Hi guys, I have a problem with orthophoto processing. I did full coverage of 100ha (1km²) area with default overlap, ~ 8cm/px GDM. Result looks like this:
I get good results with my Phantom 4 Pro drone when mapping my heavily treed 7ha site by flying missions on overcast days, at 91.4 m’ altitude with 80% front & side overlaps at 4 m per sec to minimize blurring in the direction of travel. This results in a 1/60 sec shutter speed or 6.7 cm camera travel while the shutter is open. The mission takes 18 minutes on one battery and generates about 200 photos for uploading. I review all the photos before uploading and remove any with noticeable blur. This typically removes all photos at the 9 turns in the mission which are blurred due to the rapid rotation.
You can see some of my results in my post today (10 Jan 2018) titled: “Automatic Tree Detection and Characterization with DroneDeploy”.
Your site is 14X larger than mine. So it would require 14 battery swaps if you used my settings. A challenge.
What are the key setting for your mission? Altitude, sun or overcast conditions, speed, overlaps?
I’ve only had success flying at a minimum of 80/80 for forested area. 90/90 works perfectly but the mission times and processing times are so high it makes it hard to do.
My altitude was 267m (230m above trees), side/front overlap 65%/75% , speed 15 m/s.
I will retry with more overlap. Key problem here is that I have really large area to cover (100 km²) in relatively short time (month), low budget. I’m looking for parameters to give me successfully stitched maps done with 1-2 batteries max per km².
I do a lot of forest mapping and figured I’d share what i’ve discovered. This is all purely anecdotal and I’d love to hear insight from others to see how we can improve things. We fly a Mavic.
During the summer with relatively homogenous tree canopy it seems to be much tougher to stitch due to the lack of distinguishable control points- we only have solid luck during leaf-off conditions. But in your case you do seem to have success in some homogeneous areas but not in others, so maybe our theory is wrong. All I know is that on an overcast day with snow on the ground we have 100% success. In fact, that’s the only time we bother now- we do all our mapping in the winter. Obviously that isn’t an option for many!
During leaf off, elevation needs to be higher than 130m- we usually fly at 330m (yes we are licensed for that) - because the tree distortion from the camera fish-eye effect at lower elevations makes the rendered image really blurry (we presume). But even at 330m elevation we get amazing resolution.
80/80 or 85/85 seems to be the sweet spot. Anything more than that doesn’t seem to improve image quality and just adds to processing time.
Take all of those considerations, and throw in changes in elevation. If there are more than a couple hundred meters of difference, we often lose the higher elevation, because 330m at the base becomes only 200m at the higher elevations of the parcel. In leaf-off conditions this isn’t a problem, but in the summer that seems to be part of the problem?
With those settings we can do 30-35 hectares per battery. With 6 batteries on hand and two 3-hub quick chargers, we can fly pretty much continuously, but at that scale (200+ hectares), we really need to think about getting a more efficient drone than the Mavic, like a Firefly.
I hope that helps someone. I figured it was worth posting somewhere! Happy mapping-
@VTDD Great, thank you for sharing your precious experience!
As it turned out, my dataset was corrupted. I checked all photos in pix4d desktop and that allowed me to diagnose problem - i have many photos in my dataset without overlap at all. It must be some error in DD flying app, picures were taken too sparse. Tomorrow if weather will be good I will repeat shooting.
