Yesterday, I did a test of the lag of the lag of the LCD panel on the back of the Sony a7RII. Today, I’ll do the same thing for the Sony a9, and add some notes about the EVFs on the two cameras.
Here’s a picture that pretty much says it all:
This is a crop from the center of an image made with an a7RII set to EFCS and 1/500 second. That makes the time difference of the capture, from the top of the frame to the bottom, 4 milliseconds (msec). The time difference from the top to the bottom of this crop is less than 2 msec, which is small compared to the lags that we’re observing. In addition, The scope and the display of the scope are at about the same location in the frame. Because of that, I estimate that we’re talking about a difference in time between the two of about 300 microseconds (usec). You can get a sense of that because of the verticality of the trace on the scope screen itself.
The scope sweep speed is set to 10 msec per division or 100 msec from the leftmost part of the screen to the rightmost.
Now observe the image on the LCD panel. The green band is not vertical. It is slanted, with the bottommost part of the screen being displayed 6 msec or so later than the topmost part. That shows that the LCD panel is refreshed vertically. You’ll note that at the top of the LCD the green band is about half as wide as at the bottom. That means that I got really lucky, and caught the display at the exact moment when it was halfway through displaying the top of the band. That makes the LCD lag calculation easy; all we have to do is look at the green line on the scope and see where it is. It’s three and a half divisions over, which is 35 msec. So the LCD lag is 35 msec at the top of the display, and less than 30 msec at the bottom.
But what about the EVFs on the a7RII and the a9? Do they have different lags from the LCD panels? I tried to figure out how to shoot the image in the finder and the scope itself at the same time, but I couldn’t make it work. I am open to ideas.
I did notice a few things while I was playing around with the finder images:
- The a7RII finder refresh rate is twice that of the LCD panel.
- The a9 finder looks entirely different than its LCD display when viewing the scope. It doesn’t look stroboscopic like the a7RII finder, but much more like looking at the scope itself.
- Half-pressing the shutter release button of the a7RII can double the refresh rate.
I wish I had more clarity on the EVF lag, and I’ll be thinking about how to get it. I’m sure I could gin up something with a half-silvered mirror, but I don’t that’s worth the effort.
[Added 10/11/17: I have ordered a device that mounts to the front of a lens using a 77 mm filter ring, and produces a stereo pair at the focal plane. I hope to be able to find a focus point where one half of the pair, which is looking at the scope directly, and the other half, which is looking through the camera finder, will both be sharp. Stay tuned.]
[Added later. I tried the above technique and couldn’t make it work. Sorry.]
I pre-ordered the A7rii within minutes of pre-order opening up, so even though I am more or less a weekend warrior, I have accumulated a fair amount of experience with the A7rii. I shoot birds and my daughters athletic events, usually with manual focus lenses (challenging, but my keeper rate for soccer is about the same for manual 300 f2.8 and autofocus using Alpha mount 70-400 with auto focus adapter). I can not recall ever experiencing lag with a manual focus lens (lens with no electronic coupling). To the contrary I attribute my high keeper rate with manual focus to the consistent negligible lag; when I see focus I press the shutter and capture the image before my subject has moved out of focus.
Things are less clear cut with autofocus, especially via adapter. Of course time is needed to focus, but lag can be unpredictable in auto focus mode even even when no apparent focus occurs. I am not positive, but I think I have experienced lag on full press when starting with half pressed shutter. Based on my experience I would guess either auto focus or some other body-lens communication (aperture change command perhaps) has a rare mode (possibly a coding defect) that causes occasional and unpredictable delay in initiating exposure. Sadly I expect this theory would be painstaking to test and would rate my confidence as speculative at best. Focus lag could plausibly account for 100% of incidents of lag that I have experienced. It is more the lack of lag when I use lenses with no electronic coupling that drives my speculation.
Finder lag is what I measured, not the lag between the time you press the release and the time the exposure begins. For the a7RII, according to Imaging Resources, presumably with EFCS on, full AF shutter delay is 212 msec, manual focus delay is 107 msec, and prefocused is 20 msec. That’s in cascade with the finder lag. For the a9, it’s 216 msec for full AF, 97 msec for manual, and 20 msec for prefocused. That’s also in cascade with the finder lag. Turn off EFCS and things get much worse.
Since you began the A7rii article with observation on shutter lag complaints I was offering an alternative explanation. Your results seem to me to eliminate finder lag as a potential root cause for perception of lag.
Thank you for the shutter lag measures above. I suspect the ‘manual focus’ shutter lag measures at ~100ms applies when there is an electronic coupling to the lens and the camera is controlling aperature (for the 80ms extra over prefocus). I speculate that the lag for lenses with no electronic coupling is same as prefocus at 20ms.
If I am correct total lag (finder plus shutter) is under 100ms for lenses with no electronic coupling and for prefocus. For lenses with electronic coupling, one cycle of aperature communication looks like it takes 80ms. An occasional defect that requires a retry/retransmit that adds an extra 80ms per retry might go unnoticed on one fault, but two or three in a row would be noticeable. A retry on auto focus comms would be more costly.
Based on these numbers, I’m doubling down on a defect in camera-lens communications as most likely root cause of shutter lag complaints. Fits my experience of consistent near instant capture with lenses w/ no coupling, and occasional inconsistency / hesitation with electronically coupled lenses.
So I do appreciate that you measured finder lag. I am suggesting, however, that explaining shutter lag complaints will require finding a way to measure or isolate a defect in camera-lens communication. I am in awe of your measurement skills, yet I would not wish upon you the quest of capturing the defect I hypothesize. Too much chance of sending you on a snipe hunt….
Many thanks for your blog, it is a fantastic resource. Hope my speculations are at least entertaining if not plausible.
any chance you will also do the test with the GFX? Fuji claims their display lag is 0.005s for years now and I saw a test of the X-T1 once, which supported this claim. Would be interessting to see, how the GFX works out
I haven’t figured out how to test an EVF, as opposed to an LCD back panel, yet. I thought i had something that would work, but it came yesterday and didn’t solve the problem. Sorry.
I would think an led, pulsed, with a macro lens focused on it, and with its output detected with two phototransistors, one at the led and one at the evf, fed into two channels of your scope, would work.