How to Import Livox Mid360 Point Cloud Data into RealityScan?

Development Asset Creation
,
1

Hello RealityScan Community,

I am working on importing my LiDAR scan data into RealityScan for processing and texturing, but I have some questions regarding the correct workflow. I would greatly appreciate any guidance.

My data comes from two related sources, and I’m unsure of the best path forward:

  1. Raw LiDAR Data:

    • Device: Livox Mid360
    • Data Format: A sequence of per-frame .pcd point cloud files, along with a scans_pos.json file (which I believe contains the pose/trajectory information for each frame).
    • My Question: I am unsure if RealityScan supports importing this kind of sequential, frame-based point cloud with an external pose file. What is the recommended method to merge these individual frames into a single, spatially accurate point cloud that RealityScan can use?

  2. Processed Color Point Cloud:

    • Processing Pipeline: I used the FAST-LIVO2 algorithm to perform a tightly-coupled LiDAR-Inertial-Visual odometry fusion on the raw data. This resulted in a complete, colorized, and globally consistent point cloud of the entire scene.
    • Data Format: The final output is a single .pcd file (this is the fused, full-scene point cloud).
    • My Question: Is importing this pre-reconstructed color point cloud the more straightforward approach for RealityScan? Are there any specific settings or considerations (e.g., point density, color format) I should be aware of during import for optimal results?

To summarize my core questions:

  • What is the correct preprocessing and import workflow for the raw, sequential Livox data?
  • For the complete color point cloud from FAST-LIVO2, is this the recommended method for use with RealityScan? Are there any known best practices or success stories?

Thank you in advance for any advice, experience, links to tutorials, or pointers to official documentation you can share!

截图 2025-11-03 11-23-14
截图 2025-11-03 11-23-14323×568 23.9 KB

2

Hi,
it looks like this device could be used as a aerial lidar for importing into RealityScan. But for that you will need to use LAS point cloud format to be importable.

3

Thank you so much for your response! The pointer towards using the LAS format and potentially treating the data like an aerial LiDAR scan gives me a clear direction to work towards.

To make sure I proceed correctly, I have a few follow-up questions:

  1. Regarding Point Cloud Merging: My raw data consists of multiple, sequential PCD files (individual frames) and a scans_pos.json file（this is from FAST-LIVO2） containing their pose data. Before conversion, do I first need to merge these frames into a single, complete point cloud representing the entire scanned space? If so, could you recommend any specific software or methods for this pose-based merging? For example, would tools like CloudCompare or Open3D be suitable for this task?

  2. Regarding File Format: Is the LAS format strictly mandatory, or does RealityScan also support other common point cloud formats like PLY? The final colored point cloud from my FAST-LIVO2 workflow is in PCD format, and I know that PLY often handles color information well. It would be helpful to know my options.

  3. Regarding Import Settings: Once I have the final point cloud file (whether LAS or PLY), are there any key settings or considerations I should be aware of when importing it into RealityScan? For instance, since this is from a ground-based Livox scanner rather than a traditional aerial survey, are there specific things I should check regarding the coordinate system, scale, or color channels?

Your guidance is incredibly valuable. Thank you again for your time and help

image
image445×376 7.23 KB

4
  1. Both options should be doable, for most cases is better to have an unified point cloud.
  2. Like also other formats are importable: PTX, E57, PLY, ZFS, LAS, PTS, CSV.
  3. For now the ground scans like that aren’t naturally supported, so there are no recommended settings for such case. But you can use the exact registration and georeferenced in local CS. For the other settings it depends if you have the images in the project or not.