What is the actual process of reconstructing the ray directions from the microlens images?

anup-h

@RichardChen9 The input is the (u, v) of the intersection of the ray with the lens plane and the (x, y) of the ray's intersection with the image plane, so considering the distance from the lens to the image sensor is known, it should be fairly simple to interpolate these two 3D points.

moridin22

It seems like in this image though, there are a bunch of lenses between the (u,v) and (x,y) coordinates, so do you also need to compute how the ray refracts through all of those different surfaces?

What is the actual process of reconstructing the ray directions from the microlens images?

@RichardChen9 The input is the (u, v) of the intersection of the ray with the lens plane and the (x, y) of the ray's intersection with the image plane, so considering the distance from the lens to the image sensor is known, it should be fairly simple to interpolate these two 3D points.

It seems like in this image though, there are a bunch of lenses between the (u,v) and (x,y) coordinates, so do you also need to compute how the ray refracts through all of those different surfaces?