For photos, the software takes the image and estimates the depth from each individual pixel. It then creates a left and right model, one for each eye. So, two new images are produced rather than showing you the original and a slightly different copy.
Video content is handled in a different way than images. Here the content is processed on a frame by frame basis rather than by individual pixels. This can be done in real time and the quality of the finished product depends upon the quality of the original file.
Read the full article here: http://www.theinquirer.net/inquirer/feature/2094839/lg-s-optimus-smartphone-2d-3d-conversion-technology-explained