I'm borrowing from snapchat here which uses a similar concept (tap for pics, hold for video) and tried to cut/avoid additional steps. As for focussing: why must focus and trigger be two distinct actions?
If you're thinking Snapchat and the video lasts until you release, the focus could perhaps be controlled by dragging over the area rather than tapping.
Overall, the focus isn't that big of a deal, but, to make as many people as possible happy, I would find a way to include as many features as possible. Flash is also a very important feature that wasn't accounted for. On top of that, how would the user go about toggling between the front and back facing cameras?
Front/back and Flash can be adjusted on the panel I the left.