“Alpha values carry over from the input images. Not only is the image pixel chosen, but also its alpha channel value. So, if a pixel is partially or totally transparent, the result of the Z-Combine will also be partially transparent; in which case the background image will show through the foreground (chosen) pixel.”
If Alpha is carried over should I not be able to see the bottom of the cube ?
it then explains another way to achieve the same effect with more nodes and the alpha over
Well, as you can see the transparency is there but depth buffers aren’t designed to work quite the way you’re thinking. They’re like booleans: on or off. The plane’s depth buffer is occluding or masking the depth buffer of the bottom half of the cube tthus rendering it invisible. If you want to see the bottom half of the cube then use an alpha over node and place the result of the z-combine in the bottom socket and the cube in the top.
I don’t know how other folks around here feel but I don’t like working with depth maps or z-combine because they really sort of suck. They do their jobs as well as can be expected but you’re much better off approaching compositing from a 2D perspective rather than 3D.
