Creation of mosaic dataset failed. Could not add output of type MosaicDataset

[WebinarSFCOLAB]

Hi,

I am new to Drone2map so I apologize if my question is too basic/badly explained or if it has been already answered (even if I haven't find any similar problems on the internet).

I am having trouble while processing, since the following error appears when i start the process:

[Product] Product generation failed. Could not add output of type OMWorkspaceMosaicDatase/MosaicDataset (source: C:\Users\doubl\Documents\Drone2map\Projects\2023_07_04\2023_07_04.gdb\Image_Collection)
[Processing] Create mosaic dataset failed
[Processing] Fatal Processing Error
[Misc] Processing Failed
[Processing] Create mosaic dataset failed
[Processing] Could not create mosaic dataset

I am using the default process options, you can find the template attached.

Does anyone have suggestions about what could be the problem?

Thank you in advance!!

[WebinarSFCOLAB]

I'd like to add that it does not seem to be caused by coarse overlap/overall coverage or vertical allignment issues, since I have no uncalibrated images and i have also tried modifying the Initial Image Scale value to 1 and increasing the Matching Neighborhood setting from small to large with no success. 

Useful informations:

• Version: Drone2Map 2023.1.1
• Licensing level: Drone2Map Advanced
• The images do not have real-time precision location (RTK) and no GPCs have been used for processing.

[MarkBarker1]

Hi @WebinarSFCOLAB,

Thank you for the information on the issue. Based off your pathing it looks like you are using OneDrive to store the project. This is not something we support at the moment and wouldn't recommend. Are you able to reproduce the problem without the use of OneDrive? If so, we can take a look into it and potentially provide some suggestions. 

Regards,

Mark

[WebinarSFCOLAB]

@MarkBarker1 I updated the post since I am not using OneDrive to store the project anymore.

[WebinarSFCOLAB]

@MarkBarker1 I think the problem deals with the fact that the camera stores each band as an individual image and I should assign each single band to a group name (version 2.3) or do a radiometric calibration to create a multiband orthomosaic. I though it was authomatic since I am using a MicaSense RedEdge but it is not.

[MarkBarker1 · Accepted solution]

Hi @WebinarSFCOLAB,

You are correct, it should be automatic that we detect what bands the camera is capable of. As long as you are using the Multispectral or thermal template when attempting to process the RedEdge dataset then that should be the case. When you load the images into the new project window you should either get a multispectral or thermal template option from the dropdown. If that is not happening, then Drone2Map is not recognizing the images as a multispectral sensor. This could be due to a lack of metadata, possibly the format the images are in (tiff, etc.). 

The workflow for running radiometric calibration would be as follows:
1. Add the images to a new project. The multispectral or thermal template should be the only options and one needs to be selected. 
2. Click the Flight data tab and click the Radiometric Calibration button.
3. Change the Quantity Type to Reflectance. The Panel Reflectance Factors section becomes active.
4. Click the Calculate button on the Blue band. A new Panel image editor window appears.
5. Use the browse button to add your panel image. It will appear in the view area once loaded.
6. Draw a sample area within the white sampling area of the panel. This does not need to be to the edges, generally you want it to be a smaller square in the center of the sampling area.
7. Input the Average Panel Reflectance value you receive from micasense. 
8. Click okay and you will return to the Panel Reflectance Factors list, but Blue should now have a Reflectance calculated. 
9. Repeat steps 4-8 for all the other bands. 
10. Click Ok to close the Radiometric Calibration window. 
11. Click Start on the Home tab to begin processing. 

The output orthomosaic will be corrected for surface reflectance. This can be used with raster functions we provide to generate various soil and vegetation indices.

So assuming the imagery is being recognized as multispectral, and you have gone through the steps above for the calibration, then you should not be seeing any errors. If that is the case, then I can take a look at the dataset on my end and attempt to run through to reproduce it. Let me know if you would like to do so and I will get you out a OneDrive link to share the data. 

Regards,
Mark

[WebinarSFCOLAB]

@MarkBarker1, first of all, thank you so much for your support.

The imagery is being recognized as multispectral and Drone2map atuomatically uses a multispectral template to process my dataset. Anyway, I think the problem deals with the lack of metadata, as you are saying, since the radiometric calibration fails when trying to compute the panel reflectance factor:

Do you have any other suggestion of what could be the problem? 

I can do a few more attempts before bothering you with the dataset 😅 but I really appreciate your help!!

[MarkBarker1]

Hi @WebinarSFCOLAB, 

We did actually see a bug recently for certain sensors where the panel images would not allow the calibration values to compute without first rotating the panel images 90 degrees. It may be worth a try in your case. The bug should now be fixed in our next release but its possible that's what you are running into. 

Please try rotating the panel images in windows file explorer and attempting to recalculate. Let me know if it works or if you are still running into the same problem. If it doesn't work then I can try to reproduce your issue on my end if you are willing to share the data. 

Regards,

Mark

[WebinarSFCOLAB]

Hi @MarkBarker1,
I have already tried rotating the panel image with no success...

I'm sharing with you the dataset via Google Drive, maybe we can proceed with the processing without doing the radiomatric calibration?

Thank you for your suggestions!

[WebinarSFCOLAB · Accepted solution]

Hi @MarkBarker1,The problem has been solved by: 

1.  Setting the Decimal Symbol to a dot in the region settings (my Windows OS region was set to Portuguese (Portugal) with the Decimal symbol set to comma). Once this is done the panel reflectance factor is calculated correctly.

2. Following your above-shared radiometric calibration workflow.

Thank you so much for your support!!

Best regards