How did you install ODX (docker, installer, etc.)?
Using ODX transitively via NodeODX (docker)
What's your browser and operating system? (Copy/paste the output of https://www.whatismybrowser.com/)
Google Chrome, Windows 11
What is the problem?
There's a regression in model quality. I can't pinpoint the exact change that caused this because we use NodeODX:latest. The last known healthy image we used was about a month old. Yesterday we did a re-deploy with latest image and our models turn out horrible now.
How can we reproduce this? (What steps trigger the problem? What parameters are you using for processing? Does it process with the default settings? Include screenshots. If you are having issues processing a dataset, you must include a copy of your dataset uploaded on Dropbox, Google Drive or https://dronedb.app)
I can't really provide the source images for this as they're not owned by me but the config that we're using via PyODX is:
TASK_OPTIONS = {
"skip-report": True,
"skip-orthophoto": True,
"use-3dmesh": True,
"feature-quality": "ultra",
"mesh-octree-depth": 10,
"mesh-size": 250000,
"auto-boundary": True,
"auto-boundary-distance": 5,
"crop": 0,
"pc-quality": "ultra",
"pc-filter": 2,
"texturing-keep-unseen-faces": True,
"texturing-single-material": True,
}
New, broken output:

Old, "correct" output:
I did exactly the same settings with the same input images. One workflow uses nodeodx:latest and the other one uses nodeodx:2.3.1.
That said, the issue is "resolved" for us by reverting to the older stable version. But there's clearly some sort of regression, at least for this specific setup.
How did you install ODX (docker, installer, etc.)?
Using ODX transitively via NodeODX (docker)
What's your browser and operating system? (Copy/paste the output of https://www.whatismybrowser.com/)
Google Chrome, Windows 11
What is the problem?
There's a regression in model quality. I can't pinpoint the exact change that caused this because we use
NodeODX:latest. The last known healthy image we used was about a month old. Yesterday we did a re-deploy with latest image and our models turn out horrible now.How can we reproduce this? (What steps trigger the problem? What parameters are you using for processing? Does it process with the default settings? Include screenshots. If you are having issues processing a dataset, you must include a copy of your dataset uploaded on Dropbox, Google Drive or https://dronedb.app)
I can't really provide the source images for this as they're not owned by me but the config that we're using via PyODX is:
New, broken output:

Old, "correct" output:
I did exactly the same settings with the same input images. One workflow uses
nodeodx:latestand the other one usesnodeodx:2.3.1.That said, the issue is "resolved" for us by reverting to the older stable version. But there's clearly some sort of regression, at least for this specific setup.