Awful test. It looks lke the shooter used the same lens but slightly different focal lengths, plus the tripod has moved in the second part (more to the right). Put two fingers on the escalators then fast forward to the 1DX test, you'll see the composition change a lot. Also count the window squares to the right - it'll become readily apparent.
A prime lens would have been a much better choice (e.g. a 50mm f/1.4 even @ f2,8) for a HD video test, rather than using a standard zoom shot wide open. Neither of these videos are at all sharp.
Also the colour varies, for instance, the 5D3 appears much more saturated (look at the red roof/top on model building down to the right, lower level).
The 1DX does seem to show a bit more detail - look at the window frames on the far wall (you can see most of the vertical & horizontal lines, not so on the 5D3), but that could be because of different camera position + it seems that the ambient light was brighter on the first test.
