Problems while processing MicaSense RedEdge MX Dual data (10 bands) with WebODM

WebODM
#1

Hey, i’ve been trying to process this MicaSense RedEdge MX sample data using WebODM over the last few days.

The first problem I’ve run into is that ODM doesn’t seem to support 10 bands data as of now. So I had to remove some bands from the data to get it to work. It seems that it works just fine with 9 bands, but as ODM adds an own alpha band to the final orthophoto you are only able to process 8 own bands.

My questions regarding this are:

  1. Are there any plans for 10 bands support in the future?
  2. Is there a way I can prevent ODM from creating/adding the alpha band to the final orthophoto?

Removing 2 bands from the sample data and processing 8 bands using the default ‘multispectral’ settings worked pretty good though. But when looking at the final orthophoto I noticed some problems in the image.
35EKWkj
There are a lot of those “stiches” or “seams” in the orthophoto.

  1. What could cause those “seams” effects? Is that a problem with the bands alignment?
  2. Is there a way I could get rid of those effects (by using some setting parameter)?

Thanks in advance,
John

1 Like
#2

Welcome!

Oh, that’s an interesting discovery! I wonder indeed if we can easily raise the ceiling on band numbers or not.

Do you know if many consumer-grade sensors are exceeding 9/10 bands already?

Those seams could possibly be because bands removed had important data in those areas. What other parameters did you use?

#3

Thanks for you answer!

The MicaSense RedEdge MX Dual setup is definitely an outlier, most (if not all the other) multispectral drone sensors only capture 5-7 bands. So it’s not the norm, but I could see it become more popular in the next years.

The research project I’m working on doesn’t require 10 bands, so it works just fine if I remove the bands that I do not need. But it would obviously be nice to see ODM being able to handle 10 bands, if it’s not too much work.

That could have very well been the reason. I was able to get rid of the seams by changing the resolution of the orthophoto from 5 cm to 10 cm.

2 Likes
#4

Thank you for that insight. I agree, the pattern has been to increase spatial, radiometric, and spectral resolution, so I don’t expect that to change. I’m just curious how “early” this grade of sensor might be, and it seems to be a bit ahead of the curve.

Do you have a good estimate of the survey GSD? What happens at that resolution?