Sidebar: Understanding parallax
Excerpt from Sidebar: Understanding parallax, page 27:
...Before we get to camera specifics, let’s take a closer look at a fundamental concept: Parallax is a phenomenon that occurs whenever you observe something in three-dimensional space from two or more vantage points. Hold a finger in front of your face and close your left eye. Notice what is behind your finger in the background. Now, open your left eye and close the right. Your finger appears to move in relation to the background. That perceived shift is parallax. Our brains know how to interpret this optical discrepancy, and in fact this is one of the main ways we judge distances between objects.
Because the cameras in a 360˚ rig each view the world from different locations, objects appear to move in space depending on which camera’s view you look at. In this example, the letter carrier appears to be to the right of the tree in the left image and directly in front of the tree in the right image. This is because Lens A is seeing the action from further to the left.
Unfortunately, when stitching together a spherical video frame from the output of multiple cameras, we have to manually do what our brains do automatically.
The farther the camera sensors are from each other, the more discrepancy there will be in the cameras’ viewpoints. For this reason, a larger rig will inevitably create more parallax than a small one.
Shooting in 3D exacerbates the issue even more; stereoscopic spherical video displays two side-by-side images that deliberately include this parallax discrepancy. Depending on how and where the cameras are positioned, the stitching of the left and right video streams must be done differently (and with great precision) in order to preserve natural looking depth—not just when looking in a single direction, but also when turning your head...