Bad Orthophoto Qhality

So, I’m new user in WebODM and i just made an orthophoto to test it and it was kinda bad, is that normal?
image

Wall structure is flickering, how can i solve this problem?

some informations:
drone: mavic 2 pro
height: 80 meters
overlap: 80%
speed: max possible in pix4d, something around 13 km/h

The Mavic 2 Pro has a rolling shutter which contributes to the distortion - ODM is not yet capable of compensating for the distortion caused by rolling shutters. The solution is to have the drone hover while it is taking the photograph. Not all flight planning software has this feature - I have been using Drone Harmony with great results.

3 Likes

thank u, I’ll try this, and reduce speed should help too?

You are welcome - no the drone must come to a complete stop while taking the photo. Theoretically you can fly at speeds at around 1-2 km/h without getting distortion as this speed would be below that actual readout time of the rolling shutter, but from all my practical experiments with the Mavic 2 Pro, photographing continuously at this speed is much slower than the “stop-and-hover” approach, as you can fly as fast as you want between the waypoints, which is very useful for large, high altitude mapping in particular.

3 Likes

hi @Michel_Lamana
you can tweek the parameters in WebODM,
I recommend the book The Missing Guide - https://odmbook.com/
secondly - plan the flight with cross pattern

2 Likes

Hi, I am a brand new WebODM user and am wondering if there is a planned date to support rolling shutter created images?

I’ve processed a 310 image cross hatch with a perimeter path(Mavic 2 Pro-rolling shutter) from 120ft elev 4 acre area with new building footings and formed concrete walls showing and a quick visual of the ortho seems to be quite good. I also did not apply gcps yet. I also have not yet overlaid the site plan. I just wanted to test what this would come out like. What sort of added distortion should one expect with rolling shutter images versus global shutter? Thank you very much for taking the time to read this and I hope that I have asked this in a correct location. If not, please let me know and my apologies. Cheers, Roger

Hi Roger,

Welcome. It looks like without rolling shutter correction, we are looking at 40-80 cm of error, at least anecdotally: Poor accuracy of Mavic imagery

This is a useful benchmark, but doesn’t address your core question about if there is a planned date.

Here’s the issue on github about it: https://github.com/OpenDroneMap/ODM/issues/313

But the short answer is that there is not yet a plan for fixing this. Project improvements come via three means:

In the third scenario, you can actually propose a feature to fund here:
fund.webodm.org/#/
fund.webodm.org/#/createfund

And then whip up support for the feature. I know there is a lot of interest, perhaps together we can put together the needed funding… .

1 Like

Thank you very much for the update Smathermather. Once I can confirm I am inputting my gcps in correctly and can overlay a site plan it will be interesting to see the variation between no gcp and w gcp file. I am of course still figuring out the list of settings as I am still unsure how the image exif georeferencing data gets used or doesn’t get used. Thanks again, it is great relearning this after being away from it for many years and to be able to ask questions that receive feedback is terrific.

1 Like

I wish I could be the one to help fund but that is not me unfortunately. :joy: :grinning: