Codex (gpt-5.3-codex and gpt-6.1-codex [medium]) wrote most of the below so please don't trust it too much!

Picasa Redux

Picasa Redux is a Java/Swing desktop photo browser and lightweight editor for local folders.

It provides a split view with:

• an album tree with folder/image browsing, filters, and face groups
• a thumbnail gallery for folder browsing
• an image editor for quick transforms (rotate, mirror, flip, save copy, metadata view)

Features

• Select any local folder as an album root
• Load albums in the background with a loading indicator and Cancel button
• Browse nested folders in a tree with image counts and sizes that reflect the active filter
• Check for duplicate images using file size and CRC32 hashes; browse same-size candidates with the Duplicates filter
• Filter the tree by portrait, landscape, or square image dimensions
• Run local OpenCV face detection on demand, with progress and a Stop button
• After detection, browse images with faces, without faces, or grouped by similar faces
• View folder images in a resizable thumbnail grid
• Change thumbnail ratio presets (1:true, 1:1, 5:4, 4:3, 3:2, 16:9) and size
• Open single-image editor with:
  • fit mode toggle
  • rotate clockwise / anticlockwise
  • mirror / flip
  • save edited copy alongside original
  • toggleable metadata panel (dimensions, MP, size, format, orientation, modified date, aspect ratio, colour model, transparency, pixel type, and current transforms)
  • EXIF details when available (date taken, camera, lens, focal length, aperture, shutter speed, ISO, GPS, orientation, colour space, and ICC profile)
• Right-click a tree item, or press Space / Enter on the selected item, to open it via the OS default app

Album Filters And Faces

Choose Select album folder to scan a local folder and its subfolders. Select a folder in the tree to show its immediate images in the gallery, or select an image to open the editor.

The All Images, Duplicates, Portraits, Landscapes, and Square buttons switch the tree view. Orientation filters use stored pixel dimensions; image dimensions are read through an ImageReader without decoding the full image.

Click Faces... to start face detection and grouping in the background. Once it completes, Faces, Without Faces, and Collated Faces become available. Collated Faces replaces the folder hierarchy with numbered groups such as Face 1; selecting a group shows its images in the gallery. Detection and grouping run locally using OpenCV.

Tech Stack

• Java 25
• Swing (desktop UI)
• Maven
• JUnit 5
• OpenCV via JavaCPP (face detection and similarity grouping)
• metadata-extractor (EXIF and other embedded metadata)

Project Layout

src/main/java/com/picasaredux/
  Main.java                     # public application entry point
  model/
    FileTree.java               # folder/image tree model
    DirectoryInTree.java        # folder scanning and duplicate/face bookkeeping
    ImageFileInTree.java        # image dimensions, thumbnails, hashes, face results
    ImageMetadata.java          # file and pixel metadata
    ExifData.java               # embedded metadata extraction
  service/
    FaceDetector.java           # face detection interface
    OpenCvFaceDetector.java     # OpenCV frontal-face detector
    FaceCollator.java           # face grouping interface
    OpenCvFaceCollator.java     # face similarity clustering
  view/
    App.java                    # main UI shell
    AlbumSelector.java          # folder picker
    Album.java                  # Swing tree, filters, and face groups
    VerticalSlider.java         # split pane and background workers
    ImageGallery.java           # grid view container
    ImageGrid.java              # thumbnail rendering + click selection
    GridResizer.java            # thumbnail size/ratio controls
    ImageEditor.java            # editor toolbar + metadata display
    EditableImage.java          # image transforms + save copy
    ...
src/test/java/com/picasaredux/   # entry point, model, album filter, and face grouping tests

Requirements

• JDK 25 installed
• Maven 3.8+ installed

The current Maven profiles include OpenCV native dependencies for macOS (Apple Silicon and Intel) and Windows x64. Other platforms need corresponding native dependencies and packaging configuration.

Build And Test

mvn test
mvn package

The test suite covers tree construction, duplicate hashing, thumbnail scaling, metadata/EXIF parsing, album filters, and face grouping.

Native Desktop App

mvn package also creates a double-clickable native application image for the host OS, including a Java runtime and the application's dependencies:

• macOS: target/app-image/Picasa Redux.app
• Windows: target\app-image\Picasa Redux\Picasa Redux.exe

Build each artifact on its own target OS; jpackage does not cross-package macOS and Windows launchers. The application image uses the platform's default icon until a branded .icns / .ico asset is supplied. Its app directory contains one shaded picasaredux-1.0-SNAPSHOT.jar, rather than separate dependency JARs.

Run

After mvn package, open the native application image above. It includes its own Java runtime, so users of the packaged app do not need a separate JDK.

You can also run the self-contained JAR with JDK 25 on the platform it was built for:

java --enable-native-access=ALL-UNNAMED -jar target/jpackage-input/picasaredux-1.0-SNAPSHOT.jar

To run from an IDE:

1. Import as a Maven project.
2. Run com.picasaredux.Main.

Current Limitations

• Duplicate detection uses file size plus CRC32 (fast, but not cryptographically strong)
• The Duplicates filter currently includes all images sharing a file size, so it can show candidates whose hashes differ
• Face detection and grouping are experimental: faces may be missed or falsely detected, and similar crops may be grouped incorrectly
• Face grouping uses the largest detected face in each image and assigns the image to one group; it does not identify every person in group photos
• Portrait/landscape/square filters use pixel dimensions without applying EXIF orientation
• Save action writes copies as PNG and appends an action summary to the filename
• No CLI interface; desktop-only workflow

Development Notes

• Generated and local-only folders/files are gitignored where appropriate (target/, .idea/, .DS_Store)
• Example image assets are under stock_images/

License

This is a research project. Do whatever you like with the code I’ve written here - copy it, change it, share it, or sell it. No credit required, and no promises that it works. Other people’s libraries and images have their own terms. This project isn’t affiliated with Google, and I claim no rights to the Picasa name.