I am calculating volumes on a large dataset and I have developed my own method for capturing volumes of complex areas. I wonder if there is a better way to find these volumes or a more accurate way that I haven’t thought of. I wonder if the community might have any thought on a similar situation. Here is the scenario (Idealized so I don’t have to draw pictures):

Imagine you have a large bowl with a rounded bottom, buried to the brim in sand, then you add marbles to the bowl (a random amount, but for these purposes lets say there is enough to completely obscure the bottom of the bowl, but not enough to obscure the sides). You then capture images with sufficient overlap to make a 3D model. You cant see underneath the bowl (its obscured by the sand).

How do you determine the volume the marbles?

How do you determine the volume of just the bowl?