Is it possible to (slightly) change camcorder's frame-rate in order to synchronise it with another video signal (TV, another camcorder, etc.)?
As far as I know it cannot be done with consumer camcorders, since they haven't additional "genlock" inputs.
However, I have studied service manual of my TRV900E and found out that main oscillator's frequency can be (very) slightly modified by changing the voltage on one varicap connected to the oscillator. Since I had to replace two internal mikes and two resistors, I have decided to do all three tasks at once. Therefore, I have opened camcorder, slightly modified el. circuit and added three thin wires where appropriate [In fact, a colleague of mine (by the way, he's absolute genius) helped me in disassembling and made all the soldering, since I couldn't even see the resistor which had to be unsoldered :-].
The final result does not look very professional, but it is useful :-)
The camcorder has now additional inputs by which the camcorder's frame rate can be (slightly) modified. By modifying the control voltage from 0V to 4.5V, the frame-rate changed for about 0.007Hz (faster frame-rate at higher voltage). This is relatively small change, but it probably might be used for gen-locking two camcorders.
By manually changing the control voltage, it was possible to achieve the same frame rate as the TV signal (camcorder stayed in sync with TV (within one vertical line) for more than half hour without modifying the control voltage = 4.95V). The "black bar" on the TV screen viewed by camcorder's LCD screen remained at the same position.
If I succeed to make some kind of circuit to gen-lock two camcorders automatically, I'll let you know...
July & August 2004:
OK! I succeeded to "lock" video outputs automatically by using relatively simple circuit. Scheme is here (schemes added 27. August 2004; I promise that I'll draw the schemes on my PC as soon as possible - when I learn how to do that :-).
Both pictures were synchronised within about 100ns.
21. August, 2004:
Finally I've got the second camcorder (TRV900E, of course) and connected both camcorders to my sync circuit. A PC monitor was in front of both camcorders. In order to get faster synchronisation, the camcorders were powered-on by shortening both LANC signals with ground for some shorter time (see the scheme) after forcing them into "lock" mode independently with single LANC controller (from SONY tripod).
After sync has been achieved, the camcorders were pointing to the PC screen (1024x768, 75Hz refresh rate). The appropriate zoom has been chosen on both camcorders (simultaneously by using one IR remote), focus was set to manual and shutter speed to 1/1000s. Then recording has been simultaneously started by using (one) IR remote.
The result is the following (click pictures to get larger size) [note that right picture is from left camcorder and left picture is from the right one - use "cross-viewing"]:
Anaglyph version:
By the way, do you recognize Rob Crockett's page: http://pages.sbcglobal.net/rcrock/synctest/index.html ?
By using shutter speed of 1/10000s (that's right - 0.1ms!), the picture was the following [after adjusting "white balance" (white balance still isn't so good. Next time I should adjust white balance on camcorders before "shooting")]:
Anaglyph version:
So, according to pictures, the sync is better than it takes to "draw" one line on PC monitor = better than about 15 microseconds. How much better? Unfortunately I can't tell, since even 1/10000s shutter speed is relatively long for such tests. Probably both videos are synchronised at the same rate as video output signals (within about 100ns).
There are some short clips taken in the lab and at a fountain (all of them in "cross-view" form):
Fountain video clip [It has been taken with 1/10000s shutter speed. De-interlaced with TomsMoComp at 50fps, displayed at 25fps - slow-motion].
See other camcorder modifications...