Firstly read:
http://forums.powershot-tx1.com/index.php?topic=12.0 - ask Cybercom on the TX-1 forum about his experiences - he may even give you his example footage using 4 TX-1s (stereo-squared) if you ask nicely.
Given that
http://stereo.jpn.org/eng/sdm/index.htm talks about normal stereo shots with less than 1ms difference in timing between left and right cameras, I suspect that the same technique could be used for getting the cameras properly synchronised starting video too. Cybercom said he was getting consistent 1 frame start differences on his videos between the left and right cameras - but that can probably be solved.
Lets presume that Canon have implemented the 30fps based on crystal clocks i.e. no drift in video timebase *except* due to crystal drift (testable - although very difficult IMHO - by taking videos under different conditions and measuring scatter of drift rates over time). Two crystals on any two TX-1s should have less than 100ppm difference between them, which over 5 minutes video works out to 3ms which is about 1/10th of a frame. You don't freefall¶chute for longer than 5 minutes eh?
If you really want to be a perfectionist then if you can find a friendly supplier with multiple TX-1s, you can very likely find two with very well matched crystal speeds - although measuring if they were well matched would take a bit of effort (and could drift out of sync over years).
You don't need CHDK to do this:
* Set up a reference subject you can video which allows you to measure frame to less than say 1/10th of a frame (e.g. well lit marked rotating disk with digital clock behind it?)
* Make sure all cameras are same temperature, same battery condition, same new SD cards.
* Take simultaneous video of subject with all cameras for say 15 minutes. Don't run them individually at any time, as crystal drift is temperature dependant, which is dependant on how long each camera has been running for, and the current room temp. More importantly this way your reference doesn't have to be accurate since you are only interested in measuring drift.
* measure time drift over time for each of the cameras. e.g. Take avi files and chop off start frames until all files start within 1 frame of each other, then split out all frames using
http://forums.powershot-tx1.com/index.php?topic=147.0 and then look for the cameras that best stay synchronised during the video.
* You would probably want to do more than one run and make sure everything is repeatable. It won't detect systematic errors (hopefully most can be ignored) but separate runs would detect random errors (due to unforseen factors!).
I hope that helps!
