I got two well overlapping datasets taken at the same altitude, but different flight direction angles, they suppose to be mixed together and processed, but they don’t.
The problem with them is they were taken in different time of the day and light conditions.
Another dataset taken couple hours before sunset in clear sky so the same overlapping area look quite quite different.
As there is dense forest with tree canopy only, with features to be dag in there. I guess ODM can’t find any features to match in those two pieces, to be able to add them all in the reconstruction.
My question is, looking at their difference on the last photo anybody can tell, if they can be fixed somehow on the computer (enhanced, improved photoshopped whatever it might take) and then processed again by ODM together, so they’d stitch or I just better go there and re-take the afternoon one again but in the morning, trying to catch similar light condition with the other one so they finally stitch together as they usually do.
It’s probably a good idea… I guess it’s easier to make flat one more relief than trying to flatten the afternoon one…
Thanks a lot… I will give it a try…
Anything in the processing parameters could be tweaked to help them join?
There’s diminishing returns for sure, as well as increasing RAM load and CPU time. I would not go much higher than maybe 64k. You’re already at 4x normal min-num-features.
But to answer your first question, I’m not sure what the actual input limit is, or if we even cap it.
Sky is the limit
I found quite convenient and flexible cloud computing facility where I am charged reasonably per hour of use… so for RAM and CPU demanding datasets I can stretch it to 64 cores CPU and 192GB RAM for 0.75 EUR an hour just to process something specific then reduce power down to something small and inexpensive to have run idling or just make backup image of my server and delete server itself until I need it again. Getting it set up and running in a few minutes when I need it again.
