Creating point clouds from image is surprisingly easy. Dots are translated on the z-axis using the original image’s brightness values. The resulting model is not a physically correct representation of the original image, yet it is still takes on expressive forms. In some cases, though, this technique can be used as an actual depth map—specifically forests do well, as tree tops are exposed to a lot of light, and trees become gradually darker the lower they get, as shadows are cast from surrounding trees. The code can be found in this repo.