Skip to content

3D model does not use RGB for multispectral images, only uses primary-band #1975

@maurerle

Description

@maurerle

First of all thanks for this great project!

Using a Micasense RedEdge-P multispectral camera, flying with an DJI M300 RTK, I can successfully create an orthophoto.

The alignment is best done using the https://github.com/brainergylab/micasense_imageprocessing library, but it also works to upload the whole folder of the micasense recording to ODM (with less good color quality).

The created 3D-Model however is derived from the primary_band setting, but even without this setting, and by only providing 3 channels (RGB instead of the 6 including NIR, RedEdge and Panchro) I could not create a colorful 3D-Model.

I suspect this is because of the reconstruction.multi_camera setting, which disables all images in get_photos_by_band.
As we aligned the other channels in a preprocessing, it should work to use all photos here instead.

Is there a setting to disable this behavior? Or any way to enable colorful 3D textured model calculation with OpenSFM?

The same applies to this dataset: https://community.opendronemap.org/t/micasense-red-gren-blue-near-ir-red-edge/3176

The multispectral testdataset is single-colored:

Image

vs Brighton Beach 3D model colorized:

Image

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions