Sony a7R, a7RII shutter shock

The mechanical shutter on the Sony a7RII is supposed to be better damped. It also has electronic first-curtain shutter (EFCS). How do those things make moderate telephoto pictures sharper?

The protocol:

• The cameras: the Sony a7RII and Sony a7R.
• The lens: the Leica 180mm f/3.4 Apo-Telyt-M
• The filter: Heliopan 77mm variable neutral density.
• The lighting: a single Fotodiox LED-200WA-56 daylight balanced flood.
• ISO set to 800,
• Manual focusing wide open, then lens set to f/5.6.
• Drive set to single
• Camera on an RRS TVC-43 with an Arca Swiss C1 head, RRS L brackets in landscape mode.
• Exposure protocol: ND filter to minimum attenuation,, shutter to 1/500, make an exposures, ND filter down a third of a stop, shutter down a third of a stop, make another exposures… until you get to 1/10 second.
• Make a series with both cameras, EFCS on and off in the case of the a7RII.
• Develop in Lightroom 5.7.1 with standard settings.
• Crop, export as TIFFs, analyze for horizontal edge (that's the edge perpendicular to the way the shutter travels) MTF50 in Imatest.
• Export the results to Excel and graph.

Landscape orientation makes it easier on the mechanical first curtain cases, since the tripod is stiffer in that orientation. Mounting the camera body directly to the tripod also makes it easier on those cases.

The results:



The blue line is the a7R. It does the worst even at high shutter speeds by virtue of its slightly lower-resolution sensor. The orange line is the a7RII with both curtains operated by the mechanical shutter. I still exhibits shutter shock, which is not a surprise. The grey curve is the a7RII with EFCS on. It looks pretty darned good.

Comments and questions appreciated. More focal lengths to come, if there's interest.

OBTW, even 1300 cy/ph is pretty sharp.

Jim

--
http://blog.kasson.com

JimKasson said:

The mechanical shutter on the Sony a7RII is supposed to be better damped. It also has electronic first-curtain shutter (EFCS). How do those things make moderate telephoto pictures sharper?

The protocol:

• The cameras: the Sony a7RII and Sony a7R.
• The lens: the Leica 180mm f/3.4 Apo-Telyt-M
• The filter: Heliopan 77mm variable neutral density.
• The lighting: a single Fotodiox LED-200WA-56 daylight balanced flood.
• ISO set to 800,
• Manual focusing wide open, then lens set to f/5.6.
• Drive set to single
• Camera on an RRS TVC-43 with an Arca Swiss C1 head, RRS L brackets in landscape mode.
• Exposure protocol: ND filter to minimum attenuation,, shutter to 1/500, make an exposures, ND filter down a third of a stop, shutter down a third of a stop, make another exposures… until you get to 1/10 second.
• Make a series with both cameras, EFCS on and off in the case of the a7RII.
• Develop in Lightroom 5.7.1 with standard settings.
• Crop, export as TIFFs, analyze for horizontal edge (that's the edge perpendicular to the way the shutter travels) MTF50 in Imatest.
• Export the results to Excel and graph.

Landscape orientation makes it easier on the mechanical first curtain cases, since the tripod is stiffer in that orientation. Mounting the camera body directly to the tripod also makes it easier on those cases.

The results:



The blue line is the a7R. It does the worst even at high shutter speeds by virtue of its slightly lower-resolution sensor. The orange line is the a7RII with both curtains operated by the mechanical shutter. I still exhibits shutter shock, which is not a surprise. The grey curve is the a7RII with EFCS on. It looks pretty darned good.

Comments and questions appreciated. More focal lengths to come, if there's interest.

OBTW, even 1300 cy/ph is pretty sharp.

Jim

--
http://blog.kasson.com

Click to expand...

Thank you so much for your contributions to these forums. You are a true diamond in the rough.

--

Uh, that's a Leica R series lens.

Sorry,

Jim

Jim, was hoping you'd get to the shutter vibration tests soon I'm very interested in seeing what the shutter vibration is for continuous EFCS shooting mode and how the fully-electronic shutter compares, with the real-world usage case being bracketed exposures for landscapes.

For example, on the 5DSR I found that performing a bracketed exposure in continuous mode causes vibrations on the 2nd, 3rd, etc.. exposures even with the camera in LV and with Canon's EFCS turned on (test here). The reason for the vibration is that the camera doesn't insert any delays in between exposures of the bracketed sequence, which means only the first exposure benefits from the EFCS (ie, after the first exposure, the 2nd curtain closes, the 1st curtain opens, and the second exposure starts right away, thus no settle delay and effectively no EFCS benefit). The only way to work around this on the 5DSR is to use a non-continuous mode and manually trigger each exposure of the bracket, which provides enough time for the shutter to settle between exposures and thus benefit from the EFCS. The problem with this solution is that it introduces scene-motion risk between exposures, such as the clouds, sea/lake, etc..

So for the A7rII the first experiment I'd love to see is what delay if any the camera inserts between exposures during a bracketed sequence performed in EFCS continuous shooting mode. Assuming the camera doesn't insert any delay or an insufficient delay (which in my testing on the 5DSR would be any delay < ~500ms), the next experiment would be that same continuous bracketed sequence but with the fully-electronic shutter, and then compare the shake/MTF between the two sets.

My guess is that continuous shooting with the fully-electronic shutter will yield the best bracketing solution on the A7rII. The read-noise/DR penalty from using the electronic shutter/continuous shooting vs EFCS will be immaterial since the bracketed exposures will be merged in post and provide multi-EV benefit that will allow shadows to be pushed with no noise anyway.

JimKasson said:

The mechanical shutter on the Sony a7RII is supposed to be better damped. It also has electronic first-curtain shutter (EFCS). How do those things make moderate telephoto pictures sharper?

The protocol:

• The cameras: the Sony a7RII and Sony a7R.
• The lens: the Leica 180mm f/3.4 Apo-Telyt-M
• The filter: Heliopan 77mm variable neutral density.
• The lighting: a single Fotodiox LED-200WA-56 daylight balanced flood.
• ISO set to 800,
• Manual focusing wide open, then lens set to f/5.6.
• Drive set to single
• Camera on an RRS TVC-43 with an Arca Swiss C1 head, RRS L brackets in landscape mode.
• Exposure protocol: ND filter to minimum attenuation,, shutter to 1/500, make an exposures, ND filter down a third of a stop, shutter down a third of a stop, make another exposures… until you get to 1/10 second.
• Make a series with both cameras, EFCS on and off in the case of the a7RII.
• Develop in Lightroom 5.7.1 with standard settings.
• Crop, export as TIFFs, analyze for horizontal edge (that's the edge perpendicular to the way the shutter travels) MTF50 in Imatest.
• Export the results to Excel and graph.

Landscape orientation makes it easier on the mechanical first curtain cases, since the tripod is stiffer in that orientation. Mounting the camera body directly to the tripod also makes it easier on those cases.

The results:



The blue line is the a7R. It does the worst even at high shutter speeds by virtue of its slightly lower-resolution sensor. The orange line is the a7RII with both curtains operated by the mechanical shutter. I still exhibits shutter shock, which is not a surprise. The grey curve is the a7RII with EFCS on. It looks pretty darned good.

Comments and questions appreciated. More focal lengths to come, if there's interest.

OBTW, even 1300 cy/ph is pretty sharp.

Jim

--
http://blog.kasson.com

Click to expand...

I always knew there was a reason I wanted EFCS. ;-) That, & the silent shutter, is a real benefit partially from the copper BS wiring. Thank you for the graph.

This is great news. Your efforts are much appreciated.

Many thanks Jim. Repeating those measurements in portrait orientation would be illuminating.

Also, I noticed the most severe shutter shock on the A7R with APO-R 280/4 or Vario-R 105-280/4.2 in portrait orientation. Any chance you could measure those lenses on A7R and A7RII? TIA.

JimKasson said:

The mechanical shutter on the Sony a7RII is supposed to be better damped. It also has electronic first-curtain shutter (EFCS). How do those things make moderate telephoto pictures sharper? ...

Click to expand...

Very cool. I suppose you could also test with silent (fully electronic) shutter -- I guess that would give a baseline for ambient vibration?

I assume all A7RII tests were with IBIS off? How about on?

When I got my A7II, I was actually expecting that I might see some shock with IBIS off that I hadn't seen with my A7 -- because the sensor in the A7II is not on a very rigid mount. I haven't, but I never explicitly tested for this either. If anything, my impression is that the A7II is a touch more stable, and I credit the body design. Have you ever tested the A7 and A7II? It would be interesting to see if the body design for the II versions is a significant part of the difference you're seeing between the A7R and A7RII or if it really is all due to the new shutter... (which the A7II doesn't share).

JimKasson said:

OBTW, even 1300 cy/ph is pretty sharp.

Click to expand...

Which is probably why we don't notice much of this in real-world examples from the A7/A7II even with EFCS disabled. ;-)

Well I would be interested to know if there are circumstances when the silent shutter should be used (other than when you want silence). As in perhaps high speed continuous shooting or bracketing on a tripod.

JimKasson said:

The mechanical shutter on the Sony a7RII is supposed to be better damped. It also has electronic first-curtain shutter (EFCS). How do those things make moderate telephoto pictures sharper?

The protocol:

• The cameras: the Sony a7RII and Sony a7R.
• The lens: the Leica 180mm f/3.4 Apo-Telyt-M
• The filter: Heliopan 77mm variable neutral density.
• The lighting: a single Fotodiox LED-200WA-56 daylight balanced flood.
• ISO set to 800,
• Manual focusing wide open, then lens set to f/5.6.
• Drive set to single
• Camera on an RRS TVC-43 with an Arca Swiss C1 head, RRS L brackets in landscape mode.
• Exposure protocol: ND filter to minimum attenuation,, shutter to 1/500, make an exposures, ND filter down a third of a stop, shutter down a third of a stop, make another exposures… until you get to 1/10 second.
• Make a series with both cameras, EFCS on and off in the case of the a7RII.
• Develop in Lightroom 5.7.1 with standard settings.
• Crop, export as TIFFs, analyze for horizontal edge (that's the edge perpendicular to the way the shutter travels) MTF50 in Imatest.
• Export the results to Excel and graph.

Landscape orientation makes it easier on the mechanical first curtain cases, since the tripod is stiffer in that orientation. Mounting the camera body directly to the tripod also makes it easier on those cases.

The results:



The blue line is the a7R. It does the worst even at high shutter speeds by virtue of its slightly lower-resolution sensor. The orange line is the a7RII with both curtains operated by the mechanical shutter. I still exhibits shutter shock, which is not a surprise. The grey curve is the a7RII with EFCS on. It looks pretty darned good.

Comments and questions appreciated. More focal lengths to come, if there's interest.

OBTW, even 1300 cy/ph is pretty sharp.

Jim

Click to expand...

Any thoughts about adding a line for the silent shutter? I'm surprised no one has characterized it yet.

ProfHankD said:

JimKasson said:

The mechanical shutter on the Sony a7RII is supposed to be better damped. It also has electronic first-curtain shutter (EFCS). How do those things make moderate telephoto pictures sharper? ...

Click to expand...

Very cool. I suppose you could also test with silent (fully electronic) shutter -- I guess that would give a baseline for ambient vibration?

Click to expand...

Yes, but also would drop a bit and add some RN. I will get around to trying it, though. From my tests of the Nikon D810 with EFCS vs trailing curtain studio strobe illumination, there's not a lot to be gained from omitting the trailing curtain acceleration.

ProfHankD said:

I assume all A7RII tests were with IBIS off? How about on?

Click to expand...

The camera was on a tripod. I've not found that IBIS helps with the camera on a tripod, and it often hurts:

ProfHankD said:

When I got my A7II, I was actually expecting that I might see some shock with IBIS off that I hadn't seen with my A7 -- because the sensor in the A7II is not on a very rigid mount. I haven't, but I never explicitly tested for this either. If anything, my impression is that the A7II is a touch more stable, and I credit the body design. Have you ever tested the A7 and A7II?

Click to expand...

Yes. See here:

ProfHankD said:

It would be interesting to see if the body design for the II versions is a significant part of the difference you're seeing between the A7R and A7RII or if it really is all due to the new shutter... (which the A7II doesn't share).

ProfHankD said:

OBTW, even 1300 cy/ph is pretty sharp.

Click to expand...

Which is probably why we don't notice much of this in real-world examples from the A7/A7II even with EFCS disabled.

Click to expand...

Yes.


Jim

stevo23 said:

Any thoughts about adding a line for the silent shutter? I'm surprised no one has characterized it yet.

Click to expand...

As I said to Hank, I will get around to trying it. From my tests of the Nikon D810 with EFCS vs long exposure trailing curtain synched studio strobe illumination in a darkened room (which has to be the gold standard for vibration minimization), there's not a lot to be gained from omitting the trailing curtain acceleration at this focal length.

Going to silent shutter will cost a bit of resolution and add some read noise, although probably neither will affect the slanted edge results.

Jim

JimKasson said:

ProfHankD said:

JimKasson said:

The mechanical shutter on the Sony a7RII is supposed to be better damped. It also has electronic first-curtain shutter (EFCS). How do those things make moderate telephoto pictures sharper? ...

Click to expand...

Very cool. I suppose you could also test with silent (fully electronic) shutter -- I guess that would give a baseline for ambient vibration?

Click to expand...

Yes, but also would drop a bit and add some RN. I will get around to trying it, though. From my tests of the Nikon D810 with EFCS vs trailing curtain studio strobe illumination, there's not a lot to be gained from omitting the trailing curtain acceleration.

Click to expand...

Agreed that there should be almost no positive effect vs. EFCS unless you're shooting a sequence rather than an individual frame, but you never know till you test, right? ;-)

JimKasson said:

JimKasson said:

I assume all A7RII tests were with IBIS off? How about on?

Click to expand...

The camera was on a tripod. I've not found that IBIS helps with the camera on a tripod, and it often hurts:

http://blog.kasson.com/?p=8391

Click to expand...

Logically, it seems that IBIS might be able to partially compensate for shutter shock. How do A7II tests with EFCS on tell you it doesn't? (I didn't see tests with EFCS off.) Also, wouldn't the effect, if any, be much more measurable on the A7RII?

I would certainly agree that, in light of your (and other) earlier tests, one would expect the better performance to be delivered by IBIS off + EFCS on vs. IBIS on + EFCS off. So, this is probably an entirely academic question.... Then again, I'm an academic guy and I also have very serious doubts that the control algorithm for IBIS is as smart as it could be (potential research topic?). ;-)

JimKasson said:

The mechanical shutter on the Sony a7RII is supposed to be better damped. It also has electronic first-curtain shutter (EFCS). How do those things make moderate telephoto pictures sharper?

The protocol:

• The cameras: the Sony a7RII and Sony a7R.
• The lens: the Leica 180mm f/3.4 Apo-Telyt-M
• The filter: Heliopan 77mm variable neutral density.
• The lighting: a single Fotodiox LED-200WA-56 daylight balanced flood.
• ISO set to 800,
• Manual focusing wide open, then lens set to f/5.6.
• Drive set to single
• Camera on an RRS TVC-43 with an Arca Swiss C1 head, RRS L brackets in landscape mode.
• Exposure protocol: ND filter to minimum attenuation,, shutter to 1/500, make an exposures, ND filter down a third of a stop, shutter down a third of a stop, make another exposures… until you get to 1/10 second.
• Make a series with both cameras, EFCS on and off in the case of the a7RII.
• Develop in Lightroom 5.7.1 with standard settings.
• Crop, export as TIFFs, analyze for horizontal edge (that's the edge perpendicular to the way the shutter travels) MTF50 in Imatest.
• Export the results to Excel and graph.

Landscape orientation makes it easier on the mechanical first curtain cases, since the tripod is stiffer in that orientation. Mounting the camera body directly to the tripod also makes it easier on those cases.

The results:



The blue line is the a7R. It does the worst even at high shutter speeds by virtue of its slightly lower-resolution sensor. The orange line is the a7RII with both curtains operated by the mechanical shutter. I still exhibits shutter shock, which is not a surprise. The grey curve is the a7RII with EFCS on. It looks pretty darned good.

Comments and questions appreciated. More focal lengths to come, if there's interest.

Click to expand...

Wonderful thank you!

JimKasson said:

OBTW, even 1300 cy/ph is pretty sharp.

Click to expand...

Indeed. It's always been tiresome how so much opinion polarizes immediately to insisting that it doesn't exist (despite the evidence) or alternatively makes the camera utterly worthless (despite the evidence).

JimKasson said:

Jim

--
http://blog.kasson.com

Click to expand...

I very much look forward to some handheld testing with the Zony 55 and IBIS when you get the chance. And if I might be so cheeky, perhaps even your favourite 35.

Thank you as always for your testing.

--
Blog/Website http://craigspics.net/

I put the Leica 280mm f/4 Apo-Telyt-R on both the a7R and a7RII, mounted the lens collar to an Arca Swiss C1 head on a set of RRS TVC-43 legs, set the whole thing on a vinyl tile over 6 inch concrete floor, and made some slanted edge photographs at various shutter speeds. With the a7RII, i used the mechanical shutter for one set of images, the electronic first-curtain shutter (EFCS) for another, and the all-electronic -- silent -- shutter for a third.

This is a harder test for the mechanical shutter than yesterday's, since using the lens collar as the tripod mounting point allows the camera to move more than mounting the camera body directly to the tripod. Unfortunately mounting a big, heavy lens like the 280/4 to the front of the camera unsupported is a recipe for damaging the lens mount.

The first series of images was made with the camera in landscape orientation. That's easier on the mechanical shutter than portrait orientation, since tripods are stiffer in the up and down direction than they are from side to side. Then I said "what the hey" and made a series in portrait orientation.

As a total aside, why can't all lens collars have crisp detents every ninety degrees like this Leica lens?

You want the good news or the bad news first? The bad news? I thought you were tough. Brace yourself.

Here it is. Although Sony said that they increased the damping of the a7RII shutter over that of the a7R (and I have no reason to disbelieve them) with a long lens like this it doesn't make much difference. That's because the bulk of the motion imparted to a tripod mounted mirrorless camera with no added delay between winding the shutter and firing the first curtain comes from the winding action, not the motion of the first curtain. That is especially true for a big lens which has a relatively low resonant frequency compared to a small one.

Okay, are you ready for the good news?

Thought so. The EFCS on the a7RII works great. In fact, it works so well that going all the way to silent shutter mode doesn't bring any significant improvement on single shots, although it almost certainly would on a series of exposures made in a short period.

Details follow.

First, I'll drag you through the protocol.

• The cameras: the Sony a7RII and Sony a7R.
• The lens: the Leica 280mm f/4 Apo-Telyt-R
• The filter: Heliopan 77mm variable neutral density.
• The lighting: a single Fotodiox LED-200WA-56 daylight balanced flood.
• ISO set to 800,
• Manual focusing wide open, then lens set to f/6.9 (halfway between f/5.6 and f/8).
• Drive set to single
• Self timer set to 5 seconds on the a7RII and 10 seconds on the a7R.
• Camera on an RRS TVC-43 with an Arca Swiss C1 head, RRS L brackets on the cameras, but not used for mounting. Lens collar mounted to head with RRS generic plate.
• Exposure protocol: ND filter to minimum attenuation,, shutter to 1/500, make an exposures, ND filter down a third of a stop, shutter down a third of a stop, make another exposures… until you get to 1/10 second.
• Develop in Lightroom 5.7.1 with standard settings.
• Crop, export as TIFFs, analyze for horizontal edge (that’s the edge perpendicular to the way the shutter travels) for the landscape shots, and vertical edge (that’s the edge perpendicular to the way the shutter travels)for the portrait shots. Measure MTF50 in Imatest .
• Export the results to Excel and graph.

Landscape mode:



The blue curve is the a7R, and the red the a7RII with the mechanical shutter engaged. Looks like the a7RII mechanical shutter is a bit better than the a7R one. The grey curve is the a7RII with EFCS on, and the yellow curve is the a7Rii with silent shutter on. Not much difference between silent shutter and EFCS. Above 1/160 second, you can throw a blanket over all the curves.

Portrait mode:



Not much to choose between the a7R and a7RII mechanical shutters, or between the EFCS and the silent shutter.

Let me point out a couple of things.

It's real close, but the EFCS actually appears to be a hair better than the silent shutter. What's up with that? I suspect it's because the a7RII drops to 12 bit precision in silent shutter mode and the read noise picks up.

This lens doesn't appear to be as sharp as the 180/3.4. How come? I think the main reason is that I couldn't get far enough away from the target.

--
http://blog.kasson.com

ProfHankD said:

Agreed that there should be almost no positive effect vs. EFCS unless you're shooting a sequence rather than an individual frame, but you never know till you test, right? ;-)

Click to expand...

You are right, Hank. I ran a test:

ProfHankD said:

Logically, it seems that IBIS might be able to partially compensate for shutter shock. How do A7II tests with EFCS on tell you it doesn't? (I didn't see tests with EFCS off.) Also, wouldn't the effect, if any, be much more measurable on the A7RII?

Click to expand...

I hear you, and I haven't done the requisite testing to make a definitive statement, but I expect that the upper frequency of effective IBIS operation is lower than the higher resonant frequencies of camera and smallish lenses.

The big reason I didn't test that combination is that EFCS works so well at controlling camera-motion blur when the camera is on a tripod that I consider the problem solved, at least for the photography that I do.

Really big lenses might make enough difference that IBIS could actually help on a tripod, especially in the wind.

I have experimented with mechanically shaking tripods with IBIS on and off:

ProfHankD said:

I would certainly agree that, in light of your (and other) earlier tests, one would expect the better performance to be delivered by IBIS off + EFCS on vs. IBIS on + EFCS off. So, this is probably an entirely academic question.... Then again, I'm an academic guy and I also have very serious doubts that the control algorithm for IBIS is as smart as it could be (potential research topic?). ;-)

Click to expand...

Yep.

Jim

Great data, thanks Jim. I'm troubled by the lower resolution of the silent-shutter. I don't think read noise can be the cause because at ISO 800 read noise should represent a fraction of total noise (read noise+shot noise) for most exposures. Just to be sure, it would be interesting to see a single shutter-speed sampled at ISO 100 between the EFCS and SS.

Horshack said:

Great data, thanks Jim. I'm troubled by the lower resolution of the silent-shutter. I don't think read noise can be the cause because at ISO 800 read noise should represent a fraction of total noise (read noise+shot noise) for most exposures.

Click to expand...

Yes, you're right. I'm not sure how sensitive the Burns SFR method is to noise, but I do remember that the last time I tried to measure the lost in resolution as exposure goes down, it was way better than I thought it would be.

Horshack said:

Just to be sure, it would be interesting to see a single shutter-speed sampled at ISO 100 between the EFCS and SS.

Click to expand...

I'll see what I can do. If I'm going to do only one shutter speed, I can make 16 or 32 images and compute some statistics. There's a lot of noise in the measurements, and that would help measure it and average it out.

What do you think? Portrait mode, 1/40 second?

Jim

ProfHankD said:

.... Then again, I'm an academic guy and I also have very serious doubts that the control algorithm for IBIS is as smart as it could be (potential research topic?). ;-)

Click to expand...

My limited experience so far has been that no industry-developed algorithm is as smart as it could be, but many are more than smart enough to meet design goals.

That said, I imagine the reason that IBIS ON + EFCS OFF could be better is in the case of high speed continuous shooting or non-remote/timer shooting. In this case the benefit of EFCS is potentially diminished, and the downside of IBIS (on a tripod) would be less prominent since there would be actual vibration to control.

For parity, perhaps take the first exposure from a continuous set as the baseline, then see how the following exposures compare to the first?

Interesting how at 1/200? That both mechanical shutters produce sharper results than the EFCS.

Thanks for adding some data to the conversation, it speaks louder than the trolls.