I have run 280 pictures on the webodm lightning cloudservice.
The pointcloud is 300 meters higher than real world. The quality report says that lowest point is.460 meters, but in real world the lowest point is 150 meters above sealevel.
I am flying the drone on different elevations since I aml following the terrain. Is this causing the problem ?
The drone used is a DJI Mavic 3.
Can anoyone suggest what I am doing wrong ?
Here is the options I ran:
|Processing Node:|Lightning (manual)|
|Options:|auto-boundary: true, dsm: true, dtm: true|
|Average GSD:|5,57 cm|
|Område:|989 109,3 m²|
|Reconstructed Points:|13 629 859|
Loggfile, Quality report, source photos etc can all be viewed here: Jottacloud
Thanks - you pointed me in the right direction. I took some samples from the drone and it is constantly way to high in the EXIF data. This is a brand new Mavic 3, hopefully it may be fixed with coming firmware updates.
Maybe! We’ve seen reports from various combinations of sUAS and firmware that the elevation recorded in EXIF is not always consistent (Above MSL, Above GL, Above take-off, etc) or what one would expect…
If you can, a-priori in-situ GCPs will not only greatly improve your Z-axis accuracy, but your X/Y as well. If you have access to the site and can spray field chalk Xs and record their location with a reasonable-quality GPS, you’re golden. Mats, AruCoMarkers, or immobile/immutable easily distinguished features are also good options.
Or if not, you can use something like exiftool to apply a flat offset to try and correct for the elevation offset.
Hi there, I wonder if you can expand a little on what you mean by a reasonable quality gps please as I have discovered my model and 2d orthophoto are incredibly innacurate and need to start using GCP’s.
So in the case above, you are discussing the height, but I have discovered that my model is bigger, much bigger. The obviousness is clear when you see the orthophoto overlaid on the google maps and you can toggle the opacity. And on this particular project it is really far out, but not very much at all on other projects. So to investigate further I physically went back to the site and measured a couple of measurements with a long tape measure. And both measurements were around 30% bigger on the model than in real life! This is a lot I must say. So I have been investigating using GCP’s and wonder if there is a reasonably quality gps that you can recommend without spending thousands. I wonder if you have some experience or can point my to a thread that might help improve this inaccuracy or more specifically what you use as a reasonable gps receiver.