The deep bit depth is needed only for the UV pass, and I'm sure it has to do with the sensitivity needed to replace one map with another, which is essentially what's going on with the PhotoInside process. The final output is a flash .flv file with visible compression. The whole issue is simply satisfying the requirement of their proprietary fly-render, map-tracking and replacing software. See? The UV pass bitmap that you map into your animation is a color/saturation gradient. The saftware needs to "see" this in very fine detail, in order to "know" how/where to map the user photo. The "how and where" can include not just movement but distortions, warping, exploding, reconstituting, etc etc that the animator has occuring to the UV map. 
Final formats acceptable to them are PNG and TIF. Again, only the UV pass needs to be 64 bit.
Only PhotoInside animations have these requirements. If you want to submit something not PhotoInside enabled, almost any old thing is god enuf, as long as it is video :)