
Performance Improvements & Lower RAM Usage
Immich Integration
New Before / After Compare Modes
Smart AI Masks for Copy & Paste and Presets
Job Queue
Non-Destructive Batch Denoise & Classic Denoise
Data Management & Storage
New AI Sky & Foreground Models
Sky comparison
Foreground comparison
New Mask Types: AI Text Mask & Face Segmentation
AI Text Mask
Face Segmentation
GPU Acceleration for AI Models
Auto Adjust Levels
Color-Preserving Highlight Reconstruction
More Accurate RAW Colors
Rewritten Noise Reduction & Sharpening
Rewritten Dehaze
Non-Destructive Negative Conversion
Core Improvements
Note for Upgrading Users
What to Download
| OS | Architecture | Format | Download Link |
|---|---|---|---|
| Windows | x64 | .exe
| RapidRAW for Windows |
| macOS (Apple Silicon) | ARM64 | .dmg
| RapidRAW for macOS ARM |
| macOS (Intel) | x64 | .dmg
| RapidRAW for macOS x64 |
| Ubuntu 22.04 | x86_64 | .deb
| RapidRAW for Ubuntu 22.04 |
| Ubuntu 24.04 | x86_64 | .deb
| RapidRAW for Ubuntu 24.04 |
| Android | ARM64 | .apk
| RapidRAW for Android |
Download Tethering Builds & Setup Guide (macOS & Linux)
1. Install libgphoto2 on your machine
RapidRAW uses libgphoto2 to communicate directly with connected cameras over USB. Because the tethering binary links dynamically against this library to stream live view and control camera settings, the driver must be present on your system for the app to detect and control your camera.
-
macOS (via Homebrew):
brew install libgphoto2 pkg-config
-
Ubuntu / Debian:
sudo apt update && sudo apt install -y libgphoto2-dev pkg-config -
Arch Linux:
sudo pacman -S libgphoto2 pkgconf
-
Fedora:
sudo dnf install libgphoto2-devel pkgconf-pkg-config
2. Download Tethering Build
| OS | Architecture | Format | Download Link |
|---|---|---|---|
| macOS (Apple Silicon) | ARM64 | .dmg
| RapidRAW for macOS ARM (Tethering) |
| macOS (Intel) | x64 | .dmg
| RapidRAW for macOS x64 (Tethering) |
| Ubuntu 22.04 | x86_64 | .deb
| RapidRAW for Ubuntu 22.04 (Tethering) |
| Ubuntu 24.04 | x86_64 | .deb
| RapidRAW for Ubuntu 24.04 (Tethering) |
(Note: Windows and Android are not supported for camera tethering due to OS driver limitations).
For other platforms and formats (ARM builds, RPM, AppImage, etc.), check the full asset list below.
Running on Windows and macOS
RapidRAW is not yet code-signed, so Windows and macOS may show warnings when launching the app.
-
Windows: When launching the app, you may see a Windows Defender SmartScreen warning.
- Click "More info" → "Run anyway" to proceed.
-
macOS: You'll need to remove the quarantine flag after installation, otherwise macOS may report the app as corrupted:
xattr -dr com.apple.quarantine /Applications/RapidRAW.app