I have a weird issue when running Reality Capture on Linux in Wine (which is a windows emulator). Reality Capture starts up properly and I can import images and get them visualized. When I start alignment, it seems that RC only finds features for a subset of the images. Consequently it creates a model with only these images.
If I use the same images on native Windows, I get features for all of the images and a model that uses all of the images.
Has anyone encountered something similar before? The root cause that I see seems to be that RC just doesn’t find features in these images - when run on Wine. I would be interested in learning if there are some known cases where RC does not detect features for some of the images.
In the attached image one can see that the selected image has 0 features, and only 3 of the visible images have features detected. This is the state after importing these images and clicking “Align Images”.
This happens on all of the datasets that I have tried.
