Uncompressed video is typically stored in YUV colour space. A typical pixel format, YV16 stores 1 luma for each pixel, and one Cb/Cr for every 2 pixels, which does account for the two bytes per pixel you got :)
They're probably using some type of fast lossless or nearly lossless 'compression' that cuts the really low hanging fruit out but doesn't extract quality they care about.
HuffYUV, for example, if you're going to be targeting a 4:2:0 output stream; if you're going for 4:2:2 or even 4:4:4 for some super HD stuff you might use different options.