1

Is there a way how to measure the "severity" of the rolling shutter (RS) effect for a given image sensor? By severity I mean how slow the image read-out is done for every single frame. The slower the read-out is the biger image deformation for motion objects one gets.

I guess such a method would require to take a picture of something that can be measured, probably a vertical line that would move quickly horizontally at a known speed. I would appreciate practical ideas on how to construct such a measurement.

Taking a photo of a strictly vertical object from a passenger car whose speed is measured with GPS could be one of the solutions. However I guess a distance from camera to the measured object would be needed and this may be measured with a stereo camera. Getting a working setup like this may be a bit tricky though. Better solution would be for a lab environment.

I noticed some RS sensors perform decently well (almost like global shutters) while others were really poor... I mean for a certain speed of motion RS might behave like global shutter. Do frame exposure time or frames per second (video) settings play any role in the severity of the effect? It is to be noted that RS is typical for CMOS sensors while GS for CCD sensors.

closed as off-topic by TFuto, Michael C, mattdm, MikeW, AJ Henderson Sep 29 '14 at 5:14

This question appears to be off-topic. The users who voted to close gave this specific reason:

If this question can be reworded to fit the rules in the help center, please edit the question.

  • Is there a photography problem you're trying to solve? – mattdm Sep 27 '14 at 22:56
  • Yes I want to take photo and video footage from a moving passenger vehicle with cameras pointed to all sides - front, rear and also sides where most motion will be so RS effect will be worst. If i test sensors for RS i may not need to spend more on global shutter camera. – Kozuch Sep 28 '14 at 6:17
  • Found a related post here: joancharmant.com/blog/… – Kozuch Jul 20 '18 at 10:11
2

I would print a straight vertical line and attach it to a turntable set to 45rpm. IF you set up the camera so that the frame covers the centre to the edge of where the record goes you can then measure the effect by the bend in the straight line, and compare different cameras even if the focal length changes.

  • Capturing rotation is a good idea for a lab setup. How will I exactly measure the bend? I need to start the capture when the line is exactly vertical, right? If so a video mode could be used (say 30 FPS) to catch the right frame. Also I guess a good performing (fast) RS sensor may need more RPM to show the bend at all. – Kozuch Sep 27 '14 at 10:27
  • @Kozuch keep in mind that a rolling shutter is only a problem with video when the camera has a mechanical shutter. – Håkon K. Olafsen Sep 27 '14 at 11:13
  • @Håkon K. Olafsen: I think this is not true. Small P&S cameras suffer from RS effect in video and they certainly do not use mechanical shutter for video. RS simply is reading one pixel after another during exposure which takes some time and creates image deformation of objects that are in motion. – Kozuch Sep 27 '14 at 11:38
  • Missing a comma in my sentence. What I tried to say was: If the camera have a mechanical shutter, rolling shutter will only be a problem when recording video, not taking photos. – Håkon K. Olafsen Sep 27 '14 at 11:42
  • 1
    Any shutter speed faster than the shutter transit time (the time it takes the curtains to move from one side of the frame to the other) will cause rolling shutter effects. The greater the difference between the transit time and the shutter speed, the more pronounced the effect. Most current DLSRs have a flash sync speed of around 1/200 second, which indicates a likely transit time of around 1/250-1/320 second. Assume 1/300 second. If you use a Tv of 1/2400 second only 1/8 of the frame is uncovered at any one moment. 1/8000 second means only about 1/30 of the sensor is uncovered at once. – Michael C Sep 27 '14 at 16:01

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