Can someone point me towards how to increase the resolution of my orthophotos?
And perhaps related … is there any documentation as to the various TIF file options available for the orthophoto downloads?
GeoTIFF (RAW)
GeoTIFF (rgb)
And the 4 different projections:
UTM
LatLon
Web Mercator
Custom EPSG
I have been reading Piero Toffanin’s Book “OpenBroneMap - The Missing Guide”, but I can’t find a reference to these options. There is a treatment as to how the orthophoto is created … I have some studying to do to better understand the geometry behind the creation of the mesh and the orthoprojection of that … whew!
I’m using a grid of 406 images created with my Mavic 2 Pro and 80% side overlap. Each image is 5464x3640 pixels … and roughly a 14x14 grid of images … RAW format images.
The resulting orthophoto is 6280x5474 pixels. and can be seen here.
I did NOT resize the images in WebODM … and used default settings.
Thanks for pointing me in the right direction to sort this out … still a newly here.
If you know your approximate ground sampling distance, you can set that with the orthophoto-resolution flag to increase the output resolution. The units for this are cm/pixel.
Thanks for the response. That worked like a champ. It was a good exercise for me to see that my pixel resolution was 2cm (at best) for the Mavic 2 Pro from 285ft. So I changed orthophoto-resolution to 2.5 (default was 5.0) and I ended up with a pretty good looking 12,500x11,000 pixel orthophoto.
Awesome! Yeah: sometimes I wonder if we shouldn’t increase the default. Afterall, we do cap that value to a calculated value.
We do have a new pre-built drop-down for WebODM called “Ultra Quality” which increases feature and point cloud quality to the max and sets the orthophoto resolution to 1cm. It won’t generate 1cm if your data doesn’t support it, but it’s a nice new default that is useful for folks with detailed datasets and lots of RAM.
If ultra/ultra is unachievable with your current machine, a step up from default option is high resolution option which works well and does not subdue RAM like that of ultra/ultra.
I have found that using ultra ultra is not really needed with a good dataset that has plenty of overlap and well shot images.