3M vs 3E for mapping?

[jgraca]

I'm using a 3M for mapping and processing the projects in Pix4d Matic and/or Pix4d Cloud. The results have been mediocre at best. Many of the maps show large gaps, especially when flown on sunny days. Even on ideal flying days (cloudy with flat light) I'm still not getting great results. I'm using 80-85% overlap on both front and sides and flying near 400 feet. We map rivers. These are generally challenging sites with tall trees, lots of reflection off the water and strong shadows. That said, I'm finding that the results seem to be worse than maps generated using our old Phantom 4 Pro.

Is anyone else having issues with mapping using the 3M?

Does anyone have a strong opinion on 3M vs 3E for mapping? The camera specs appear to be the same but other threads suggest that the drones come from different DJI "departments."

 

[adm_geomatics]

It allegedly has the same RGB sensor as the M3E. What is "mediocre"?

 

[jgraca]

When shooting in bringt sun, around noon, in river corridors with large trees and shadows, there are often small or large patches of orthomosaic that are completely missing. One orthomosaic I processed in Matic had 90% of the ortho missing (see attached). Others have been better but still missing 10%+ of the orthomosaic.

It's hard to know if others orthomosaics from the Phantom 4 Pro were actually better because I have not yet compared the two drones side by side.

Are there other variables that might be different between the 3E and 3M RGB cameras?

 

[adm_geomatics]

When shooting in bringt sun, around noon, in river corridors with large trees and shadows, there are often small or large patches of orthomosaic that are completely missing. One orthomosaic I processed in Matic had 90% of the ortho missing (see attached). Others have been better but still missing 10%+ of the orthomosaic.

It's hard to know if others orthomosaics from the Phantom 4 Pro were actually better because I have not yet compared the two drones side by side.

Are there other variables that might be different between the 3E and 3M RGB cameras?

Trees and especially water are problematic. Even on a "windless" day there are almost no features that can be captured repeatedly so no tie points are created in aerial triangulation. Matic is for very large projects and speed so it's no wonder it struggled more. The only way I have been able to overcome this consistently is post-processing the photos for exposure correction to pick up points and use a photogrammetry software that allows for manual tie points. It's an arduous process of hunt and peck for alike pixels/features but I have saved many maps with it. Your results would have been the same with the M3E.

 

[jgraca]

I've tried manual tie points in Pix4D Matic on several projects but have seen very little improvement. Maybe I'm doing it wrong? What kind of success have you had with MTPs, in short?

So you've post processed photos to pull out features and then reprocessed - and had success?

 

[adm_geomatics]

When doing manual tie-points you need to ensure you have a minimum of 4 images/pixel if you expect noticeable improvement. There are always occasions where there's just nothing you can do about it but it has never not made a map better. If you have to you might just need to edit the ortho and clone appropriate content or use a photogrammetry software with mosaic editing.

As for post-process I don't have to do it near as much as I use to because cameras have gotten so much better but I immediately know when I hit the site what I am going to have to do later. You get a feel for it. I use ACDSee and have preset filters depending on the type of correction that needs to happen and it non-destructively batch processes. In these instances I have picked up as much as 10-15% more tie-points vs the natural images. Like I said I pretty much know it ahead of time so there is no reprocessing.