This is required to create the intended light field, but the light field approach isn't the only one, you can do simple deconvolution also. I wonder how well that would work.