September 17, 2018 at 4:07 pm #18186
Thomas SharplessIVRPA Member
- Forum Posts: 33
At the National Stereoscopic Association meeting, I learned a lot about synchronizing exposure on digital cameras. Mainly that you are pretty much at the mercy of the camera maker; and that to have any idea what is going on with your rig you have to make some fairly technical measurements.
There are two basic ways a cam’s s/w can detect a release signal or button push: polling and interrupt. Polling means it samples the signal at regular intervals, typically the refresh period of the display screen (often 16 msec). This means a response lag of anywhere from zero to the sampling period, half that on the average. With interrupt the lag varies less, typical range 1 – 4 msec. This is enough to stop “walking” motion effectively, but not fast motion.
With both methods there are occasional much longer lags, due to the fact that the software has disabled polling or interrupt, while it deals with some unusual situation.
One way to get the polling cycles of two cameras in phase is to power them on at the same time. They will then release at nearly the same time for at least a few seconds, sometimes minutes.
There are a couple of gadgets that work with cameras that use polling and whose firmware is hackable — that is, Canon SLRs — to get shutter lag down to the level of interrupt driven cams like the Sony Alphas. It is much trickier to make two cams, whose polling has drifted out of phase, fire at the same time, though I believe that has also been done.
The only way to do better is full genlock, where the 2 cameras effectively run as one. Not available on any still camera that I know of, but doable on some pro video cameras.
Or, shoot film with blade shutters, that can be synced to half a msec.
Rolling shutter effects are also very important. Almost all digicams have a rolling shutter, which means that the exposure happens at a different time on each row of the sensor. So if you place your cameras bottom to bottom, the images will only be in sync in the middle and will be out of sync by the whole readout time (on the order of 50 msec ) at top and bottom, even if shutter release sync is perfect.
There are several ways to measure exposure lag and sync. For lag, you have to start a timer when the release signal is issued. The best way to do that is to have your timing device issue the signal. Then you time how long it takes the camera to start the shot, for example by monitoring its flash trigger contact (or actual flash). For pixel sync you need to photograph some kind of timer display. A blitz flash can work for that; more elegant ways involve using an oscilloscope or CRT computer display. In any case you need to take lots of measurements under different conditions to really know what your cameras do.
The bottom line is that without genlock you can never count on any given shot being perfectly synchronized; so it is a good idea to take extra ones.
- You must be logged in to reply to this topic.