Timemark Plugin

Import photo sets from Timemark (Android stamp camera, com.oceangalaxy.camera.new) without losing either the pixels or the data — and work around the fact that Timemark will not give you both.

The problem this exists for

No Timemark export preserves the full-resolution image and the field values together.

Input	Full-res pixels	Field values	Tags
Original images (a Photos_from_Timemark ZIP is just these bundled, byte-identical)	yes	filename text only	filename only
Photosheet XLSX	no — 720×960, ~2% of the bytes	yes, as real columns	dropped
Work Report PDF	no — 617×823	prose, time cut to the minute	yes

So an import is a reconstruction: the images and the photosheet, joined on the capture timestamp to the second. The PDF can never participate — it drops the seconds, so two frames from the same minute are indistinguishable.

The practical consequence, and the reason the first skill is a gate: if only half the export arrives, the other half is not recoverable later from what you have, and the photos usually get cleared off the phone.

Skills

Skill	Does
timemark-import	the gate — verify both halves are present, then join. Run first, always
timemark-burn-in	write the values into the images as XMP, so each photo is self-describing
timemark-archive	build a self-contained bundle: full images + JSON sidecars + an index

Burn-in and archive are complementary, not alternatives.

Use

/timemark:import ./exports

Or directly:

python3 scripts/timemark_archive.py photos/ \
  --photosheet Photosheet_2026-08-26_to_2026-09-01.xlsx \
  --out inventory-2026-09-01 --embed-xmp --zip

Producing timemark-archive/1:

inventory-2026-09-01/
├── index.json          every record: image path, values, tags, sha256, provenance
├── README.md           generated — explains the format to whoever opens it
└── images/
    ├── <original filename>.jpg        full resolution, untouched, unrenamed
    └── <original filename>.jpg.json   sidecar, Google Takeout convention

Scripts are stdlib-only Python 3 — no openpyxl. XMP embedding needs exiftool.

Traps this handles for you

• exiftool silently drops undeclared XMP tags. -XMP-xmp:ItemNo=1 exits 0, prints nothing, writes nothing. Custom fields need a generated -config, so the plugin generates one per export and reads the tags back to confirm.
• XLSX images are not in row order. xl/media/imageN.jpg numbering is unrelated to data rows — the real mapping is the anchors in xl/drawings/drawing1.xml. Index-order zipping mislabels every photo, silently.
• Filenames must be preserved verbatim. Tags exist only there, and the timestamp is the join key.
• GPS is written zeroed, not omitted, when location was off — testing for tag presence puts every frame on Null Island.
• OffsetTimeOriginal ignores DST, so UTC conversion off it is an hour out.
• UserComment is a 5–7 KB encrypted blob. The fields are in there and are not readable. Don't spend time on it.

Full measurements: reference/export-formats.md.