Good question! I vaguely remember that originally they needed to be handled differently, but could indeed be converged into a single class. When I refactor it to remove the QuickTime stuff and replace it with AVFoundation, I'll also tidy-up the front-end classes and do this.
I guess along the same lines, there isn't a Camera implementation for Linux, unless I missed it somehow. If I have to implement a Linux version of the camera functionality, I'll send it to you for inclusion.