I downloaded the youtube video and finally grabbed the frames I wanted.
I wish I could run the first image in slow motion so you can get a sense of the body language expressed by the cops that are reacting to the "Threat".
I'm working on that.
I know angles can be deceiving, however it looks to me that they see a threat on the passenger side of the car.
The second set of images, clearly shows a passenger.
Source video is here:
http://www.youtube.com/watch?v=NJm8Z5iqhe0
At ~00:07 is the first image.
At ~ 00:29 is the next set
Thanks for doing those caps.