Unable to import Aerial Lidar data

RealityScan 2.1 auto-detects all LAS point clouds as “Mobile LiDAR”, including genuine aerial LiDAR from both drone sensors and manned aircraft surveys. The LiDAR type field is locked and cannot be changed to Aerial. This prevents the aerial LiDAR workflow (virtual camera generation, etc.) from being used.

Environment

  • RealityScan version: 2.1

  • OS: Windows 10/11

  • GPU: NVIDIA (CUDA enabled)

Steps to Reproduce

  1. Open RealityScan 2.1, create a new project

  2. (Optionally) import georeferenced aerial images with EXIF GPS

  3. Go to WORKFLOW tab → click LiDAR Scan button

  4. Select a LAS file from aerial LiDAR source

  5. The LiDAR Import dialog opens with “LiDAR type: Mobile LiDAR (only supported mode for this f…)” — the field is locked/grayed out

LiDAR Sources Tested

Both of the following are misclassified as Mobile LiDAR:

1. DJI Zenmuse L3 (drone-mounted aerial LiDAR)

  • Exported from DJI Terra as merged LAS

  • LAS 1.2, Point Data Record Format 3 (GPS time + RGB color)

  • CRS: EPSG:4326 (WGS 84) — embedded as GeoTIFF VLR keys + WKT OGR

  • 138M points, 5 returns, classification (ground + unclassified)

lasinfo header confirms valid georeferenced aerial data:

version major.minor:        1.2
point data format:          3
point data record length:   34
number of point records:    138363771
number of points by return: 135968330 2080469 281524 29889 3196
scale factor x y z:         0.00000001 0.00000001 0.0001

GeoKeyDirectoryTag:
  key 1024 - GTModelTypeGeoKey: ModelTypeGeographic
  key 2048 - GeographicTypeGeoKey: GCS_WGS_84

WKT OGC COORDINATE SYSTEM:
  GEOGCS["WGS 84", ... AUTHORITY["EPSG","4326"]]

2. Norwegian national aerial LiDAR (høydedata.no / Kartverket)

  • Manned aircraft survey data from the Norwegian Mapping Authority

  • Standard aerial LiDAR distributed as LAS/LAZ

  • Also misclassified as Mobile LiDAR

What Happens After Import

When clicking OK with Mobile LiDAR classification and “Use camera poses: From camera pose priors”:

  1. Verify Coordinate System dialog appears: “You have specified a geodetic coordinate system. The app can import scans in this coordinate system but they are usually registered in projected coordinates.”

  2. Clicking Import Anyway“No images found” error, even when images are loaded in the project

  3. Clicking From component requires an aligned component, but even with aligned images, the Mobile LiDAR path doesn’t produce the virtual camera generation options that the aerial workflow should provide

Expected Behavior

Based on the Aerial LiDAR Support tutorial and the Alcatraz case study, RealityScan should:

  1. Auto-detect aerial LiDAR from LAS files, or allow manual override of the LiDAR type

  2. Offer the “Generate aerial poses” option for virtual camera generation

  3. Successfully create .lsp files from the aerial point cloud for alignment and meshing

Questions

  1. What criteria does RealityScan use to distinguish between Mobile and Aerial LiDAR in LAS files? Is there a specific header field or metadata it looks for?

  2. Is there a way to force the aerial LiDAR import path — e.g., a specific LAS version, point record format, or metadata tag that triggers aerial detection?

  3. Could the LiDAR type dropdown be made editable rather than locked, so users can manually override the auto-detection?

Any guidance on getting DJI L3 aerial LiDAR into RealityScan would be greatly appreciated. This is currently blocking a drone photogrammetry + LiDAR fusion workflow for simulation projects.

Thanks!

Hello @OddbjornL

Since 2.1 version all “movable” scans were grouped as Mobile LiDAR. The only other option is Terrestrial LiDAR. So, all aerial LiDARs are considered as mobile one in 2.1. So, that is OK for both your datasets.

What were your other import settings? Do you have set the coordinate system properly? As you mentioned that you have also images in the dataset, are they georeferenced and aligned?

As you mentioned the option Generate aerial poses, are you missing that in the settings? It should be there automatically for selection.

Regarding the No images found warning, are those images georeferenced?