Messages

1

Bug Reports / Exported point cloud in chunks has local coordinates, not real-world coordinates

« on: May 09, 2024, 07:46:57 PM »

Problem: When exporting a point cloud into chunks, the coordinates are changed to local coordinates without telling you (even if the output projection is set to real-world coordinates).

I'm running into a situation where when I export a point cloud as a single tile in UTM coordinates, the resultant file is fine (i.e. it has the correct extent, resolution, and point density). However, when I try exporting the point cloud by tiling into 1km chunks, the resultant cloud has a density of 0 points / m2 and the extent appears to be switched to local coordinates (even though the CRS is correctly set).

I found one feature request (https://www.agisoft.com/forum/index.php?topic=15683.msg67887#msg67887) that appears to be somewhat related, but I don't know for sure if this behavior is a bug or if this is intentional. If it is the latter, can there be a warning or something that if people are exporting a point cloud to chunks, that the coordinates will NOT be what they expect?

2

General / How to align orthomosaics using external point cloud?

« on: February 07, 2024, 04:51:09 PM »

Hi,

I am struggling with aligning orthomosaics using an external point cloud.

This is my initial workflow:
- upload georeferenced photos to Metashape
- align photos and create point cloud
- export point cloud

I then align the photogrammetry-derived point cloud to a LiDAR-derived point cloud using CloudCompare. This involves first retiling the photo-point cloud to a standard grid of 844 tiles.

What I want to do is then use this (now-external) aligned point cloud to create an aligned orthomosaic. However, this is not working. I've tried the following attempts:

Attempt #1
- upload aligned point cloud (844 tiles) and merge into single cloud
- create DEM, then orthomosaic
- Result = not aligned

Attempt #2
- upload aligned point cloud as laser scans
- create orthomosaic directly from the scans
- Result = exactly aligned, but the resolution is coarse and there are multiple blurry spots (still apparent even when using a higher-resolution point cloud)

Attempt #3
- (following this tutorial: https://agisoft.freshdesk.com/support/solutions/articles/31000168474)
- create new chunk
- upload aligned point cloud as laser scans, group them, change to Fixed, disable lock transform
- re-upload photos
- align the photos (now Metashape uses the laser scan info)
- build a new point cloud, then DEM, then orthomosaic
- Result = not aligned, but slightly better than Attempt #1


What I want is to get the output of Attempt #2 but with the orthomosaic using the photos. It seems to do that I would need to directly transform each individual photo but I have 844 transformation matrices and >1000 photos.

Is there another way I can get a fully aligned orthomosaic? I know the new point cloud is fully aligned but I can't figure out how to do this.

3

General / Re: Correct workflow for aligning photogrammetry orthomosaics

« on: February 05, 2024, 07:26:26 PM »

Bumping this, can anyone either at Agisoft or otherwise verify if the workflow is correct or not?

4

General / Correct workflow for aligning photogrammetry orthomosaics

« on: January 25, 2024, 04:15:44 PM »

Hi,

Context:
- we want to difference two orthomosaics to see where NDVI changes, so we need the orthos to be perfectly aligned. However, this is apparently more difficult than I thought.
- We are using Metashape Pro v2.0.3.16960.

This is my current workflow. I do this separately for each flight.
- load photos and align
- build point cloud
- export point cloud, align to a standard one we have in CloudCompare software
- start a new chunk and re-upload the aligned point cloud as laser scans.
- re-upload the photos and align again (this time I understand it uses the laser scans).
- build a new point cloud
- Build DEM, and then build orthomosaic.

The result is that one side of the study area is almost aligned while the other side is not. I can confirm that the CloudCompare outputs are aligned because I've checked on DSMs created from them and they are perfect.

I have tried the following without success:
- aligning the two flight chunks
- uploading the aligned point clouds as point clouds instead of laser scans, merge to single cloud, then build DEM via extrapolation
- following the tutorial as described here for terrestrial lidar scanning: https://agisoft.freshdesk.com/support/solutions/articles/31000168474-terrestrial-laser-scans-processing-in-metashape-2-1
- building the orthomosaic directly from the laser scans *does* produce aligned orthomosaics, but the resolution is worse and there are many artifacts in the image that make comparison difficult.

My questions:
- am I missing something in my workflow that would help the alignment?
- does anyone have any recommendations for something else to try?

Thanks!