UI overhaul to a darktable tone-and-manner and a set of features adapted from
darktable's proven patterns (reimplemented in our Electron/TS stack; no GPL code).
Design reskin:
- Dark neutral-gray palette + amber accent, flat/squared corners, no card shadows,
compact darktable-style top bar (logo + pipe-separated view tabs), denser 15px base
- Done via design tokens (Tailwind slate/brand/radius/shadow remap) — minimal churn
Metadata & collections (Phase A/B):
- exifr now captures GPS + camera; asset table ALTER-migrated (gpsLat/gpsLon/camera,
metaVersion backfill on re-index)
- Collection facet bar (year timeline / camera / color-label) filters the grid
Map & relation finder (Phase C):
- Leaflet + online OSM map tab; geotagged photos as markers
- relationService: related photos by place (GPS<1km) + time (+/-2d) + CLIP similarity
Easy mode (Phase D):
- easyMode setting (menu / onboarding); scales the whole UI (rem) + bigger thumbnails
+ large icon nav with plain labels (4050 accessibility)
Library usability:
- Video thumbnails (representative frame capture in the inference worker)
- Media filter (All / Photos / Videos) to separate them
- Clearer culling labels ("Good shots" / "To cull") + explanation tooltip
- Multi-select tiles -> Export selected to a folder (copy, best-cut extraction) and
Delete to Recycle Bin (shell.trashItem) behind a confirm dialog
- ONNX Runtime wasm bundled locally (offline) via copy-ort-wasm + asarUnpack
Docs: DARKTABLE_REVIEW (feasibility + roadmap A->D). All typecheck/tests/build green;
boot smoke verified each phase.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Builds the "indexed library" foundation and first intelligent features on
top of the organizer (sql.js index, non-destructive in-place indexing).
Phase 0 — Library index:
- sql.js (WASM SQLite) index DB; contentHash-keyed assets, resumable indexing
(skip by path+mtime), batch persistence (chosen over native better-sqlite3
which fails to build on Node 24 / Python 3.12)
- Library folders (in place, non-destructive) + background indexer w/ progress
- Thumbnails generated in the AI worker (canvas->webp), cached in userData;
served via photoai-media://thumb by hash; thumbnail grid w/ pagination
Phase 1 — AI quality assessment & culling:
- Focus (Laplacian variance), exposure (histogram), eyes-open (face-api EAR)
computed in one analyze pass alongside the thumbnail
- Culling filters (candidate/rejected) + quality badges
- Adjustable thresholds (live SQL re-classification from stored raw scores,
no re-analysis) + manual star rating (0-5) and color labels (usermeta)
Phase 2 — CLIP natural-language / similarity search:
- @huggingface/transformers (WASM/WebGPU, no native build)
- CLIP image/text embeddings (lazy-loaded); Korean queries auto-translated
via opus-mt-ko-en into the English CLIP
- Embeddings stored as SQLite BLOBs; "build search index" batch w/ progress;
brute-force cosine search; new Search tab
- Note: models download from HF Hub on first use; fully-offline ORT-wasm
packaging and KO search-accuracy tuning are follow-ups
Tabs added (Organize / Library / Search). All typecheck/tests(12)/build green;
boot smoke verified across phases.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
koriweb