The problem is evident with moving objects: they leave a "shadow" behind them. It's more or less the same effect of scrolling text on slower (~= cheaper) LCD TV/screens, that becomes unreadable.
With a colleague we tried to imagine the technological culprit, be it the slow CCD sensor or the videocamera processor or even the SD card speed.
Speaking of the FS200 only, it has three recording quality settings: LP, SP and XP. They are equivalent to 3, 6 and 9 Mbps. Is this the data rate from the CCD to the MPEG processor, or the top writing speed to the SD card? Probably it's the latter since the user's manual states that the FS200 uses VBR encoding.
In order to get closer to a plausible answer I need to create a relative speed between the camera and an object that is repeatible an infinite number of times. The speed must always be the same, so that it's possible to evaluate artifacts in the recorded output (on a CRT screen, if possible!).
Excluding all human being movements (non-repeatible speed) and those controlled by a human being (like a car), I came up with these constant-speed sources readily available at home:
- an R/C car at full steam after the initial acceleration distance, within few runs of fully charged batteries
- the swinging weight of a pendulum clock within a short timeframe (say, 1 hour or so)
The experiment will be as follows:
- mount the camera on a tripod and point it to the moving target path
- record the moving target at each recording quality settings
- change the SD card to a faster/slower one and repeat the three recordings
- compare on a fast screen
3 comments:
There's one more way to get an object move at the same speed several times: free fall! I am thinking of a light ball falling off a table.
Have you got a record player? If so, put a small object on the turntable and film it rotating -- it'll have pretty good long-term speed stability, and you'll probably have a choice of speeds.
Good one Adam! A record player is even harder to get than a pendulum clock in my opinion! Got to call parents...
Post a Comment