0
\$\begingroup\$

I have a Hue light and want to measure the flicker frequency (due to the PWM) without buying an expensive photometer. Is it possible, to calculate it with such an Image, by counting the bars from the rolling shutter effect?

enter image description here

I've already done some work to gather further information:

  • Captured by a Samsung Galaxy S23, which got an Isocell GW3 image sensor
  • Sensor area 8.160x6.144 pixels à 1µm
  • 27 FPS @ 50 MP (but I think thats irrelevant for picture taking)
  • ShutterType: Electronic rolling and global reset

enter image description here

GRR (Global Reset Release Shutter)- diagram - https://docs.baslerweb.com/electronic-shutter-types

So theoretically, together with the given exposure time (i think it was 1/20000s here) and the information on how fast the sensor reads the individual rows, it should somehow be possible to determine the frequency with the top image, shouldn't it? Unfortunately, I can't find the Temporal Offset tRow [µs] () for this sensor anywhere.

PS: I'm a little confused, how rolling and global release shutter could create such an Image, shouldn't the black bars have a brightness gradient and not be completely black? That's more like a classical Rolling shutter effect.

\$\endgroup\$

1 Answer 1

1
\$\begingroup\$

If you know the sensor readout speed you only need to count the bright lines to get cycles/ms which can be easily converted to cycles/sec or Hz.

PS: I'm a little confused, how rolling and global release shutter could create such an Image, shouldn't the black bars have a brightness gradient and not be completely black? That's more like a classical Rolling shutter effect.

That is a rolling shutter effect. Global reset can only be used when flash is used; i.e. when the camera is set to flash. And the reset only affects the lines that received exposure; the light lines, not the black ones.

\$\endgroup\$
16
  • \$\begingroup\$ So I can just say: 20000/16 = 1250Hz? (Then I've made it much more complicated than it is lol) \$\endgroup\$
    – iwab
    Commented Jan 21 at 15:21
  • \$\begingroup\$ But it's strange. With a shutter speed of 1/20000 I can count 16 bright bars. With a shutter speed of 12000 I can also count 16 bright bars but they appear thinner and the dark ones get bigger. That would result in a very different frequency \$\endgroup\$
    – iwab
    Commented Jan 21 at 15:37
  • 1
    \$\begingroup\$ @iwab, I messed up my answer (edited now) The speed you need to know is the maximum sensor readout speed, which will be the same for all shutter speeds faster than that. \$\endgroup\$ Commented Jan 21 at 17:17
  • 1
    \$\begingroup\$ I would say so... the fastest sensor rolling readout speed I am aware of is ~ .004s. and the slowest I know of is .08s...whereas somewhere around .03s is much more common. Being that the phone sensor is very small, it is probable to be towards the faster end. \$\endgroup\$ Commented Jan 21 at 17:47
  • 1
    \$\begingroup\$ Cell phone cameras do all sorts of weird things without necessarily telling you... it's not uncommon for them to actually record multiple images and combine them as one. Or it might be pixel binning (combining pixels) for reduced noise at lower resolution. But yes, a higher readout speed is better if there is motion. \$\endgroup\$ Commented Jan 21 at 21:45

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.