|
|
|
|
|||||||||||||||||||||||||||||||||||||||||
|
|
…make your good video even Gooder™ |
|
|||||||||||||||||||||||||||||||||||||||||
|
|
How SteadyHand Works: Step 3 Break the Camera-Motion into Components |
|
|
||||||||||||||||||||||||||||||||||||||||
|
Working with the complete camera motion that has been separated from the action in the scene, SteadyHand further breaks it down into 4 components:
Separating these requires a complex set of proprietary calculations combined with statistical and artificial intelligence techniques. It consumes the bulk of the CPU time when you run SteadyHand, but it is totally worth it. While visiting the lab in Germany, I saw a demo where this step was skipped. The result was clearly worse – it looked more like the simple corrections inside a camcorder. Don’t be surprised that SteadyHand does more than a camcorder – a camcorder doesn’t have the luxury of a powerful Pentium crunching at its leisure. The arrows in the images, below, show the components isolated in the analysis phase by SteadyHand.
Armed with these 4 measured camera movements (for each frame of the video), SteadyHand is ready for the next step – figuring out how the camera “should” have moved.
|
|
|||||||||||||||||||||||||||||||||||||||||
|
|
|
|
|
||||||||||||||||||||||||||||||||||||||||
|
|
|
|
|||||||||||||||||||||||||||||||||||||||||
|
|
|||||||||||||||||||||||||||||||||||||||||||
|
|
|
||||||||||||||||||||||||||||||||||||||||||