CamReview v0

Fast, local-only review of trail camera media on Windows. No cloud. No Docker. Safe deletes.

Features

• One-at-a-time review with Keep / Delete / Favorite
• Swipe gestures and big buttons (mobile-first)
• Undo last action
• Arrow-key browsing + on-screen arrows (no decision required)
• Desktop library drawer with search/filter + critter columns
• Desktop-only batch detect with live per-image feedback
• Resizable library pane on desktop
• Safe deletes (moves into dated Trash_YYYY-MM-DD/)
• Favorites, Keeps, and Deletes move into dated folders immediately
• Persistent JSON metadata
• Video streaming with phone-friendly fallback
• Optional AI batch critter detection (desktop)
• Optional AI captions with editable text

Quick start

1. Copy config:

copy config.example.json config.json

2. Edit config.json and set mediaRoot to your trailcam folder.
3. Install dependencies:

npm install

4. Start the server:

npm start

The app listens on http://0.0.0.0:3000. On your phone, open http://<your-lan-ip>:3000.

For auto-reload during development:

npm run dev

Controls

• Swipe right = Keep
• Swipe left = Delete
• Swipe up = Favorite
• Buttons always work
• Arrow keys browse the queue
• K = Keep, D = Delete, F = Favorite

Desktop vs mobile

• Desktop view is power-user mode: library drawer, resizable pane, batch detect, full keyboard shortcuts, and hover hints.
• Mobile view is streamlined: big buttons + swipes, minimal chrome, and no batch detect to keep it fast and thumb-friendly.
• Desktop plays the original MP4 with sound. Mobile uses a lighter preview stream when playback is flaky, so it may look choppier but stays reliable.
• Both views share the same review queue, metadata, and safe delete behavior.

Non-goals

• No filename changes or renaming.
• No permanent deletes (everything is a move you can undo in Explorer).
• No auto-organized albums or collections.
• No cloud sync, accounts, or multi-user features.
• No editing, filters, or color adjustments.

How deletes work

• Delete moves files immediately into Trash_YYYY-MM-DD under mediaRoot.
• Favorite moves files immediately into Favorites_YYYY-MM-DD under mediaRoot.
• Keep moves files immediately into Keep_YYYY-MM-DD under mediaRoot.
• No permanent deletes in v0.

Metadata

The metadata file is trailcam_review.json and is SQLite-friendly (flat rows). It stores:

• sessionDate
• path
• status
• reviewedAt
• caption
• ai (reserved)
• critter, critterConfidence, critterCheckedAt, critterModel

sessionDate is created automatically on the first review action and is used for the dated folder names.

If the metadata file is corrupt, a backup copy is created and a fresh file is used.

Video behavior

• If a video fails on mobile, CamReview auto-creates a smaller H.264 MP4 on the server (requires ffmpeg) and plays that version.
• Previews are generated at previewFps up to previewMaxFrames.
• You can also drop a sidecar preview image (.gif, .jpg, .jpeg, .png) with the same base name or inside .camreview/previews/.

AI critter detection (optional)

CamReview can call OpenRouter in batch mode (desktop only). It runs through images without critter data, moves no-animal photos into Trash immediately, and writes captions for the animal photos.

Captions are always editable inline after generation.

Requirements:

• Set OPENROUTER_API_KEY in the server environment.
• Optional: set OPENROUTER_MODEL to override the default model.

Results are stored as critter and critterConfidence fields in the metadata.

Configuration

See config.example.json. Options:

• mediaRoot (required)
• ffmpegPath (optional, full path to ffmpeg.exe)
• previewFps (optional, default 2)
• previewMaxFrames (optional, default 24)
• openrouterModel (optional, default openai/gpt-4o-mini)

API (v0)

• GET /api/items -> list unreviewed items
• POST /api/action -> keep/delete/favorite and move immediately
• POST /api/undo -> undo last action
• GET /api/preview-frames?path=...&generate=1 -> list or generate preview frames
• POST /api/transcode -> create phone-friendly H.264 MP4
• GET /media?path=... -> stream image/video
• POST /api/detect-critters -> detect animal presence for the current image
• POST /api/caption -> save a caption for a file
• POST /api/caption/generate -> generate a caption for the current image

Tests

API tests run against a temporary copy of a few sample files from your C:\Users\danny\iCloudDrive\trailcam folder (originals are never modified).

Run:

npm test

Troubleshooting

• If ffmpeg is installed but not found, set ffmpegPath in config.json.
• If mobile video still fails, check the server console output for ffmpeg errors.