I was trying to familiarize myself with the Multiple Exposure mode on the Nikon D3, when I realized I really can't understand what it does (at the pixel level, not in terms of the resulting effect).
For those unfamiliar, according to the manual, the Nikon D3 allows to "record a series of two to ten exposures in a single photograph, [...] and produce results with colors noticeably better than photographs combined in an imaging application".
I tried to do some casual experimentation to see what happens when one uses 5 multiple exposures combined in camera compared to a single exposure for a static subject:

I took the resulting raw files (nef) and increased exposure in ACR in order to bring out the noise. Clearly, there are benefits in terms of noise. However, I don't understand what exactly happens to the raw files in camera. Beside some super-secret proprietary algorithm, does anyone have any clue as to how each image is combined? I wonder if this is equivalent to Guillermo Luijk's Zero Noise technique?