This is the first post in a series on the Sony alpha 7 Mark III, aka the a7RIII. I’ve received some complaints that it’s hard to navigate in this blog, what with its couple of thousand posts and many multi-part series. I’m trying something new. I’ve created a Category called a7RIII. You should be able to find all the posts about that camera in that category. I’ve also added a widget called Category List to the right sidebar, below the Articles widget. There’s a drop-down menu there that you should be able to use to get to all the posts in this series.
The a7RIII has a silent shutter mode. When running silent, the mechanical shutter is open and the sensor does all the work. The mechanical shutter is a rabbit, completing an exposure in faster than 1/250 second (for shutter speeds much faster than that, a slit moves across the sensor, and we’ll see in this post how long it takes to make its journey). The fully electronic shutter takes longer to do its job.
How much longer? Stay tuned.
I set up an analog oscilloscope with an input from a function generator. If you’re going to try this at home, don’t try it with a digital scope; it won’t work. I set the time base to 1 msec/division, and the amplitude of the generator to just over the maximum excursion for the scope’s vertical sensitivity. I mounted a Sony 90 mm f/2.8 lens on the camera. I put a generic RRS plate on the camera and clipped it into an Arca Swiss C1 head on a Gitzo tripod. I set the camera for the mechanical shutter, with silent shutter off. I set EFCS to on and set the file format to uncompressed raw, with the camera on manual exposure, the shutter mode single shot, and the shutter speed at 1/1000 second – that’s one msec.
I used RawDigger to demosaic the images, since ACR, Adobe DNG Converter, and Lightroom don’t yet support the a7RIII.
The mechanical shutter on the a7RIII, like the shutters on all the a7 series, travels from the top of the image to the bottom. Looking at that another way, it travels from the bottom of the camera to the top. The speed is not constant; if it were the diagonal green band would have straight edges. The shutter accelerates as it moves across the sensor. Both curtains accelerate at about the same rate, so that the exposure — the horizontal width of the green stripe — is about the same from the bottom to the top of the image. The width of the stripe is one division, or one millisecond. You will recall that I set the shutter speed for 1/1000 second. 1/1000 second is one millisecond.
If we project where the trailing edge of the green band intersects the top and the bottom of the frame onto the top and bottom of the oscilloscope graticule, and the horizontal distance in divisions between those dots will give us the shutter travel time. I put it at 3.4 divisions or the same number of milliseconds. That’s about 1/250 of a second, which is about average for mechanical shutters these days.
Now, let’s leave the scope set up the same way, switch the camera to silent shutter, and make another picture:
The electronic shutter travels in the same direction as the mechanical one, from the top of the image to the bottom, or the bottom of the camera to the top. You can see that the electronic shutter is much slower. To find out how much slower, we’ll have to change the scope’s time base. I set it to 10 milliseconds per division, and made another exposure:
I put the scan time at 7 divisions or 70 milliseconds. That’s about 1/15 of a second, or about the same speed as the a7RII silent shutter.
But if we set the file format to compressed raw, and the shutter mode to continuous (C), things move a bit faster:
That’s almost twice as fast, or about 1/30 second.
what?? that’s twice faster than mark ii, is this even possible?
Apparently so…
Sony claimed that the sensor is the same, so they were using only 50 percent of its peak performance in mark ii? Seems a little bit extravagant…
It does not appear to me to be the same sensor. ADC characteristics are different. Noise is different. Spectral sensitivity is different, if you can believe the compromise matrices in the EXIF data.
Hi Jim,
“But if we set the file format to compressed raw, and the shutter mode to continuous (C), things move a bit faster…”
Are we still 12 bits in continuous modes?
Best regards
Erik
Continuous is 14 bit when uncompressed RAW is used and 12 bit with compressed RAW.
Double performance WAS Sony’s claim when the camera was announced.
Unfortunately, Sony seems to have only done this for certain use cases – for others (such as video) there is no change from the R2 as far as I can tell – double performance should have allowed for full-sensor-width video sampling with no pixel skipping.
After all – it’s reading a full frame in 1/30 second, it should be able to read a 16:9 crop at 24fps with no issue – probably even 30fps although I’d accept if Sony had a mild crop at 30 like they do with A6300/A6500/A9.
In s35 mode, electronic shutter is faster? Thanks
It is in the a7RII and the D850. I would expect that it would be in the III as well.
Is elettronic shutter in apsc mode faster? Thank you.
It is in the a7RII, and probably is in the a7RIII. I’ll test it.
Interesting results. So in APS-C and 12bit compressed RAW mode electronic shutter may be around 1/60s fast enough for shooting slow/medium moving action without significant distortion.
Thanks for the test.
P.S. Will be nice if you will had the chance also for testing Olympus E-M1 II electronic shutter. They claim a 1/50s … 1/60s sensor readout speed.
Sorry, but don’t have access to Olympus gear.
Is it still reading out only 12 bits in compressed continuous? Maybe that’s where they made their gains.
Yes.
Jim, thanks again 🙂
I guess that the “compressed continuous” mode gives 12bit files, instead of the 13bits of “compressed_single” .. no ?. This would explain the 2X speedup 😉
What is the bitdepth of “uncompressed continuous” ?.
12.
Interesting findings, thanks Jim. I’m guessing the A7rII is also capable of 1/30 readout but Sony didn’t bother to run the configure the sensor/ADC to utilize it. Also, the fact the A7rIII can run at 1/30 in 12-bit (in combination with the faster imaging ASIC) tells me the A7rIII is capable of full-sensor readout for 4K video in FF mode , vs only in S35 mode on the A7rII. However, based on Sony’s recent segmentation decisions on the A9 I presume Sony didn’t take advantage of this for the A7rIII, to keep separation with the upcoming A7s refresh.
Have you done the same test with a9? What is the scan speed of a9 e-shutter?
http://blog.kasson.com/the-last-word/how-fast-is-the-sony-a9-electronic-shutter/
Curious, any test update on the A7r iii readout speed in APSC mode? Would be great to know, as its a key feature for me in thinking about upgrading (I work on film sets where I need to shoot silent mode all the time). Thanks again for all your tests – they help alot!
I suspect it will be half again as fast, like the a7RII:
http://blog.kasson.com/the-last-word/sony-a7rii-silent-shutter-speed-in-aps-c-mode/
Whether it’s half again as fast in continuous compressed will need verification. I’ll test it.
Yes, I had a think and then did some math. If it scans at the same speed in each mode, then it should take around 1/50 second to scan the shorter height of an apsc crop zone. Just multiply 30 milliseconds x apsc sensor height (16mm) / full frame sensor height (24mm) = 20 milliseconds or 1/50.
So what do images look like with silent shutter? Would a car driving down the street look blurred? Can you set 1/500 of a second on silent shutter? Or is everything shot at 1/30th of a second. I’m confused at how this works practically.
Fine, except for distortion from either subject or camera blur.
No, (assuming the shutter speed is fast enough) but there might be some distortion.
Sure. Just like with the mechanical shutter, which doesn’t traverse the focal plane in 1/500 second, either.
Just like with a mechanical focal plane shutter, speeds faster than the scanning time of the shutter are achieved by having a (virtual or real) slit that traverses the focal plane. As the shutter speeds get higher, the slit gets narrower.
I’m confused at how this works practically.
Read this: https://en.wikipedia.org/wiki/Focal-plane_shutter
Then look at this example of focal plane shutter distortion:
https://en.wikipedia.org/wiki/Focal-plane_shutter#/media/File:Bundesarchiv_Bild_183-1991-1209-503,_Autorennen_im_Grunewald,_Berlin.jpg
Or read this and look at the pictures of the fan:
https://www.premiumbeat.com/blog/how-camera-shutters-work/
Jim
Thanks Jim.
‘But if we set the file format to compressed raw, and the shutter mode to continuous (C), things move a bit faster’
Jim, what about uncompressed RAW and which continuous mode did you use (Hi+ ?)
Thank you
To get the highest frame rate, you need compressed raw and the fastest continuous mode. That’s what I tested.
I read this article and had hoped that using uncompressed and continuous mode might let me shoot at a higher shutter speed and still avoid banding with dimmed LED lights. I did a test and I got the same banding in both compressed and uncompressed modes when shooting continuous.
If the sensor readout is faster in uncompressed mode why is this not effecting the banding caused by a dimmed LED?
The sensor is not faster in uncompressed mode. It’s the opposite.
To get the highest scan rate, you need compressed raw and the fastest continuous mode. In that mode, the width of the bands with a LED should be wider.