local photo intelligence console · pipelined vision engine
IDLE
0 / 0
processed
—
photos / min
—
sec / photo
—
eta
0
failed queue
0
tagged total
awaiting scan0%

◉ optical feed — pipeline frame

NO SIGNAL
—
model output will appear here

▚ engine log — telemetry stream

--:--:-- PHOTON pipeline ready

◫ category density

◸ recent results — last 8 photos (click to inspect/edit)

No recent activity yet. Start the pipeline to populate.

◫ classification matrix — categories total

⌕ photo library — search, browse & bulk edit

—
Loading tagged photos…
—
—
—
—
—

◷ audit history — accuracy over time

0
total graded
—
overall accuracy
0
passed
0
failed

Blind Accuracy Grader

This mode selects up to 100 random tagged images from your database. You will grade their description and category accuracy blindly to calculate a quality score for your prompt/model configurations.

◐ appearance — accent color & browsing preferences

picks the highlight color used across borders, buttons and badges. saved locally on this browser.

⚙ engine configurations

qwen3.5:9b has best quality (thinking disabled). glm-ocr is fastest scout classifier.
smart router
scout classifies screenshot vs real photo, routing to specialized model branches.
photos downscaled before analysis. lower = faster, original image files untouched.
0 = deterministic classification, higher = more descriptive variety.
skip already tagged
preserve file dates
keep _original backups
off = atomic metadata write. on = copies originals (doubles disk usage).
dry run (no writes)
tag videos too
videos are included in the same pipeline run — several frames sampled across each clip go to the video model in one call for real motion-aware description.

◫ custom categories manager

🧠 smart recategorize — auto-discover categories from your library

Analyzes descriptions already sitting in your journal (no images re-scanned, no vision model) to propose a richer category list when one bucket is swallowing the library, then reclassifies every tagged photo into it using a fast local text model.

🛡 trust & safety console

verify pixel integrity
checks image pixels pre/post write via double-BMP double-hash. Restores backup on mismatch. 100% corruption proof.
organize aliases
creates Portable Relative Symlinks inside folder categorized in _organized/. Originals remain completely untouched.
telemetry failures 0 FAILED

Removes all embedded PHOTON descriptions, categories, and keyword markers recursively. Clears organized symlinks and journals.