Stereoscopy
We see the world in 3D because we have two eyes that are spaced slightly apart. These two views are processed in our brain to create a single three dimensional view, and this is how we perceive our world and judge depth.
Filming in 3D is all about providing the brain, via our two eyes, with the two slightly spaced apart views of a scene, that it needs to construct a single 3D view. So stereoscopy is the art of filming with two (stereo) cameras in such a way that the brain can fuse these two slightly different views.
The slight variation between these two views is known as (Parallax) and there are two main variables that we use to manage the 3D effect: interaxial distance and convergence.
Interaxial (or interocular) distance is the distance between the two cameras. While you would think that the perfect distance between the cameras would always be equal to the actual distance between your eyes, that is not the case. You see our eyes don't zoom in and out like a cameras lens can, and therefore we have to adjust the interaxial distance between the cameras as the camera lenses zoom in and out. This is further complicated by the focal distance. So the interaxial distance will be much smaller if the camera is zoomed in on a subject that is close to the camera than it would be if the subject was far away.
The convergence point is the point at which the two camera's views meet. That means that the two cameras are never parallel to each other, they are always slightly aiming towards each other. The point of convergence is essentially the screen. Therefore if the point of convergence is behind the subject then the subject will appear to come out of the screen (a common effect in 3D films), and if the subject is behind the point of convergence it will appear to the viewer that it is 'in' or behind the screen.
So when shooting in stereo, the camera person not only has to worry about the exposure, focus and composition, but now also has to consider the parallax, interaxial distance and point of convergence. Clearly this is just way to much geometry for any 'normal' person to calculate on the fly, all the while driving around the African savannah in search of wild animals. So WE commissioned the creation of a 3D camera rig that would calculate the interaxial distance and convergence point based on the focal distance and zoom that the cameraman chose. Of course the cameraman can change the parallax and therefore creatively adjust these parameters via the onboard computer, but in the heat of the moment the computer will do the mathematical heavy lifting.
Coming off the safari vehicle is a side-by-side "frame compatible" 25 frame per second HD video stream. This is a strange video format which is basically two video streams (left eye and right eye) that have been joined together and squeezed into one normal sized, and shaped, video stream (frame compatible).
This video feed is sent from Djuma Game Reserve to our cloud based Master Control server in Ireland via Mumbai in India on the Bit Gravity CDN (Content Delivery Network). The Master Control archives the stream at full HD and splits off the left eye, stretches it back to its original shape, archives it, and then translates the stream for iPhone, web and TV. The Master Control also outputs a scheduled TV channel in full 3DHD using the LIVE incoming feed/s and the archived safaris, which it reinserts into the BitGravity CDN for global delivery to our TV clients.