A7Rii Shutter Lag

[Ed_Ingold]

<p>I devised an interesting test for EVF and shutter lag times. I have a metronome app for my iPad which has a sweep hand with an optional click and flash every beat. I set the frequency to 60 bps, so the sweep hand covers four seconds with four divisions per second. Since the pattern is repetitive, I can rely on my musician's training to anticipate clicks, and as a photographer, to anticipate the position of the sweep hand or the flash. The flash is very brief, probably on the order of 1/60 second (my oscilloscope is buried at the moment). The shutter speed was 1/100 sec throughout. The camera was a Sony A7Rii with a Sony 90/2.8 Macro lens in locked focus mode (half-press).</p>

<p>The tests were as follows. (1) Click and flash, using the silent (electronic) shutter, (2) electronic front shutter. (3) flash only, no finder and (4) Click and flash, full shutter. With the EFS and full shutter, the camera would capture all or part of the flash, obscuring the scale. For clarity, I made some shots with the flash turned off.</p>

<p>My conclusions are as follows. The silent shutter (electronic only) has about a total lag time of about 1/4 sec. The total lag time of the electronic first shutter is about the same as the full mechanical shutter (two closures), between 1/60 and 1/30 sec (interpolated). The shutter lag time, not using the finder, is too small to be measured in this manner (< 30 msec). There is no significant difference observing the flash through and without the viewfinder (<30 msec). By inference, if you can't capture the "peak of interest" with the A7Rii, don't blame the camera.</p>

<p>Silent Shutter, flash only<br /> <img src="http://d6d2h4gfvy8t8.cloudfront.net/18116450-lg.jpg" alt="" width="600" height="400" /></p>

<p>Silent shutter, click and flash<br /> <img src="http://d6d2h4gfvy8t8.cloudfront.net/18116449-lg.jpg" alt="" width="600" height="400" /></p>

<p>Full Shutter, click and flash<br /> <img src="http://d6d2h4gfvy8t8.cloudfront.net/18116448-lg.jpg" alt="" width="600" height="400" /></p>

<p>Full Shutter, no finder<br /> <img src="http://d6d2h4gfvy8t8.cloudfront.net/18116447-lg.jpg" alt="" width="600" height="400" /></p>

<p>You will find all the test photos in my portfolio under "Metronome." (http://www.photo.net/photodb/folder?folder_id=1085224)</p>

<p> </p>

[eric_arnold]

<blockquote>

<p>By inference, if you can't capture the "peak of interest" with the A7Rii, don't blame the camera.</p>

</blockquote>

<p>taking shots at Thom Hogan, are we now? i'm not sure how useful your test is, however, because it's a static lab test -- Hogan's experience was in the field under active conditions, and EVF lag or blackout is a real concern for sequential shooting in continuous focus mode with all current mirrorless cameras (except maybe the new Leica bhemoth, which isnt even commercially available yet). if you aren't shooting consecutive frames under those conditions, how does this test help someone who is or wants to? as far as i can see, all you have done is confirm there is an EVF lag (which is different from shutter lag BTW, as was earlier pointed out). i don't think this test actually solves the problem. </p>

[Ed_Ingold]

<p>I forgot to mention that some tests were made using observation of the flash only in the viewfinder.</p>

<p>By timing my shots listening to the click, any delay is due to the shutter lag. If I time the shots by observing the flash in the viewfinder, the observed delay is a sum of the EVF lag and shutter lag. The variation between consecutive measurements is on the order of 30 msec. I also timed my shots by observing the flash without clicks, looking over the top of the camera. There was no significant difference between those results and those listening to the clicks (which are also independent of the EVF lag. Most of the shots with the flash turned on, using the mechanical shutter captured the flash in progress. When using the click, I turned the flash off so I could get a reading of the delay from the sweep hand.</p>

<p>I also observed the metronome flash at the same time as the viewfinder. I observed a slight lag, too short to measure with this test method, less than the experimental error.</p>

<p>A metronome is a lot more reproducible than a football play. With due respect to Thom Hogan, my test is more objective. As far as sports action, I was shooting basketball games from under the net before Thom was a gleam in his father's eye, thirty years before auto focus was commonplace, and continuous shooting required a nimble shutter finger punctuating a quick flip of the wind lever. Tri-X was rated at 200. Tell me about anticipating the action ;) Sports editors aren't the forgiving sort.</p>

<p>The blackout from an EVF is no longer than that of a DSLR, and probably shorter. Furthermore, you don't see the image jump, like you do as the mirror starts to flip.</p>

[david_smith110]

<p>Interesting test Edward. Thanks for taking the time to conduct it and post the results. The great thing is EVF's will only get bigger and better as technology improves. This cant really be said for OVF's which are sort of at a dead end technologically. I cant say that for certain of course, since you never know what innovations will be released. But right now it is hard to imagine any breakthrough in OVF tech.</p>

[Jochen_S]

<p>Interesting! Thanks Edward. How or where to get hold of a similar target for Win or Linux laptops? - I'd love to evaluate the cameras I have here.</p>

[Ed_Ingold]

<p>It takes time to get used to an EVF, even a very good one like the A7. Unlike an OVF, there is no depth, no 3D. Everything is in one plane. If something's not in focus, your eye can't see through the plane and fix things. Once focus is locked, every view is a DOF check (the diaphragm stops down). There is also a stroboscopic effect. The A7, for example, updates about 60 fps, but each frame represents an extremely brief interval, which increases as the light grows dimmer. You don't see a rapid movement, no blur until the light is almost too dim to see anything outside the finder. Instead you see three or four positions in the blink of an eye. Your mind must learn to fill in the blanks.</p>

<p>So there are issues with EVFs, just not the ones we've heard so much about. After a few thousand shots, many involving action of some sort, they don't seem like problems any more.</p>

[Ed_Ingold]

<p>Search for applications under "metronome". This particular one is called "Clockwork" for iPhone, but works on an iPad too. I toyed with the idea of using a turntable, but this little app proved to be ideal, since it has both audible and visual cues and a sweep hand.</p>

<p>Another project I'm considering is an objective way to measure the sound level of shutters and such. The key is to get a "standard" sound pressure source which could be used to calibrate an ordinary microphone and recorder. I have an audio editor (Wavelab) which displays wave forms, will do a frequency analysis (fast Fourier transforms), peak and averaging.</p>

<p>There are sound pressure meters, some relatively inexpensive ($25) as well as apps for a smart phone. However they're geared to fairly loud source like jets, air hammers and movie previews (clocked at 106 dB unweighted by your's truly). For an anechoic chamber, a friend from MIT suggests your own back yard on a quiet afternoon. (No reflective surfaces worth mentioning.)</p>

<p>Science on the cheap ;)</p>

[eric_arnold]

<p>so you're just going to ignore the part about EVF lag in continuous shooting, then? </p>

[Ed_Ingold]

<p>I think it is a moot point. There is a moment, at the instant of shooting or in continuous shooting, where the image is frozen just before it blacks out. It all happens in the blink of an eye. At that point you are committed to the shot and nothing you do will alter it. Either you chose the right moment or you didn't. In continuous shooting, this sequence repeats itself and you get about 5 fps as long as you hold the shutter release or the camera runs out of memory. Pressing harder or less will not alter that sequence. The effect in a DSLR viewfinder is disrupted in a similar fashion, only the image seems to move vertically because the mirror is moving.</p>

<p>At this point in time, many DSLRs have a faster continuous rate than the A7. You can bet that will change, as it has changed with the Leica SL (601). Whether 4K video (12 MP frames) at 30, 60 or 120 fps will prove useful in sports remains to be seen. The A7Rii will do that, and you can see everything through the viewfinder in real time, with full control over focus and exposure.</p>

[Ed_Ingold]

<p>I took a closer look at the continuous shooting mode of the Sony A7Rii, using the metronome tool described above. The viewfinder displays a series of still shots separated by an almost imperceptible blackout. Each still shot represents the last moment before the image is captured, and does not wait for processing of captured image. There is no perceptible delay in the viewfinder from the time the shutter release is pressed until the first image is displayed.</p>

<p>This contrasts with the optical finder of a DSLR, in which the image is blurred as the mirror moves out the way and back again, with a brief period of clarity as the shutter recovers. If the DSLR is more effective, it is because most are about twice as fast as the 5 fps of the Sony. Since I am shooting uncompressed RAW images, the buffer fills up rather quickly at 82 MB per pop. The obvious solution would be to shoot JPEG only, still at 5 fps but for a longer time.</p>

[Carl Stone]

<p>This would be really useful if you were trying to sell images of your metronome. Not so much if you were trying to sell sporting event images to SI. </p>

[Ed_Ingold]

 

 

<blockquote>

<p>This would be really useful if you were trying to sell images of your metronome. Not so much if you were trying to sell sporting event images to SI.</p>

</blockquote>

 

 

I'm trying to explain what happens in the finder, not sell images of a metronome. That aside, do you have something to contribute?

 

<p><a name="pagebottom"></a></p>

[Carl Stone]

<p>Yes, your findings in your lab test, didn't match the experience that Thom Hogan had when he used the Sony for sports shooting. THAT calls the value of your lab testing into question. </p>

[Ed_Ingold]

<p>There are no inconsistencies. Thom Hogan said he had trouble anticipating action. While he blamed the problem on viewfinder lag, we see that that is not an issue. The viewfinder behaves different from that of an SLR. Chalk up Thom's analysis to lack of experience with the camera and not recognizing the root cause of the problem.</p>

<p>If there were a lag, the sweep hand would be past the top mark in the exposure. We see that to be the case if the silent electronic shutter is employed, about 250 msec in fact. When the mechanical shutter is used, front and back or back only, the shutter lag about 0.1 divisions (interpolated), or 25 msec. I get the same results whether listening to the clicks, or observing the flash through the viewfinder with the click turned off. The first depends only on shutter lag and the latter records the viewfinder lag plus the shutter lag. Consequently the viewfinder lag is significantly less than the shutter lag, or no more than about 15 msec.</p>

<p>This experiment is designed to eliminate human reaction time. Since the beats occur at regular intervals, each beat can be anticipated very accurately, click or flash. The flash has a duration of about 1/30 second (2 frames at 60 fps), and most of the time, the flash itself was captured.</p>

<p>Of course in football things happen only once, so reaction time of the photographer is very much a factor, typically between 100 to 250 msec. At 5 fps, the Sony captures the action in 200 msec intervals. In football, a lot can happen in a fifth of a second.</p>

<p>I said the EVF behaves different from a DSLR. When you press the shutter release, the first image is captured almost instantly (< 30 msec) but the last view is frozen in the viewfinder. It is held until the camera is ready for the next shot, or in continous mode, next image is captured, and that image is displayed, etc. With a DSLR, you only see what happens BEFORE the image is captured. During the actual capture, the viewfinder is blind. With the Sony EVF, you see what you actually captured. In single mode, it's just for an eyeblink, but that's enough to know if you need to try again. Armed with this knowledge, life gets a lot easier on the sidelines.</p>