FMX 2013 - Camera Physics
Date: 24.Apr 2013 | Reading Time: 2 minutes, 10 seconds
This talk was very theory heavy with lots of formulas and photos of curves that summarize pretty badly. I still gave it a try here.
The session started with the history of camera tech. From the first wooden boxes with a hand crank.
A hand cranked camera had severe restrictions in that it obviously only could film where a human operator could go and there wasn’t even adjustable focus. Since then we have come a long way. Huge VistaVision camera rigs, car cranes with two seat at the top end of the crane (Titan crane) and Technocranes, which was the first camera crane that didn’t need someone to look through the lens allowing much greater range of camera freedom. This was thanks to Jerry Lewis idea of a video feed.
After that came the motion control rigs, which not only allow repeatable motion to shot many matching passes of a shot, but also allow the combination of live footage with camera matched 3D footage or miniature footage.
Setting The Scene
- it’s imperative to set up your film back accurately
- after that setting the focal length should give you the correct field of view
- you need to set up your nodal point on set correctly
- live action usually is not nodal, meaning there is a parallax while panning
- this must be matched in the 3D camera
Real Life Camera Motion In 3D
Cameras must adhere to physics, which means there is a limit to the acceleration of an object. Not speed. Acceleration. Meaning sharp changes in speed, up or down, directional changes.
The acceleration can be derived by the second order derivative of the position change (translation curve). Ideally, you want your 3D curves to accelerate gradually with no more the 9.82 m/s or 1G. Some motion control sYstems can handle more or less.
Of course, there is also a certain speed limit involved.
Problems Capturing Live Motion For Repeatability
A problem arises when you have a camera motion, usually SteadyCam shots, that you need to transform into a motion control shot. For example, you film your actors with a SteadyCam in a greenscreen setup and then you need to repeat that move with a motion control rig on a different set or a miniature.
The way to go about it is to match move (track) the shot, which will give you a rather noisy result, at least in motion control terms, which you won’t be able to program. You can smooth the result, which will loose you accuracy, but enables you to program the shot. The trick is to filter enough, without leading to misalignment.