Measuring rolling shutter: put a number on this issue!

(I read this and downloaded the clip, but haven't had time to run the numbers, and won't have it for at least a week more, maybe two)
 
Okay please note but these are 25p (I hope that's okay) from my RX100-V, which is supposed to be a noticeable improvement on the RX100-IV, so let's see if that's true...
4k, SteadyShot On: https://vimeo.com/222000789 (Note I'm not a Pro member so the original files can only be downloaded for a couple of weeks.)
4k, SteadyShot Off: https://vimeo.com/222001054
FHD, SteadyShot On: https://vimeo.com/222001319
FHD, SteadyShot Off: (To follow as I've used this week's upload allowance)
HFR250 (note 250fps conformed to 25fps; Quality priority; 1/250 sec !), SteadyShot On: (To follow)
 
Last edited:
My upload limit resets early Sunday, so they other two will be a while I'm afraid. (I can WeTransfer them if you'd like, just PM me an e-mail address to send them to.)
 
Dr_Jon said:
Okay please note but these are 25p (I hope that's okay) from my RX100-V, which is supposed to be a noticeable improvement on the RX100-IV, so let's see if that's true...
4k, SteadyShot On: https://vimeo.com/222000789 (Note I'm not a Pro member so the original files can only be downloaded for a couple of weeks.)
4k, SteadyShot Off: https://vimeo.com/222001054
FHD, SteadyShot On: https://vimeo.com/222001319
FHD, SteadyShot Off: (To follow as I've used this week's upload allowance)
HFR250 (note 250fps conformed to 25fps; Quality priority; 1/250 sec !), SteadyShot On: (To follow)
Click to expand...

I finally got to measuring this, and wow, is that a big improvement. It's gone from the bottom of the chart to a solid mid range. I could already see it was good by looking at the footage but the numbers are pretty impressive.

I don't see a measurable difference between steadyshot on or off, so I'm putting those together.

Thanks a lot!


RX100 V 250 fps ---- 3.8 ms (3.8-3.8)
RX100 V 1080p ----- 13.9 ms (14.4-14.0---14.2-13.1) (first 2 with stabilization, last 2 without)
RX100 V 4k -------- 17.5 ms (18.1-16.3-17.7---17.7-16.9-18.5) (first 3 with stabilization, last 3 without)

RX100 IV 1080p ---- 16.9 ms (16.5-17.3) (without stabilization it's slightly faster: 16.1)
RX100 IV 4K ------- 36.6 ms (36.4-36.7) (without stabilization it's slightly faster: 35.7)
 
Thanks, that's interesting. I assume as the 250fps is so much faster it's skipping a lot of pixels (of course it HAS to be less than 4ms or it would be in trouble). Although I should have said it's actually 1,824×1,026 scaled up to FHD so it is usually described as a mild sensor crop - actually the FoV seems identical to the standard FHD to me. Also it has two 250fps modes, the one I used is quality priority which should perhaps be mentioned (the other is speed priority and shoots at 1,824×616, but for nearly twice as long).

Ah, some maths says 1,824×1,026 is exactly 1 pixel in 3 horizontally and vertically (so 1 in 9 in total, although not necessarily evenly spaced due to the Bayer array). Also the rolling shutter is about a third, so I guess that's what happens.

Long version...

Quality Priority:

NTSC/PAL
240fps/250fps @ 1,824×1,026 (about 4 seconds recording) 1 pixel in 3 H and V
480fps/500fps @ 1,824×616 (about 3 secs) 1 pixel in 3 H, 1 in 5 V
960fps/1000fps @ 1,244×420 (about 3 secs)

Shoot Time Priority:

NTSC/PAL
240fps/250fps @ 1,824×616 (about 7 seconds recording)
480fps/500fps @ 1,292×436 (about 7/6 secs)
960fps/1000fps @ 912×308 (about 6 secs)

Note all are stored at FHD (by scaling) and can be saved at 24p/30p/60p (NTSC) or 25p/50p (PAL) - these set the degree of slowmotion, so 250fps shooting and 25p on card is 10x, etc.
 
Last edited:
Definitely welcome. In fact, I'd gladly stop running my measurements if somebody else ran and published accurate numbers for most cameras. Hopefully imaging-resource will test video mode too... (that post is only about stills).
 
They do say they have a video article in preparation in the text.

BTW I just wrote this somewhere else and thought worth a mention...
If you can read the sensor fast enough you can presumably simulate an ND filter for video, internally in the camera, by taking 2, 3 or more electronic-shutter exposures during your shutter open period and merging them, to get the correct blur.
E.g. the A9 can read the full sensor in 6.68ms so let's assume it can read the 16:9 crop in 5.64ms. A 24 fps 180 degree shutter exposure is 20.8ms so you could have 2 or 3 exposures of as short a time as you'd like and merge them together.
For example 3 stops (20.8ms/8 = 2.6ms):
Start first exposure. Wait 1.3ms from top line enabled and read out image.
Wait until 1.3ms before end of 20.8ms period and start second exposure, 1.3ms after top line enabled start reading data out.
Merge data to get more accurate blur.
There's 13.86 ms doing nothing in the middle, so could do a 3rd exposure then (or a 4th as only want 2.6ms active time total and the final read-out is after the end of the exposure period) , e.g. to help with flickering lights, which could be an issue.
 
The problem with that is that you're unlikely to have the fps you need for that. With the a9, even if you can read the full sensor in 7ms, you still can only do that 20 times per second (even though you can fit 150 "7ms" bits inside a second). Make it 16:9 and you can do 24 fps with 6ms shutter. But you'd need 72 fps in order to implement what you comment.

Either that, or you have to temporally store your analogue charge and add to it before reading its value on the ADC, but then you're talking about pixel design, and the total readout time is mostly irrelevant.
 
Curious about the Sony A9. Thank you so much for this resource!

From :
http://www.imaging-resource.com/new...the-sony-a9-rolling-shutter-is-pretty-minimal

"Our previous (and as yet unpublished; stay tuned!) rolling shutter testing concentrated on video recording, where we found typical values ranging from 1/40 - 1/75 second. We hadn't previously paid a lot of attention to still-shooting electronic shutters, as they didn't strike us as having very strong use cases. With the A9, though, the e-shutter is the only way to get to it's astonishing 20fps full-resolution continuous-shooting frame rates, and Sony has been touting the advantages of truly silent (electronic-shutter) shooting for noise-sensitive situations and truly candid people-photography."
 
Yes, the a9 definitely would be interesting to test.

Meanwhile, another interesting camera added to the list: the C200 clocks in at 16.1ms, both in 4k and 1080p, always at 25 fps.
 
Very interested if anyone has found info on the D850? Curious about performance in 4K full frame.
 
Just added the a7R III to the table. It's a very good performer: 4k FF is not bad (22.8 ms), 4k S35 is very good (16.0 ms, in line with the C200 and the BM 4.6k), and for those shots that need minimal RS, you can just swap to 1080p, which is super fast (7.7 ms in FF and 6.1 ms in S35).

For comparison, the a7S II was 30.4 ms in FF 4k mode. Nice to see these improvements rolling in.

Thanks for keeping the sample videos coming my way <3
 
You must log in or register to reply here.
Back
Top