This 'HDR video' was obtained with two Canon 5D2 shooting in video mode with a 4 stops exposure gap. They managed to syncronize the sequences obtained and post processed them with some HDR tone mapping software to create an 'HDR video': .
What I see here is not any HDR video, and the reason is that the scenes being represented don't have a high dynamic range. Not at least for a Canon 5D2 with an output video format in mind.
In the beginning we can see sky and shadow areas between the buildings, a good amount of contrast, but still not enough to need overexposed shooting on a 5D2; shadows can be lifted without noise being a real problem. The guy talking is just a low dynamic range scene. Finally the panorama is a medium to low dynamic range scene.
What these guys claim to be an 'HDR video' is just in my opinion a tone mapped video. They could have saved the beam splitter and one of the Canon 5D2. Just shooting the highlights preserving sequence, then creating a replica of it overexposed by 4 stops, and mixing them in exactly the same way as they did with the two sequences, the result would have been the same.
It is a question of knowing what HDR is about and when extra shooting is really needed to capture the entire DR of a scene. Too bad that they didn't choose real high dynamic range scenes, that would have been a more interesting experiment IMO.
Regards
