Very small Orthomosaics

Hello,
I am having a strange problem with very small (44 × 336 pixels) orthomosaic being generated with WebODM 1.9.11 The source images were captured using my DGI Phantom 4 Pro v2 and are scaled to 2048 pixels on the long edge.

I have looked online and tried setting the following values:

dem-resolution (float) : 2.0
orthophoto-resolution (float > 0.0) : 2.0

Here is a sample of my outoput.

orthophoto (3)

Any help greatly appreciated.

2 Likes

Issue resolved. When exporting my images I whacked exif information.

3 Likes

Oof, yeah, that’s a nasty one! You’re kind of in uncharted waters in terms of reconstruction dimensions when your images lack EXIF, especially if you don’t add in GCPs to keep the reconstruction sane.

Glad you got it sorted!

Can we get a peek at the proper output? Looks like that survey came out nicely!

1 Like

Of course! Yea it came out great.

4 Likes

Sweet! Looking great!

Hello, I am experiencing same problem before you sorted it out. Could you please inform how to check if our EXIF information is not corrupted or damaged.
I have import my geotagged photos to QGIS and confirmed that coordinates were ok, but when i import it into webODM, many photos were ignored so the results just a small area.

This is QGIS overlay of my geotagged photos

Im using DJI Mavic Mini 1 drone with no GCP’s.
Thank you for your reply

And this is geojson that i downloaded from WebODM results

And this is orthophoto result from the above

It should be covered more area, but i dont have any clue what caused this.
Thank you for your reply.

Could you hand out the flight path? And flight configuration like speed and height? Have you stopped for your photos or was this a fly through?

1 Like

Thank you for your reply, I will try to obtain my last flight path. But as I can recall, i was flying automatically with about 5,5 m/s of speed and about 30 meter of height and that was a fly through.

1 Like

In this case you probably suffer from rolling shutter.
The DJI has a Readout time for the sensor, so if you are low on height and high on speed, the sensor will have a shift in pixels. That will happen to nearly all cmos Sensors

2 Likes

Thank you Maurice for your suggestions. I will try with another configuration.

1 Like

You are welcome, I suggest to stop at every picture.

2 Likes

Welcome!

Maurice gave you some great best-practices for mitigating survey/data collection issues!

In addition, you’re over a tight forest canopy and very close. You should significantly increase overlap/sidelap. It looks like something like 60% overlap/sidelap. That will not be sufficient, unfortunately.

1 Like

Yes Saijin, you are right. I was using 65% overlap for that pass (it was my first mission, though). :smiley:
Next time i will try with different configs as you guys suggested.

1 Like

BTW, that was a sugarcane field, but yeah, i will try to fly higher maybe 50 meters and increase the overlaps. Thanks for your suggestions

1 Like

You don’t need To fly Higher when you stop for ever picture and increase overlap.

1 Like

Flying higher can enhance feature detection and matching, and can also reduce the cost (airtime) of a survey. You often do not need leaf-level GSD for things like this, so a higher flight makes things much more simple.

Ok guys, thanks for your advices. Next time will surely try another configuration on that location. Meanwhile, on another field we were succesfully create orthophoto with ODM
[DJI Mavic Mini 1; Height at about 30 mtrs; Speed at 5,5 m/s; Overlap 75% for both sides]

1 Like

You are right, but it will also increase the cm /px and therefore the resolution and it will not be good as. But the question is always: what’s the target of the mission?

1 Like