Changelog

0.2.0 - 2026-09-02

Moved pixel and perspective processing into cancellable worker jobs with progress, stale-result suppression and a bounded main-thread fallback.
Added precise manual corner fields, OCR word geometry overlays, hOCR/TSV export and combined multipage text export.
Declared image/PDF/OCR input, export and optional worker/camera capability profiles for Toolbox handoff.

Add bounded multi-image and camera-picker ingestion with real decoded-pixel checks
Add draggable four-corner projective correction and keyboard-accessible handles
Add rotation, manual and estimated deskew, brightness, contrast, grayscale, Otsu threshold and manual threshold processing
Add retained debounced previews and bounded high-resolution PNG/JPEG exports
Add drag-to-reorder page assembly and A4, Letter, or source-sized PDF output
Add opt-in local English OCR using bundled same-origin Tesseract.js runtime and model
Add editable OCR text export, explicit confidence and memory/accuracy disclosures
Add Toolbox shell integration, nested-path offline PWA support, documentation, deterministic packaging, and Chromium/Firefox coverage