Still quite disappointed with webODM results

WebODM
#1

Hi,

This is my second venture into webODM, I’m using a Mavic Air to create georeferenced maps of forest and bog habitats.

Unfortunately I am finding the resulting georeferenced orthos disappointing. I ran a test on 46 jpegs, the image quality using the highest quality settings seems to produce quite poor, grainy results with weird edge effects, the georeferencing isnt too bad but could be much better. It also “chops off” the edges of the aerials. It also took 40+ minutes to produce the geotiff.

Ive found using Microsoft ICE to produce the mosaic is very quick (sadly no command line option) and produces way better, sharp images and it processed the images in less than 5 minutes. Then georeferencing is just a matter of either using GCPs or programmatically georeferencing the tiff using a bit of trigonometry from the exif data (in my case using R).

Has anyone any general advice on how to speed up the process, increase the image quality and georeferencing accuarcy? I love the idea of ODM and appreciate all the work people have but unless I’m missing something, in its current form its still a bit a way from useable for these purposes.

Thanks

1 Like
#2

The --fast-orthophoto option may be what you are looking for rather than a full photogrammetric process, if you like MS ICE. But even then, it won’t be as fast as ICE, as OpenDroneMap is a more complete and robust approach.

FYI, with a photogrammetric approach, there will always be clipping on the edges. ICE isn’t as flexible, and doesn’t work in as many conditions as OpenDroneMap, but for some use cases is the perfect tool. @Saijin_Naib has done some pretty cool work with it, if memory serves.

3 Likes
#3

Thanks smathermather,

I’ve experimented with different settings, I had accepted the “resize image” option which caused the artifacts, using full sized images removed that issue. Also the “forest” setting seems to work best for georeferencing, much better now.

Regards

2 Likes
#4

Fantastic. And thanks for sharing back your findings.