Fig. 6.

Histogram of voxel densities of the porous aluminum test object. By setting the average occupancy of the charcoal test object to 1 and then multiplying by the refractive index of aluminum, the porous aluminum test object was inadvertently created with voxel densities exceeding the actual density of aluminum. This means that the test object was more strongly refracting than a true aluminum object would be, but this does not affect our measurement of the convergence or timing properties of the algorithms tested.