Fragmented Point Cloud Issue in WebODM on VPS

[kemaluzel]

I have a dataset consisting of 240 images. When processing it on a VPS ( WebODM 2.7.2) with 4 CPUs and 16GB RAM using the following parameters: “auto-boundary:true, rolling-shutter:true, dem-resolution:0.4, dsm:true, feature-quality:high, optimize-disk-space:true, max-concurrency:2, tiles:true, gps-accuracy:0.1, no-gpu:true, orthophoto-resolution:0.2, use-exif:true”, the point cloud appears fragmented.

However, when I run the same process on Desktop WebODM without “no-gpu”, the point cloud is clean. ( WebODM 2.7.2)

If I enable “no-gpu” on Desktop WebODM, the point cloud also becomes fragmented, just like on the VPS. The main issue is that even if I remove “no-gpu” or “max-concurrency:2” on the VPS WebODM, the point cloud still gets generated in a fragmented manner. I would greatly appreciate any help in resolving this issue.

(edit: transfer retired, so link removed -- edit by @smathermather)

[kemaluzel]

I performed a clean installation of WebODM 2.8.0 on Ubuntu Server 22.04, but the same issue (fragmented point cloud) persists. Could you please help me resolve this problem?

[harunalt]

I'm experiencing the exact same issue described here. When I run WebODM on a cloud server, this problem occurs. Could the developer team please help resolve this issue?

[pierotofy]

This is due to GPU vs. non-GPU runs on ODM. Moving to the ODM repo.