The side by side images are different views from the two lenses of a 3d camera, if you look closely you’ll notice each of the images is slightly different. What YouTube does is combine the 2 images into 1 stereoscopic picture via their 3d player. The YT3D player outputs the picture in various 3d formats such as anaglyph (red/blue) or line polarized (for 3D polarized monitors that use REALD glasses). When you initially upload to YouTube their encoder takes a still image for the video which is going to be of the side by side images before they are processed by the player. If anything isn’t clear let me know 😉
