Rolling Shutter and Shutter Speed / Is there a comparison

[HDNitehawk]

With the R3 coming on Wednesday I have been giving thought to the 30 fps and the effects of the Rolling Shutter.

My thought is that just like IS is described by the number of stops, the limitations of the electronic shutter could be described in the same manner.

My first thought was how was it reduced from the R5 to the R3, and my conclusion is that in part it was probably reduced by the speed of the processor clearing the sensor. The faster a sensor is cleared the less distortion that is introduced.

So this got me thinking, if you shoot something in motion the faster the shutter speed the less motion blur. The same should be true with rolling shutter distortion, the faster the sensor is cleared the less distortion due to a moving subject. Most of us who shoot much wildlife know what we need for shutter speed to get the results we want. The same would be true of most all sports. So my thought is if we know it takes 1/1000 fps to freeze the action of a particular subject, would the distortion be unnoticeable or acceptable? Would the maximum be less, say a subject that only requires 1/500?

[Kayaker72]

Sounds interesting and, absolutely, faster readout speeds will result in less distortion. I am wondering if we may not ever see a global shutter as readout speeds get to faster than 1/500th of a second, it will likely be good enough. Granted, some of us may always want a true global shutter.

In terms of rolling shutter, I do not think shutter speed will impact it. The readout speed of the sensor will stay the same regardless of shutter speed. For the R5, something like 1/60 of the second. For R3, something like 1/180th of a second.


Shutter speed can be very fast, but think of it in terms of how long each row is turned "on" to receive light. The top row (time 0 for the read out) might only be turned on for 1/1000th of a second, so the exposure is only 1/1,000th of second, but 1/60th of a second later (on the R5) the bottom row will be turned on for 1/1000th of a second.

An old article from DPReivew may say it better:

"However, although each part of the image is only made up from, say, 1/16,000th of a second, the slow shutter rate means each part of the image is made up from different 16,000ths of a second. Essentially, you're capturing the very short slices of time that your shutter speed dictates, but you're capturing many different slices of time. And, if your camera or subject have moved during that time then that distinction becomes apparent. This effect, where the final image is made up from different slices of time as you scan down it is known as the 'rolling shutter' effect."

[HDNitehawk]

I wasn't implying that shutter speed would have an effect on the rolling shutter distortion, what I was saying is that just like we can use shutter speed to give the appearance of a frozen in time subject, the faster the electronic shutter clears the distortion can be unnoticeable.

You mentioned 1/500th of a second, 1/60th and 180th of second. Do we know how fast the electronic shutter actually loads and clears the sensor?

I do not think a true global electronic shutter would be possible, but one so fast that there is no noticeable distortion in the image would be. When one that fast arrives it would be called a global shutter.

[DavidEccleston]

This effect, where the final image is made up from different slices of time as you scan down it is known as the 'rolling shutter' effect."

The part of this that is confusing is that we know from HSS requirements for the flash that for a shutter speed over 1/250s, the shutter open and shutter close overlap. You need multiple rapid flashes to illuminate the subject while the two shutter curtain moves and different parts of the sensor are receiving light... which means any shutter speed over 1/250s technically should have the same rolling shutter effect. With flash and low ambient light you might even expect to get multiple distinct chunks for each flash exposure as opposed to the continuous jello look you get from slow readout.

I'm not invested enough to bother doing any experiments though, and a quick google search didn't reveal anything interesting. All the links talking about shutter curtains are about 2nd curtain sync flash timing to get trails, not anyone talking about artifacts.

[Kayaker72]

You mentioned 1/500th of a second, 1/60th and 180th of second. Do we know how fast the electronic shutter actually loads and clears the sensor?

The 1/60th and 1/180th of a second is ballpark what I have heard people estimate as the readout speed when using electronic shutter for the R5 and R3. For the R3, here is DPReview estimating it at just under 1/200th of a second.
https://youtu.be/oYFLrRj8ODE?t=854

The part of this that is confusing is that we know from HSS requirements for the flash that for a shutter speed over 1/250s, the shutter open and shutter close overlap. You need multiple rapid flashes to illuminate the subject while the two shutter curtain moves and different parts of the sensor are receiving light... which means any shutter speed over 1/250s technically should have the same rolling shutter effect. With flash and low ambient light you might even expect to get multiple distinct chunks for each flash exposure as opposed to the continuous jello look you get from slow readout.

I'm not invested enough to bother doing any experiments though, and a quick google search didn't reveal anything interesting. All the links talking about shutter curtains are about 2nd curtain sync flash timing to get trails, not anyone talking about artifacts.

You are not wrong.

It seems electronic shutter disables flash options in the R5. I am traveling right now without my R5 so I cannot verify, but per R5 manual page 165 "In flash photography set (shutter mode) to option other than Electronic." Page 174 flash sync speeds are 1/250th for Electronic first curtain and 1/200th for mechanical. And, if you need more, Page 249: "Precautions when set to (Electronic):...AEB and flash photography are not available"

[HDNitehawk]

The 1/60th and 1/180th of a second is ballpark what I have heard people estimate as the readout speed when using electronic shutter for the R5 and R3. For the R3, here is DPReview estimating it at just under 1/200th of a second.
https://youtu.be/oYFLrRj8ODE?t=854

Doing a little cave man style testing I think those numbers may be right. Regardless that is still slow.

[Kayaker72]

Nothing wrong with a little cave man testing. After all, they gave us wheels, fire, and BBQ.


Sent from my iPhone using Tapatalk