Omera v0.4.3

Phase 6 T4 Advanced Transform Controls, Resilient In-App Updater & Windows NSIS Integration

Version 0.4.3 delivers Phase 6 T4 (Advanced Image Transcoding Controls #118, #184), fortifies the in-app GitHub updater against API rate limits, deepens Windows installer integration with Start Menu folder indexing and process safety hooks, introduces ModelScope ultra-fast domestic model mirror downloads for WD14 AI Auto-Tagger, and unifies adaptive UI contrast across 7 languages:

⚡ Phase 6 T4: Advanced Image Transform Controls (#118, #184)

• Target Size Search (Byte Budget Optimization): Iterative search algorithm (bounded to 4 cycles) that converges image quality and dimension downscaling toward an exact target file size (e.g. 1MB/2MB sharing targets) without overshoot or degenerative cycles.
• Scale Percentage Downscaling: Aspect-ratio-preserving Lanczos3 scaling supporting 75%, 50%, and 25% downsampling constraints across both managed vault imports and bulk gallery optimizations.
• Pixel Multiple Alignment: Enforces 8x (codec DCT block friendly) or 16x (AI tensor/latent friendly) dimension alignment to eliminate padding artifacts and optimize compression efficiency.
• One-Click Reusable Presets: Built-in presets for Web Fast Sharing (WebP 80% / 1920px), Archive Grade (AVIF 65%), Target 1MB, and Lossless WebP.
• Adaptive UI Integration: Seamlessly integrated into BatchTransformModal.vue and ImportTransformModal.vue with real-time responsive form controls and estimated savings feedback.

🔄 Resilient In-App Updater & GitHub Fallback (#184)

• Rate-Limit Resilience (GitHub 403 Fallback): Added custom Omera-Updater user-agent and an automated fallback engine in src/utils/updater.ts that gracefully recovers from GitHub REST API anonymous rate limits (HTTP 403) by querying releases via HTML 302 tag redirection.
• Canonical Multi-Platform Asset Matching: Robust asset matcher for Windows (.exe / .msi / legacy installers), macOS (.dmg universal / aarch64), and Linux (.AppImage / .deb), ensuring seamless one-click update detection from v0.4.1 and v0.4.2 onward.

🪟 Windows NSIS Installer Customization & Desktop Integration (#184)

• Start Menu Folder Organization: Explicitly sets startMenuFolder: "Omera" in NSIS configuration, registering shortcuts cleanly under %AppData%\Microsoft\Windows\Start Menu\Programs\Omera.
• Optional Desktop Shortcut: Configured NSIS page allowing users to toggle desktop shortcut creation during installation.
• Process Safety Hooks (hooks.nsh): Pre-install termination hook (taskkill /F /IM omera.exe) to prevent file lock errors during updates, and clean uninstall hooks to remove empty parent application directories.
• In-App Desktop Shortcut Creation: Added one-click "Create Desktop Shortcut" button in Settings (SettingsModal.vue) with dynamic Shell COM path resolution for both local and OneDrive redirected desktops.

🤖 WD14 AI Auto-Tagger UX & Mainland China Mirror (ModelScope) (#184)

• ModelScope Domestic Mirror: Added high-speed direct download support for WD14 Tagger models (BerryUIKI/wd-v1-4-convnext-tagger-v2) via ModelScope (modelscope.cn), solving slow or blocked network downloads in mainland China.
• One-Click Auto-Download & Progress Monitoring: Seamless empty-state onboarding card that downloads models with a single click, streaming real-time download progress, downloaded bytes, and speed (MB/s) with full cancellation support.
• Light Theme Contrast & Accessibility: Replaced hardcoded slate/white text in AutoTagModal.vue with semantic CSS variables (var(--color-text-primary), var(--color-text-secondary)), guaranteeing crisp readability across all themes.

🌐 Complete 7-Language Internationalization

• Multi-Language Coverage: Added localized strings across English (en), Simplified Chinese (zh-CN), Traditional Chinese (zh-TW), Japanese (ja), German (de), French (fr), and Spanish (es) for all new transform presets, mirror sources, installer options, and download states.

---

For complete release details and commit history, see CHANGELOG.md.

Omera v0.4.2

Phase 6: Native Image Transcoding, Managed Import Presets & Safe Gallery Optimization

Version 0.4.2 delivers Phase 6 of the Omera roadmap (#118, #182), introducing pure-Rust image transcoding and compression engines, instant export size estimation, configurable vault import optimization presets, non-destructive gallery batch transformation with safe original disposition, and full internationalization across 7 languages:

⚡ Native Image Transcoding & Compression Engines (#118)

• Pure-Rust AVIF & Lossless WebP: Built-in support for AVIF encoding (ravif) and fast lossless WebP transcoding (image / webp) in crates/omera-scan with tuneable quality and speed tradeoffs.
• Format Fallback & Reliability: Graceful fallback strategies during transcoding to ensure zero image corruption and verifiable file outputs.

📊 Export Estimation & Live Preview (#118)

• Instant Size Estimation (estimate_export_file): Added fast single-file simulation command calculating exact estimated byte sizes, space delta, and compression percentage before running full batch exports.
• Export Modal Comparison Card (ExportModal.vue): Interactive visual before/after comparison showing source size, projected output size, and space savings percentage.

📥 Managed Vault Import Presets & OS Drop Handling (#118)

• Import Optimization Modal (ImportTransformModal.vue): Configurable import workflow for managed vaults with resolution downscaling constraints (e.g. max width/height while preserving aspect ratio), target format conversion (WebP, AVIF, JPEG, PNG, Original), AI metadata stripping or preservation, and filename collision handling.
• Direct Sidebar Import & Drop Routing: One-click import button on the sidebar and native OS drag-and-drop routing directly to managed vault folders with automatic preset transformation.

🔄 Gallery Batch Transformation with Safe Disposition (#160, #182)

• Batch Transform Workflow (BatchTransformModal.vue): Gallery multi-select action to transcode and optimize existing library images in bulk with real-time progress tracking, cancelability, and aggregated savings summaries.
• Non-Destructive Original Disposition: Supports three user-chosen policies:
  • Keep: Retains the original file alongside the transcoded file.
  • Archive: Moves the original into a hidden .omera_archive/ subfolder preserving relative file structure.
  • Trash: Safely relocates the original to the OS system Recycle Bin / Trash without permanent deletion.
• Linked Folder Read-Only Protection: External linked library folders are strictly safeguarded against destructive original disposition (trash / archive disabled, forcing non-destructive output).
• Metadata & DB Seamless Sync: Automatically preserves AI generation parameters and updates SQLite database records and thumbnail caches without losing favorites, ratings, or tags.

🌐 Complete 7-Language Internationalization

• Multi-Language Coverage: Added localized strings across English (en), Simplified Chinese (zh-CN), Traditional Chinese (zh-TW), Japanese (ja), German (de), French (fr), and Spanish (es) for all import, export, and batch transformation dialogs.

---

For complete release details and commit history, see CHANGELOG.md.

Omera v0.4.1

Theme Contrast & Design Tokens Unification, Aggregated Facet Counts & High-Load Performance Caching

Version 0.4.1 is a maintenance and performance patch release fixing theme contrast issues across dark and light modes, unifying semantic design tokens, aggregating filter/tag statistics into single-pass database queries, and caching distinct metadata models and samplers to eliminate CPU overhead under heavy library load:

🎨 Theme Design Tokens & Contrast Unification (#177, #178)

• Eliminate Hardcoded Colors: Replaced all hardcoded dark and light hex values across SortBar.vue, TitleBar.vue, SettingsModal.vue, TagModal.vue, AlbumModal.vue, and form controls with semantic CSS custom properties (--color-text-primary, --color-text-secondary, --color-bg-surface, --color-bg-hover, --color-bg-card, --color-border-subtle).
• TitleBar Version Visibility: Resolved muted version subtitle contrast in Light mode, ensuring crisp readability across all operating systems.
• Select & Dropdown Contrast: Fixed native <option> background and text contrast across custom select inputs in light mode to prevent unreadable white-on-white text.
• Automated Theme Regression Test: Added comprehensive theme contrast tests in tests/theme-contrast.test.mjs verifying design token adherence across all components.

⚡ High-Load Facet Aggregation & In-Memory Caching (#180)

• Single-Query Library Summary Counts: Replaced separate queries with a unified SQLite query calculating total files, favorites, and NSFW flags simultaneously (SELECT COUNT(*), COALESCE(SUM(is_favorite), 0), COALESCE(SUM(is_nsfw), 0) FROM files), cutting startup and navigation IPC latency.
• Aggregated Tag Counts: Replaced potential N+1 tag frequency lookups with an indexed GROUP BY tag_id aggregation (get_tag_counts), directly displaying counts on Sidebar and TagModal chips.
• Distinct Models & Samplers Facet Cache: Added thread-safe RwLock in-memory caching for list_distinct_models and list_distinct_samplers, eliminating repetitive full-table JSON parsing across 50,000+ files on repeated drawer opens. Cache is automatically invalidated upon folder scan completion.
• Sidebar & Tag Badges: Added live count badges for Favorites and Sensitive navigation items, as well as file count indicators on tag chips.

---

For complete release details and commit history, see CHANGELOG.md.

Omera v0.4.0

Omera Identity Transition, Single-Instance Lifecycle, Hardened Security & Verified Transformations

Version 0.4.0 is a milestone pre-release completing the official project identity transition to Omera (com.berryuiki.omera), enforcing single-instance desktop lifecycle, restricting WebView CSP and local protocol boundaries, introducing verified image export and managed vault transformation engines, auto-detecting adult content with user override preservation, and embedding a comprehensive offline feature documentation drawer:

🌟 Omera Canonical Identity & Migration Bridge

• Canonical App Identity: Fully transitioned binary naming, application metadata, OS bundles, and packaging to Omera (com.berryuiki.omera, repository BerryUIKI/Omera, primary SQLite database omera.db, local preferences prefix omera_).
• Zero-Data-Loss Migration Bridge: Integrated automatic discovery for legacy Berry AI Studio installations (berry.db, berry_* local storage keys, and legacy snapshots), migrating them seamlessly into Omera with verifiable receipts (berry-to-omera-v1).
• Core Crate Restructuring: Reorganized core architectural components under crates/omera-* (omera-domain, omera-metadata, omera-scan, omera-storage, omera-tagger, omera-clip).

🪟 Desktop Lifecycle & Security Hardening

• Single-Instance Enforcement (#124): Integrated tauri-plugin-single-instance = "2", preventing duplicate database locks and taskbar clutter by unminimizing and focusing the active application window upon secondary launch.
• Defense-in-Depth WebView CSP (#107): Enforced strict Content Security Policy denying untrusted remote scripts, inline object embeds, and arbitrary frame nesting, while permitting necessary local asset rendering and verified API endpoints.
• Scoped Local Asset Protocol (#107): Removed wildcard scopes (scope: []) and dynamically authorized local asset protocol access strictly to user-registered library folders and the managed WebP thumbnail cache.

📤 Verified Export & Transcoding Precision (#158)

• Format & Extension Agreement: Synchronized container encoding for ExportFormat::Original with resize or privacy options across JPEG, WebP, and PNG, preventing mislabeled file signatures.
• Strict Privacy Sidecars: Ensured .txt prompt and .json metadata sidecars strictly follow effective sanitization policies (StripAll completely omits sidecars; StripAllAiMetadata scrubs all AI workflow chunks).
• Collision Suffixing: Safe collision detection appending numeric increments (_1, _2) for both directory and ZIP package exports, preventing accidental file overwrites.
• Accurate Outcome Reporting: Counted exported items only after verified completion of image and sidecar disk writes, reporting granular per-file errors.

🔄 Staged Image Transformation & Managed Vaults (#159, #160)

• Managed Import Transformation (T2, #159): Optional image compression and conversion during import into managed vaults, with staged decode verification before database indexing; external source images remain unmodified by default.
• Library Batch Compression & Safe Disposition (T3, #160): Gallery batch action to convert and compress existing managed media assets. Supports explicit user-directed original disposition (Keep, Archive to .omera_archive/, or move to OS Trash without permanent deletion fallback). External linked folders are strictly read-only.

🔞 Rule-Based NSFW Auto-Detection (#128)

• Rule-Based Keyword Analysis: Integrated automated prompt and WD14 tag analysis in omera-metadata to flag sensitive content on initial ingest.
• User Override Absolute Priority: SQLite upsert policies ensure explicit user curation (is_nsfw manual toggle) is permanently preserved across library rescans and folder walks.

📖 Global In-App Help Drawer & Feature Documentation (#137)

• Offline Help & Feature Guide: Lightweight, zero-network documentation reader (HelpGuideDrawer.vue) featuring 11 feature chapters, quick keyword search, and keyboard shortcut matrix.
• Smart Context Deep-Linking: Auto-detects active workspace context (burst stacks, video media, settings, models, CLIP, search) to highlight relevant documentation.
• Global Keybinding: Summonable from any screen via F1 or Ctrl+Shift+H / Cmd+Shift+H, and accessible via the top MenuBar across all 7 supported UI locales.

⚡ Asynchronous Command Concurrency & Schema Evolution (#102, #135)

• Asynchronous Command Scheduling: Offloaded blocking filesystem operations, network requests, and ONNX inference to tauri::async_runtime::spawn_blocking, keeping database lock durations minimal.
• Large-Library Storage Benchmarks: Verified keyset cursor indexing (idx_files_cursor_mtime, idx_files_cursor_size, idx_files_cursor_rating) and documented query plan metrics across 1k, 10k, and 50k item libraries in docs/benchmarks/LARGE_LIBRARY_BENCHMARK.md.

---

For complete release details and commit history, see CHANGELOG.md.

Berry AI Studio v0.3.0

Cloud Sync & Export Utilities, Keyset Deep Pagination & Multi-Database Team Studio (Milestones 12, 13, 14)

Version 0.3.0 is a major milestone release delivering batch format conversion and privacy stripping, standalone HTML showcase export, S3 and WebDAV snapshot backup and incremental mirroring, keyset cursor deep pagination for 500,000+ assets, directory fingerprint benchmarking, and enterprise multi-database collaboration support:

☁️ Cloud Sync & Export Utilities (Milestone 12)

• Batch Transcoding & Packaging (12.1): Multi-threaded format conversion to WebP, JPEG, and PNG using Rayon; pure-Rust ZIP archive packaging and directory export.
• 4-Tier Privacy Sanitization (12.1): Configurable metadata stripping (KeepAll, StripPromptOnly, StripAllAiMetadata, StripAll pixel-only sanitization) removing embedded prompts, generation parameters, and sensitive EXIF/ICC chunks.
• Filename Pattern Templating & Sidecars (12.1): Flexible export naming with tokens ({name}, {date}, {model}, {id}, {rating}) and optional .txt prompt or .json metadata sidecars.
• Standalone Interactive HTML Showcase (12.2): Zero-dependency, single index.html export containing responsive dark-theme gallery, fullscreen pan/zoom lightbox with keyboard navigation (Esc, ←, →), prompt inspector with one-click copying, and instant keyword filter.
• S3 & WebDAV Snapshot Backup & Restore (12.3): Automated and manual snapshot creation targeting AWS S3 (SigV4), Cloudflare R2, MinIO, Backblaze B2, and WebDAV servers (Nextcloud/Synology). Includes hot SQLite VACUUM INTO, schema verification, pre-restore rollback backup, and live database connection hot-swapping.
• Incremental Remote Asset Mirroring & Delta Sync (12.4): ETag and streaming SHA-256 change detection for bidirectional asset sync, with a token-bucket RateLimiter aggregate bandwidth throttle, atomic cancellation, dry-run simulation, and live animated UI progress.

⚡ Keyset Cursor Deep Pagination & Scalability (Milestone 13)

• Keyset Cursor Deep Pagination: Virtual gallery scrolling directly integrated with keyset cursors (search_files_cursor_page and search_files_by_query_cursor_page), eliminating SQLite OFFSET N scans and decoupling window counting for (O(1)) row traversal (< 1 ms at 40k+ assets, an 84x speedup over offset paging).
• Directory Fingerprint Benchmarks: Automated benchmark suite (crates/berry-scan/benches/directory_fingerprint.rs) evaluating 4 directory traversal strategies, showing parent directory mtime gating delivers a 2.5x speedup locally and 5.6x speedup over network shares (SMB/NFS/WebDAV).

👥 Multi-Database Support & Team Studio (Milestone 14)

• Storage Engine Abstraction: Abstract StorageEngine trait supporting SQLite, MySQL 8.0+, and PostgreSQL 14+ backends.
• Storage Root Mapping: Cross-platform path normalization (storage_roots table and client mount configurations) for multi-user team collaboration across Windows, macOS, and Linux.
• Optimistic Concurrency Control: Row-level version tracking (version column) detecting concurrent modifications with last-write-wins and set-union conflict resolution.
• Real-Time Collaboration Sync Engine: Zero-DevOps change log journal polling engine (CollaborationSyncEngine) with automatic cadence adaptation and cross-client event dispatching.
• Team & Database Settings: Dedicated configuration tab with live database connection testing, latency ping diagnostics, and storage root mount mapping.

---

For complete release details and commit history, see CHANGELOG.md.

Berry AI Studio v0.2.1

Large-Library Performance & Gallery Virtualization (Milestone 13)
Version 0.2.1 delivers comprehensive performance optimizations across gallery rendering, thumbnail caching, database queries, filesystem indexing, and application bundles to ensure fluid navigation with tens of thousands of local images:

⚡ Virtual Gallery & Layout Optimization
Responsive Fixed-Width Columns: Maintained stable card widths during viewport resizing, dynamically adjusting column counts instead of stretching images.
Frame-Coalesced Scrolling: Coalesced scroll events through requestAnimationFrame, limiting reactive updates to at most one per rendered frame.
Per-Column Binary Search: Replaced full dataset filtering in Waterfall view with per-column binary search for visible items.
Filter-Aware Stack Summaries: Kept stack counts, heroes, and expanded members aligned with active text, structured, and semantic search filters.
📦 Incremental Data Delivery & Deferred Payloads
Bounded 400-Item Paging: Replaced full table materialization with bounded 400-item SQLite queries returning exact filtered window counts.
Deferred Raw Metadata: Omitted heavy prompt strings and ComfyUI workflow JSON from gallery page IPC; loaded detailed records on-demand with a 64-entry LRU cache.
Stale Response Invalidation: Discarded in-flight page queries when navigation filters, folders, or sort orders change.
🖼️ Smart Multi-Tier Thumbnail Pipeline
Persistent Thumbnail Manifest (Schema v11): SQLite-backed manifest tracking image IDs, modification times, size tiers, and codecs with rate-limited access tracking.
Configurable 2 GB Cache Budget: Automated LRU eviction maintaining disk usage within the configured budget during generation and background migration.
Rendered-Size Tier Selection: Selected the smallest sufficient thumbnail tier covering rendered card dimensions across Grid, Waterfall, and compact Table rows.
Existing Tier Reuse: Reused existing larger cached tiers for the same revision to eliminate redundant decoding.
Cancelable Decode Priority: Prioritized visible thumbnails before speculative look-ahead, using monotonic viewport generations to drop stale decoding jobs.
📂 Streaming Ingestion & Filesystem Watcher
Native Directory Watcher: Registered long-lived platform watchers with a durable SQLite journal, 750ms quiet-period coalescing, and targeted path reconciliation.
Streaming Folder Scans: Streamed media entries into bounded database batches without keeping full directory trees in memory.
Throttled Scan Progress: Coalesced scan progress IPC events to at most one per 64 files or 100 ms.
Startup Scan Cooldown: Added a six-hour default cooldown preventing repetitive full-tree walks on app launch.
🚀 Bundle & Resource Reduction
Async Modal Bundle Splitting: Dynamically imported 18 infrequent modals and drawers, reducing the entry JavaScript bundle by ~103 KB and CSS by ~87 KB uncompressed (~25 KB gzipped).

Berry AI Studio v0.2.0

See CHANGELOG.md for full release details.

Berry-AIGC-Toolbox v0.1.3

See CHANGELOG.md for full release details.

Berry-AIGC-Toolbox v0.1.2

See CHANGELOG.md for full release details.

Berry-AIGC-Toolbox v0.1.1

See CHANGELOG.md for full release details.