Under what circumstances will multiple folders be generated under sparse and dense directory (for example, 0, 1, 2, ..., etc.)

[hxboxy-xh]

I wonder why there could be multiple folders under sparse and dense (for example, 0, 1, 2, ..., etc.) Or is it possible to merge these using 'model_merger'

[ferreram]

Colmap will try to register all the provided images into a single model. However, if at some point it cannot register any new image into it while there are remaining unregistered images, it will start the reconstruction of a new model with these remaining images. As this can happen multiple times, you will have as many folders as the number of models created. You can try using the model_merger command to merge models together but they need to share at least 3 images in order to work if I remember correctly and this will most often not be the case.
If you want to avoid having multiple models, you should ensure that you have enough overlapping between the images in your dataset.

[hxboxy-xh]

but with the same images, use default options, Reconstruction -> Automatic reconstruction create 3 folders under 'sparse' folder(0, 1, 2) and Processing -> Feature extraction && Processing -> Feature matching && Reconstruction -> Start reconstruction create only one folder under 'sparse' folder(it 's name 0)

so why? @ferreram