Saving and exporting

The source .copc.laz is never rewritten, through any path. Edits and findings stay in a separate file, a sidecar.

Autosave

Edits autosave to localStorage per dataset and restore when you reopen the same data. They survive closing the browser.

That said, this is a local copy tied to that device and browser. To hand it over as a deliverable, or continue on a different device, export it to the sidecar below. The server side of collaboration persists nothing — once everyone leaves, the room is discarded.

The sidecar

"Save (.oniyanma.json)" in the ✎ Edit panel downloads <dataset>.oniyanma.json (command: saveEdits). Loading it back is the same panel's "Load".

The contents are the Command log plus a dataset signature.

{
  "format": "oniyanma-edits",
  "version": 3,
  "datasets": [{ "name": "kasado.copc.laz", "pointCount": 95169313, "bounds": [/* real-coordinate AABB */] }],
  "frame": { "offset": [/* x,y,z */], "alignYaw": 0 },
  "crs": { "epsg": 6669, "verticalDatum": "orthometric" },
  "commands": [/* hide / delete / reclassify / colorOverride … each command carries an actor, and a batch if it has one */],
  "entries": { "FindingType": [/* … */], "Finding": [/* … */], "Batch": [/* … */] }
}

Key	Role
datasets	A lightweight signature (name, point count, real-coordinate AABB) for detecting a version mismatch. No full hash is taken of a huge file
frame	The definition of what basis the predicates' coordinates use. scene = Rz(-alignYaw)·(real - offset)
crs	The coordinate reference system's identifier (recorded only — never reprojected)
commands	The non-destructive-edit log. Each command carries its issuing actor
entries	Findings, finding types, batches (findings added in version 2, batches in version 3). For batches, see → Non-destructive editing

When loading is refused

If conditions don't match, the sidecar states the reason and doesn't apply, rather than applying it. Silently applying edits to the wrong points would be worse.

Message	Meaning
These edits may be for different data (signature mismatch)	datasets' signature doesn't match the current data
Not applied because the coordinate frame differs	frame differs. Reload the data in the same order
Not applied because the coordinate system (CRS) differs	crs differs
This sidecar has no coordinate-frame record	An old format. A warning that positions may be off if the frame differs
This is a newer-version sidecar	The saved version is newer than this app. Refuses to load, since opening it with an older version would lose edit history
The sidecar's format is invalid	The JSON is broken / a different format

Predicate coordinates are in the aligned frame, so applying a sidecar with a missing or different frame would have the same condition select different points. This is the crux of what gets checked.

An older-version sidecar is migrated before it's loaded (edit/migrate.ts). Only opening a newer version with an older app is impossible.

Exporting selected points

"Export" in the ▚ Select panel, or the command exportSelection. Choose CSV (default) or GeoJSON.

await oniyanma.execute('exportSelection', { format: 'geojson' })

• What comes out is real coordinates (m), classification, and RGB
• It's resident-based (the LOD sample currently on screen) — the count grows as you get closer
• Deleted points are excluded, and the count comes back in excludedDeleted
• Besides the file download, the return value also carries the content (the full text if small, or a preview of the first 2000 characters if large). An AI can't see the file dialog, so it can't confirm the result unless it comes back in the response

Exporting findings

exportFindings exports to CSV / GeoJSON. The default is published only (the form you can hand to the client); view: 'draft' includes drafts too. → Filing findings

Sharing view state

To hand over "what it currently looks like" rather than edits, use 📁 Data's "Copy share link". The data URL, camera, color mode, point size, EDL, grid, and section all ride in the URL's hash. → URL parameters