Dronedeploy + P4 camera + metashape : holes in tie point

Hello Everyone,

Currently working on a big mapping project.

tiles are 500x500 m of sand and rocky deserts.

Gear :

Phantom 4 camera
Dronedeploy
Metashape

Dronedeploy was set in automatic settings :

75% front overlap
65% side overlap
pictures are taken with F/2.8, ISO 100 and 1/4000 shutter
fly altitude : 45 m
flight speed : 5m/S

I have pretty big holes in my tie points resulting in some errors in the orthomosaic. I think is probably linked to the sandy nature of the ground. There is not a lot of colour variations so after pictures alignement i often have bunches of pictures that metashape was’nt able to align.

So according to you guys, what parameter should I modify to get a better tie point?
increasing the overlap?
lower flight altitude?
slower flight speed?
changing camera settings?

I want to be able to do one tile using 3 to 4 batteries so i won’t be able to modify all parameters. So wich one would be more effective on the results?

Thanks a lot!

Can you share a couple images that are indicative of the scene? Normally you would fly homogenous scenes higher in order to get more objects in frame but you also run the risk of being forced to fly at an altitude where the quality isn’t acceptable. In my opinion 45m is about as low as I would map even a normal scene with a P4P. Are those blue Markers GCP’s? MTP’s?

If you don’t mind sharing I like to mess around with these kinds of datasets to see what can be done with photo editing and other photogrammetry software. Just PM me a link to download if you’re up for it. Please ZIP the files or allow download of the folder. :wink:

Blue markers are gcp’s.

I can’t really fly higher because I need a max GSD of 2cm/pix

Can’t really share pictures unfortunately but here is the resulting orthomosaic errors i got.

I assume this is due to the gaps in the tie point cloud.

Thanks a lot for your help

1 Like

No worries. Must be Area 51… :smile: Have you thought about trialing something else? I think you could probably get it quite a bit better but it’s going to take some time and patience to create the necessary MTP’s regardless of what software you use and unfortunately cloud processing solutions don’t have anything like that. More GCP’s would have helped. The problem with flying really low is that a lot of float can be introduced when there are a dense bunch of images in between GCP’s.