Hey all,
I downloaded a sample set of images from a photogrammetry vendor of the new DJI Mavic 3 multispectral. I was able to process the ndvi of the ms imagery but it failed when I tried the rgb ortho.
I’m processing with WebODM Lightning. The RGB imagery of the dataset includes 350 images for a total of 3.2 GB. I’m guessing this is something that Lightning can’t handle? Do I need to downsample the images or is there some option that I need to include to run these.
Thanks
What resolution (MP) are the RGB images?
5280x3956
oh sorry 20MP
Can you provide the full Task log?
It’s been awhile. Can I access this somewhere besides the dashboard. I didn’t write out the output to the dashboard.
350 images at that resolution should be no problem memory-wise. But yes the task log would help a lot to understand what happened.
I’m running it again using the lightning node. I’ll capture the logs this time. It’s a very uniform ag field with a kind of vignette thing on the corners of the images. It may just struggle due to lack of diversity in the imagery.
It completed this time but it took 7 hours and 40 minutes. The log is attached. The difference this time is I used the Lightning node instead of the Auto choice. It doesn’t look like I can upload the log file due to not being one of the proper extensions
You can post a link to it on a sharing site perhaps?
The console log is just a text file.
Even with: min_num_features: 10000, you are typically finding over 100,000 features per image, so I don’t think there is a need to increase that number.
1 partial reconstructions in total
is good
Maybe try on your machine using ultra feature extraction and PC quality?
This topic was automatically closed 30 days after the last reply. New replies are no longer allowed.
