Samsung Galaxy phones (and others) can have GPS EXIF tags with null
coordinate values when location is disabled. The server returned
_latitude: null, and the client guard used !== undefined which passed
for null, causing null.toFixed() to crash React with no ErrorBoundary.
- Server: validate GPS array values are actual numbers before computing
- Client: use != null guard (catches both null and undefined)
- App: add ErrorBoundary to prevent white screens from any future crash
Integrates feature/multi-image-ux branch with 20 commits including:
- Multi-image viewer with arrow navigation and filmstrip thumbnails
- Batch processing across all tool settings components
- File store rewrite with FileEntry model for multi-image support
- Resize settings redesigned with tab-based UI (presets, custom, scale)
- Side-by-side comparison for resize results
- Per-file metadata caching in strip-metadata
- Client-side ZIP extraction via fflate
- SSE progress correlation via clientJobId
Consolidate all access to localhost:1349 — Vite dev server serves on 1349
and proxies API calls to an internal dev port (13490). Production API
defaults to 1349. Also includes strip-metadata improvements, UI component
updates, and compress operation fixes.
Replace technical AI model names with user-friendly controls:
- Subject type buttons (People/Products/General) with icons
- Quality selector (Fast/Balanced/Best) with time estimates
- Passport/ID photo checkbox for portrait-optimized processing
- Models selected automatically based on user's choices
Upload occupies 0-15% of the bar, server processing 15-100%.
SSE progress from the server is scaled into the 15-100 range.
The bar now moves smoothly from start to finish.
The Zustand setError() setter has a side effect: it sets
processing=false. When called after setProcessing(true), it
immediately killed the processing state, so the ProgressCard
never rendered. Fix: call setError(null) before setProcessing(true).
The /api/v1/jobs/ endpoint was blocked by auth middleware.
EventSource doesn't support custom headers so auth tokens can't be
sent. The jobId is a random UUID (unguessable capability token), same
security model as the download endpoint.
Replace AIProgressBar and inline progress indicators with the new
ProgressCard component across all 5 AI tool settings. The three
useToolProcessor-based tools (remove-bg, blur-faces, upscale) now
destructure `progress` from the hook. The two custom-fetch tools
(erase-object, ocr) gain inline XHR upload tracking + SSE processing
progress with elapsed timer.
Replace Loader2 spinner buttons with ProgressCard in all 13 non-AI tool
settings components that use useToolProcessor. Each now shows upload
progress, elapsed time, and processing phase during operations.
Extract clientJobId from multipart form data in all 5 AI route handlers
(remove-background, upscale, blur-faces, erase-object, ocr) and forward
progress callbacks to the SSE system via updateSingleFileProgress.
- Set up semantic-release with zero-touch CI pipeline on push to main
- Add version sync script to keep all package.json files and APP_VERSION
constant in sync automatically
- Consolidate Docker publishing into single tag-triggered workflow that
pushes to both Docker Hub and ghcr.io with semver tags
- Add help dialog with keyboard shortcuts, getting started guide, and
resource links
- Sync all versions to 0.2.1 to match Docker Hub latest
BiRefNet-Lite times out on first load (~60s+ for 973MB model).
U2-Net works in 2 seconds. Users can still select BiRefNet for
higher quality when they're willing to wait. Added timing hints
in model descriptions and increased timeout for BiRefNet models.
Rewrites all documentation with accurate project details (Fastify, port
1349, single-container Docker, all 33+ tools, full database schema).
Adds getting started guide and configuration reference. Updates help and
settings dialogs to link to the docs site.
- Switch default from birefnet-general (973MB, 4min) to
birefnet-general-lite (faster, still SOTA quality)
- Fix Python script stdout pollution — progress messages now go to
stderr so the JSON result parser doesn't break
- Pre-download birefnet-general-lite in Docker build
- Switch default model from U2-Net to BiRefNet (state-of-the-art)
- Add 6 model options: BiRefNet, BiRefNet Lite, BiRefNet Portrait,
BRIA RMBG, IS-Net, U2-Net
- Add animated progress bar with stage indicators (loading model,
analyzing, removing, refining edges) and elapsed timer
- Add intuitive background color presets (Transparent, White, Black,
Red, Green, Blue) as clickable buttons + custom color picker
- Handle background color compositing in Python (PIL alpha composite)
- Add checkerboard pattern to before/after slider for transparency
- Pre-bake BiRefNet model (973MB) in Docker image for instant use
Siddharth Kumar Sah