From left to right in the first row, we show the input image, the initial segmentation (random colors), the depth-ordered layers (blue is farthest, red is nearest). Clicking on an image will display it in full size.
The second row shows the resulting parallax animation next to the anaglyph image.
This is example demonstrates a failure of our method: since the pots and the pans are hanging from the ceiling, their bottom coordinate is a poor indication of their depth. As a result, the depth layering fails to produce a correct depth ordering of the segment. Furthermore, no ground fitting is possible. Interestingly, flipping the image vertically, and applying our method (with the ceiling playing the role of the ground!) and then flipping back, yields a much more reasonable result.