MICASENSE Rededge Projects/Options/Results

Hello :slight_smile:
It’s been some few days i am testing WebODM.
With 3 smalls Rededge sets, i always have the same kind of pattern in the orthomosaic.
Below is an example
http://localhost:8000/public/task/712a84f4-c476-4fa0-bd68-053497e5c1f8/map/
image
Using Multispectral option, resizing at 2048, camera radiometric calibration ticked…
image
I join in the set calibrated reflectance panel pictures before and after the flight. Do WebODM is using this ? By the way there could be an option to setup panel albedo values. I read a post dated 6 months ago that reflectance panel might be added soon in the workflow.
Sorry i did not download the log report and don’t know how to share.
I would be glad to discuss about Rededge results and options to optimise ODM :slight_smile:
Thanks !

2 Likes

It’s not. There’s still some work to do to include support for reflectance panels… and camera vendors don’t really seem to agree to any standard way to do it, making a generic solution more difficult to implement.

We would love to get help (pull requests especially) with this.

3 Likes

@pierotofy thanks for your comments.
My question is now :
Do those non uniform patterns, that are artefacts, that are create during the process are the consequence of non radiometric calibrated process ?
Do Micasense REDEDGE sets are currently possible to process correctly or not ?
Also, may I guess that any other rendering option ticked would mitigate these artefacts ?
In this example, and others I have ran to compare, I have the same useless results
Hence, i can only tell that processing RGB imagery with WebODM is quite good, close to the commercial process when considering high res orthophoto.
However, there is lot of noise in the DSM result.
The community and developers are doing a great job. Keep going on :slightly_smiling_face::pray:

Yes. Perform radiometric correction before feeding the images to WebODM, then skip the radiometric calibration setting. Camera vendors often have scripts/programs to perform that.

Probably not; they might be the results of poor meshing results / noise in the DSM. Try increasing texturing-nadir-weight.

3 Likes

Great to know, i have heard about those scripts and i am trying to run them. I will keep you update :slight_smile:

4 Likes

image

I have increased to 30. The result is not better.
I have start to dig into available online Micasense Radiometric Calibration tutorials, with everything one need to do…It’s doable but that will take hours and days to understand each step because i need to understand Python scripts.
However there is a huge and nice tutorial available. I only know the basic parameters and it’s challenging, interesting, to understand what is behind the scene of the workflow dealing with multispectral images…
There is one thing i would like to notice ;
the band order from Micasense RedEdge images, from 1 to 5, is Blue, Green, Red, Rededge, NIR;
In WebODM workflow, i have noticed that the 4 and 5th band are swapped.
So, when we look at Plant Health, we need to choose the band order : BGRNRe, if not, the vegetation indices containing the 4 and 5th band are not correct.
I guess that depend on type of multispectral camera, and in the MICASENSE case, the Redege band is the 4th band, the NIR is the 5th :slight_smile:
If someone, and i have not any doubt, can help me in the comprehension of each Python scripts available from the Micasense Image Processing tutorial, it could be possible to use finally panel and radiometric correction.
Also, are you sure that band alignment is included in the ODM workflow ?
Feel free to comment,
Thanks !

3 Likes

For 10 days, succeded to use Micasense Image Processing tutorials for Alignements, Warp, re-insert the exif etc…
i have file names as xxxx_1.tif (+2,3,4,5th band) with geotags.
Now i run WebODM and it is not recognize as a multicamera (multispectral dataset )!
However, with raw multispectral pictures, WebODM runs correctly, excepted that above we get those weird flourescent green pattern.
I hope alignment made last days could enhance the ortho ?
But, how and why WebODM recognize an image set as a multispectral one in the end ??

1 Like