I Recently started using Meshroom to photoscan buildings. While the raw Meshroom output is hardly ever usable directly (with my cheap camera anyway), does provide an extremely nice basis for modeling the building.
You can import the camera-reconstruction and the resulting pointcloud into blender and use that as as base for your model.
The nice thing is, you can even project the original photo's back onto your handcrafted low-poly model and you can get very nice results that way.
Meshroom outputs it's remapped images (with camera distortion compensation) in OpenExr format. I can open these and use them as camera backgrounmd images in Blender, and they show up just fine.
When I select 'apply camera image' when texture Painting nothing happens, the camera background images does not get projected on the model.
The strange thing is: if I convert the OpenExr image to JPG or PNG and use that as a camera background image it suddenly does work.
Does anybody know if this is a bug? Or is there something extra I need to tweak somwhere to get this to work? I'd very much like to skip the extra conversion step as all those images are taking a lot of HD space.
I did try various color space settings for the camere background image, but none have any affect on the projection (but the background images does get rendered differently depending on the colorspace).
I understand this is a highly specific question, but I hoped maybe someone else happens to use the same Meshroom->Blender workflow.