Skip to content

v1.3.0 Release

Choose a tag to compare

View all tags
@peterchei peterchei released this 14 Oct 21:14
· 48 commits to master since this release
v1.3.0
96c051e

Full Changelog: v1.2.0...v1.3.0

PhotoSift v1.3.0 Release Notes

Release Date: October 14, 2025

🎉 What's New in v1.3.0

🌍 Unicode Path Support (Critical Fix)

The Problem: Users with Chinese, Japanese, Korean, or other non-English characters in their folder paths experienced a critical bug where ALL images were incorrectly marked as 100% duplicates, even when they were completely different.

The Solution: We completely rewrote the image loading system to properly handle Unicode file paths. PhotoSift now works seamlessly with folder and file names in ANY language!

Who Benefits:

  • International users with native language folder names
  • Anyone with special characters in file paths (accents, symbols, etc.)
  • Users who organize photos with non-English naming conventions

🎚️ Adjustable Similarity Threshold

Take control of duplicate detection with our new Similarity Threshold Slider (80%-99%):

  • 98-99%: Find only identical images (pixel-perfect matches)
  • 95-97%: Near duplicates (default - recommended for most users)
  • 90-94%: Similar images (different angles, crops, edits)
  • 80-89%: Loosely related images (experimental)

Benefits:

  • Fine-tune detection sensitivity for your specific needs
  • Fast re-grouping: Change threshold and see results instantly (10-20x faster than re-scanning)
  • Quality guide helps you choose the right threshold
  • Real-time percentage display

⚡ Improved Scanning Workflow

Start Scan Button: We separated folder selection from scanning for better control:

  1. Select Folder → See image count
  2. Adjust Threshold → Set your preferred sensitivity
  3. Start Scan → Begin analysis when ready

Auto-Select Results: After scanning or re-grouping, all duplicate groups are automatically selected and displayed - no more manual clicking to see results!

🐛 Bug Fixes

Critical

  • Fixed Unicode path handling: Images in folders with Chinese/Japanese/Korean characters now process correctly
  • Resolved false duplicate detection: 100% similarity on different images is completely fixed

Minor

  • Improved error handling for corrupted image files
  • Better memory management during large batch operations

🚀 Performance Improvements

  • 10-20x faster re-grouping: Change threshold without re-processing images
  • Optimized image loading: PIL-based loading with LANCZOS resampling for quality
  • Efficient caching: Embeddings stored in memory for instant threshold adjustments

📊 Technical Details

Changed

  • Image loading now uses PIL (Pillow) instead of OpenCV for Unicode support
  • LANCZOS resampling for high-quality image resizing
  • Improved embedding extraction workflow

Maintained

  • All 33 regression tests pass ✅
  • Backward compatible with previous versions
  • No new dependencies required
  • All existing features work exactly as before

📥 Installation

New Users

Download and run the installer:

PhotoSift-1.3.0-Setup.exe

Upgrading from v1.2.0

Simply run the new installer - it will automatically upgrade your installation while preserving:

  • Your settings
  • Trash folders
  • Previous classifications

Important Note for Upgrading

If you previously had issues with folders containing Unicode characters:

  1. After upgrading, re-scan those folders
  2. You'll now see accurate duplicate detection
  3. Previous false positives will be resolved

📚 Documentation

New documentation added in this release:

  • CHANGELOG.md - Complete version history
  • UNICODE_PATH_FIX.md - Detailed technical explanation of the Unicode fix
  • THRESHOLD_ENHANCEMENT.md - Guide to using the threshold feature
  • START_SCAN_BUTTON_ENHANCEMENT.md - Workflow improvements documentation
  • AUTO_SELECT_GROUPS_FEATURE.md - Auto-select implementation details

🧪 Testing

This release includes comprehensive testing:

  • ✅ 33 regression tests (all passing)
  • ✅ Unicode path testing with Chinese characters
  • ✅ Similarity threshold validation
  • ✅ Memory leak testing
  • ✅ Large dataset processing (1000+ images)

🔄 Upgrade Path

From v1.2.0 → v1.3.0: Direct upgrade (recommended)
From v1.1.0 → v1.3.0: Direct upgrade (tested)
From v1.0.0 → v1.3.0: Direct upgrade (tested)

🙏 Acknowledgments

Special thanks to users who reported the Unicode path issue and provided test cases with Chinese folder names. This feedback was instrumental in identifying and resolving this critical bug.

🐛 Known Issues

None identified in this release. If you encounter any issues, please report them on our GitHub issues page.

📞 Support

📅 What's Next?

We're already planning v1.4.0 with:

  • Video file support
  • Batch export functionality
  • Custom classification categories
  • Cloud storage integration

Download: PhotoSift v1.3.0 Release

Full Changelog: v1.2.0...v1.3.0

Assets 4
Loading