Estimating the cameras synchronization can be tricky because the values we are trying to measure usually in the millisecond range.
Traditional clocks (including those who display "milliseconds") will typically have a refresh rate of 30 to 60 Hz and we need to be able measure time differences that can be less than 1 millisecond (ie 1000 Hz).
That is why we created a custom high frequency timer using leds.
That timer has 3 rows of 10 leds that can increment at any frequency between 1Hz and 10 KHz. The frequency and duty cycle can be set using 4 push buttons below the display.
To measure the camera sync, we take a picture of the timer and compute the relative timestamp on each image based on the leds that are illuminated.
In the example below, we set the frequency to 1000Hz, meaning:
- first row will increment every millisecond
- second row increments every 10 milliseconds
- last row increments every 100 milliseconds
In order to take precise measurement, the cameras should be set with an exposure time which is less than the refresh rate of the timer. In this case, an exposure time of 1/1000 was used. If the exposure time is too high, several leds will be illuminated on the first row and measuring will be tricky...
NB. You might need to adjust the ISO and aperture to make sure you can still determine the column number of the led that is illuminated.
Once the picture is captured, you will need to compute the relative timestamp of each image by adding the times from each row (see example below).
To get the synchronization you will simply need to compare the extreme values of the timestamps for each image.