Smart Delta SmartPano Gen2

SmartPano Gen2 captures: panoramas sitting flat at the bucket root with one trajectory CSV beside them, and a coloured point cloud from the SLAM session.

Folder layout#

<dataset bucket>/
├── GPS_Trajectory.csv          exact filename, exact case
├── 0001.jpeg                   panoramas at the bucket ROOT
├── 0002.jpeg
└── slam_session/
    └── merge_color/
        ├── cloud_001.laz
        └── cloud_002.laz

Two things people get wrong here:

Panoramas are .jpeg and live at the bucket root, not in a subfolder. Only files directly at the root are seen.
The point cloud lives at exactly slam_session/merge_color/. That is the coloured output of the SLAM run, not the raw session folder.

A missing GPS_Trajectory.csv skips the panorama half with a note; a missing slam_session/merge_color/ skips the point cloud half. Neither is fatal on its own.

GPS_Trajectory.csv#


Separator	Comma
Header row	Required, and columns are addressed by name
Extra columns	Ignored — only the eight below are read
Blank cells	Read as empty strings, not nulls

Column	Type	Meaning
`imagename`	text	Photo filename including the `.jpeg` extension
`x`	number	Easting, in the dataset's CRS
`y`	number	Northing
`z`	number	Height
`heading(deg)`	number	Degrees
`pitch(deg)`	number	Degrees
`roll(deg)`	number	Degrees
`UNIX Timestamp`	number	Epoch seconds, read as UTC

The parenthesised suffixes and the space in UNIX Timestamp are part of the names — they are matched literally. A missing or misspelled column fails the run.

Unlike several other templates here, imagename is used verbatim: no extension is appended. Write 0001.jpeg, not 0001.

Example#

UNIX Timestamp,x,y,z,heading(deg),pitch(deg),roll(deg),imagename
1765631346.704,182518.396,477934.743,37.326,0.0,0.0,0.0,0001.jpeg
1765631348.113,182519.204,477936.881,37.341,1.4,0.1,-0.2,0002.jpeg

Tolerance#

Up to 10 % of CSV rows may reference missing photos. Past that the run fails immediately, naming the threshold.
Photos at the root that no CSV row mentions → listed at the end of the run and left unindexed.

Ingest parameters#

Key	Required	Values	Default
`height_above_ground`	yes	number, metres	—
`anonymisation`	yes	`already_anonymised`, `auto_anonymise`, or anything else for neither	—
`tile_pointclouds`	yes	anything except `no` enables grid tiling	—
`set_name`	no	string	`1`

tile_pointclouds is required here even though it is optional elsewhere. Pass no explicitly when you don't want tiling.

Coordinate reference system#

Nothing is reprojected. Positions are stamped with the dataset's SRID and point cloud boundaries come from the LAS headers unchanged. See Coordinates are not reprojected.

See Rotation angles.

What you get back#

Indexed panoramas with positions and orientations, plus blurred derivatives if anonymisation is auto_anonymise.
Indexed point clouds and a merged COPC at copc/pointcloud.copc.laz.
Grid-tiled point clouds under pointcloud_tiles/ at 100 m when tile_pointclouds is on — the tiles are then what gets indexed.

Common failures#

Symptom	Cause
"GPS_Trajectory.csv NOT found"	Wrong name, wrong case, or nested in a subfolder
Every photo reported missing	`imagename` omits the `.jpeg` extension, or the photos are in a subfolder rather than at the root
More than 10 % of photos missing	Filename case mismatch, or `.jpg` delivered where `.jpeg` is expected
No point cloud indexed	The LAZ files are not under `slam_session/merge_color/`