D6?

[ilkka_nissila]

Quick question: Thom Hogan's review of the Z 7 is up, and he reports (and I'd not realised) that it resorts to electronic shutter above 5.5fps (in addition to locking metering). That's possibly tied with the normal requirement to close, then reopen the shutter (he mentions that the shutter lag on the Z 7 is worse than the D850, which surprised me - I'd expected the mirror to be the limiting step). While I like having a true silent mode, electronic shutters currently have a downside in terms of rolling shutter effects, which still gives an edge to the cameras with fast mechanical shutters (like the D5).

 

I didn't even think about it in the past, but when pressing the shutter button on the Z7, it does create a sequence of sounds (in normal mechanical shutter sound) and there would then be two shutter cycles for each picture taken in this way. Jim Kasson reports that the Z7 shutter causes more vibration than the DSLR if EFCS is not used, this could be because of the additional vibration due to shutter closure before reopening, also it could be due to the differences in the body structure and how vibration is managed in the body. So, for intermediate shutter speed work, with mid teles or longer focal lengths, EFCS is probably something that should be used by default with the Z7, to avoid vibration. It feels very odd indeed that a mirrorless camera would create more vibration than a corresponding DSLR, but that's what is reported. So, it's probably best to turn EFCS on, unless you run into some limitation of it (in the Nikon DSLRs, there is a limit of 1/2000s max shutter speed in EFCS mode if I recall correctly, and there may be uneven exposures when EFCS is used with tilt-shift lenses with movements in use).

 

I wonder how long it will take to develop a high-resolution sensor with fast read time (to avoid or minimize rolling shutter) and what drawbacks it might have (dynamic range and high cost have previously been reported as a drawbacks of global shutter sensors). Once global shutter without any image quality drawback is possible and economical then the mechanical shutter can be removed. But it may also be that slowly rolling shutters always have a dynamic range advantage and in that case the mechanical shutter is likely to stay in mirrorless cameras, with two shutter cycles per picture (!).

[heimbrandt]

Nikon seem to have more or less sorted out the viewfinder issue, the Z7 EVF is really good. I wouldn't have thought I would say this, based on my experiences with other EVFs, which I've felt were awful...

More or less is relative. I would use less to describe a viewfinder that lags when you pan (and has parts in the image that move fast, such as the wheels on a car) and a viewfinder that always superimpose some information over the image. In my book the Z7 still has some basics to sort out before I call its viewfinder really good. (I am not blaming Nikon as much as I note that we still have a long way to go before mirrorless is truly better than DSLRs in all respects.)

[mike_halliwell]

Why use a mechanical shutter at all, unless your subjects are moving A LOT and FAST?

 

My J5 could do 60fps, until the buffer fills after 20 pix either JPEGS or NEFs.....:(

 

...and it's totally SILENT!

 

The only way you know it's done it, is the buffer is full and the little green BUSY LED glows.

Edited December 4, 2018 by mike_halliwell

[Andrew Garrard]

EFCS is probably something that should be used by default with the Z7, to avoid vibration. It feels very odd indeed that a mirrorless camera would create more vibration than a corresponding DSLR, but that's what is reported. So, it's probably best to turn EFCS on, unless you run into some limitation of it (in the Nikon DSLRs, there is a limit of 1/2000s max shutter speed in EFCS mode if I recall correctly, and there may be uneven exposures when EFCS is used with tilt-shift lenses with movements in use).

 

The tilt-shift thing is news to me, but Thom is also suggesting EFCS where possible. I assume EFCS has a fast way to "reset the sensor" line by line, racing the speed that the shutter passes it. I'm sure the speed limitation (including 1/200s synch) is merely a design choice of the Z series (although IIRC they did have to design a special slimline shutter) rather than anything fundamental about mirrorless.

 

I wonder how long it will take to develop a high-resolution sensor with fast read time (to avoid or minimize rolling shutter) and what drawbacks it might have (dynamic range and high cost have previously been reported as a drawbacks of global shutter sensors). Once global shutter without any image quality drawback is possible and economical then the mechanical shutter can be removed. But it may also be that slowly rolling shutters always have a dynamic range advantage and in that case the mechanical shutter is likely to stay in mirrorless cameras, with two shutter cycles per picture (!).

 

I wondered whether Sony would get there with the stacked sensor tech. IIRC the older CCD sensors effectively had a per-pixel "storage area" for the charge which wasn't exposed to light, but it took up some of the sensor area (this could be very confused). Stacking circuitry behind the sensor area ought to allow something similar without actually losing exposure area. This is distinct from just doing a very fast read-out; the problem may resolve itself by sensor read-out getting fast enough.

 

Why use a mechanical shutter at all, unless your subjects are moving A LOT and FAST?

 

Having the sensor active contributes to heating and has been known to affect static images. I do track birds and insects enough that I'd be cross if they got stretched out across a 1/9s exposure, not that a 1/250s mechanical shutter is perfect either. Sadly, for now, phase detect AF is my main motivation with the D850 - although the lighting conditions I shoot in might not play all that well with a Z7, especially absent cross-type AF points. People have complained about the shutter noise, so I might have to go back to my D810 some of the time.

 

My J5 could do 60fps, until the buffer fills after 20 pix either JPEGS or NEFs.....:(

 

...and it's totally SILENT!

 

And the D850 can do 6fps silently or 30fps in JPEG 8MP. Small sensors have advantages for speed of data transfer and heating issues.

[ilkka_nissila]

Why use a mechanical shutter at all, unless your subjects are moving A LOT and FAST?

 

With 1/15s read time of the Z7 sensor, the movement would not have to be very fast at all to see with distorted geometry in the recorded image.

[ilkka_nissila]

The tilt-shift thing is news to me

 

It is listed by Nikon as a limitation of EFCS. The mechanical shutter curtain is in front of the sensor and the electronic curtain is within the sensor plane, so there is a certain asymmetry to how oblique rays are rendered around the two curtains. The 1/2000s shutter speed limit of EFCS is likely because the asymmetry would be too great to render a visually consistent image.

 

I wondered whether Sony would get there with the stacked sensor tech.

 

I don't know. After they made the stacked 24MP sensor for the A9, they subsequently made the A7 III which also has a 24MP sensor but with slower read time and better low ISO dynamic range. I don't know if it is possible to go around this or if fast read sensors will always trail behind the slow ones in this respect.

[ilkka_nissila]

More or less is relative. I would use less to describe a viewfinder that lags when you pan (and has parts in the image that move fast, such as the wheels on a car) and a viewfinder that always superimpose some information over the image. In my book the Z7 still has some basics to sort out before I call its viewfinder really good. (I am not blaming Nikon as much as I note that we still have a long way to go before mirrorless is truly better than DSLRs in all respects.)

 

I was playing around with my friend's Z7 in my apartment and the viewfinder didn't show any of the noisy pixelation that I've seen with Sony and other cameras' electronic viewfinders in live view, when panning around in indoor artificial light, there was also no flickering that I could see, in fact I had to look hard to see any sign of the electronic nature of the view. There is a slight delay when panning but that's not my primary concern about EVFs; I look at people's faces and I want to see their expressions and emotions evolve in time, which I use to determine when the shot should be taken, and the artifacts in previous viewfinders really prevented me from being able to work in this way. If the viewfinder flickers, there is a band that rolls from top to bottom slowly (due to some interference between the update and the flickering of lights), and there is obvious pixelation and jaggies, I can't work with that. The Nikon viewfinder seemed completely clean in this respect, so it is quite possible that I would be able to (and willing to) work with it. However, I understand that people who photograph fast action are interested in the refresh time and how much lag there is, which is a separate consideration than mine (of course I too want a fast refresh, and don't want to look at something that happened in the past). Anyway, from my point of view the Z7 EVF is the first EVF which doesn't completely turn me off the prospect of using the camera.

[Andrew Garrard]

Thanks, Ilkka - I wondered whether the asymmetry was the reason.

 

I've heard good things about the live view. I don't know whether there are any dynamic range concerns (I've yet to see a Z body), which has been my biggest worry - I gather it's a lot less bad than some old screens. Other than the option of some additional information on the finder, I don't desperately feel the need for an EVF, especially with live view on the rear screen or HDMI readout as an option, and I like the low-power nature of a mirror, but I don't find the idea toxic. I'm glad technology is improving.

[ilkka_nissila]

Thanks, Ilkka - I wondered whether the asymmetry was the reason.

I don't desperately feel the need for an EVF, especially with live view on the rear screen or HDMI readout as an option, and I like the low-power nature of a mirror, but I don't find the idea toxic. I'm glad technology is improving.

 

I still prefer the optical viewfinder but for some applications there are benefits from the EVF. I have no plans on buying a Z camera now, but may get one later once the native telephoto lenses are available (mainly 85/1.8 which I would pair with the currently available 35/1.8) mainly for the quieter sound of the camera relative to DSLR, for concert applications and such things where quietness is a priority. Nonetheless the F mount system with its DSLRs are likely to be my main system in the foreseeable future even if I get a Z camera for special needs.

[mike_halliwell]

So, err, how do you have a shutter speed of say, 1/16000th second (my lovely J5 again) and a 'read time' of 1/15?

 

Where does that put subject movement?

 

1/16000 is fast enough to freeze movement with a flash, why not a shutter?

[Andrew Garrard]

So, err, how do you have a shutter speed of say, 1/16000th second (my lovely J5 again) and a 'read time' of 1/15?

 

Same way you can have shutter speeds faster than the 1/250s flash sync on a dSLR, kind of, or why there's rolling shutter on video. You reset the lines of the sensor independently, and read the results of some of them out while resetting others - 1/1000 of the sensor is still a few scan lines. I assume the resetting is faster than the read-out, which is why the EFCS can clear a line as fast as the mechanical shutter can move down the sensor (it can't be "all at once" or you'd have different exposures across the frame).

 

Where does that put subject movement?

 

Still moving as the different lines are read out. :-) (Mechanical shutters have the same issue, just less badly because 1/250s is quite quick.)

 

1/16000 is fast enough to freeze movement with a flash, why not a shutter?

 

Because the flash illuminates the whole frame at once. It's not that you'll get blur, it's that different scan lines are seeing the frame at different time offsets, so if something is moving it'll likely get stretched or squashed.

[mike_halliwell]

So it's not really a shutterspeed of 1/16000ths then? In realtime?

 

It's a 1/15th exposure made of 1100 x 1/16000th second strips.

 

So there's no movement within a strip, but there is between them?

 

Because the flash illuminates the whole frame at once. It's not that you'll get blur, it's that different scan lines are seeing the frame at different time offsets, so if something is moving it'll likely get stretched or squashed.

 

So, if I fire a low powered flash (say 1/20000 th duration) in the middle of a 1/2 second exposure in the dark, i'll get the electronic shutter equivalent of shutter-blade-shade?

 

ie most certainly NOT like film!

[BeBu Lamar]

Andrew,

Shun already has the Z6 why don't you have the Z7 yet?

[Andrew Garrard]

So it's not really a shutterspeed of 1/16000ths then? In realtime?

 

It's a 1/15th exposure made of 1100 x 1/16000th second strips.

 

So there's no movement within a strip, but there is between them?

 

I doubt the electronic shutter exposes a whole strip at a time, just as the mechanical shutter doesn't judder. The line is reset (in the same way the whole sensor is prior to any exposure), then read back however much later is needed for the nominal exposure duration. Without a true global shutter, you can't do this for the whole frame at once (you might be able to clear it, but the read-out time is finite), but you can ensure that you only start exposing a line 1/16000th of a second before you'd be able to read it. If you're taking 1/15th of a second to pull the data off the sensor, each line would get reset 1/16000th of a second before it was going to be read, effectively one line at a time. The reset and read follow each other across the sensor like the first and second curtains of the shutter.

 

At least, I assume. I've never been anywhere near the implementation of one of these.

 

So, if I fire a low powered flash (say 1/20000th duration) in the middle of a 1/2 second exposure in the dark, i'll get the electronic shutter equivalent of shutter-blade-shade?

 

If it's a 1/2 second exposure, the whole sensor should be exposed concurrently, like a normal mechanical exposure longer than the flash sync speed. (This may not be true of some compacts or smartphones that compose longer exposures out of multiple shorter ones to manage motion blur, but that's not the issue here.)

 

You'd get a partial exposure if you fired a fast flash part way through an electronic exposure shorter than the sensor read-out time (which we've been claiming is 1/15s-ish); even with a 1/15s exposure one part of the sensor would be just starting its exposure and the other would be just finishing, giving a variable blur from ambient light part way between first- and second-curtain flash depending on where you are in the frame. I believe Nikon disallow flash with the electronic shutter because they know the results are going to seem weird at common shutter speeds.

 

ie most certainly NOT like film!

 

It'd be what you got from film if you had a mechanical shutter speed with a 1/15s flash sync speed. I've never met one, but 1/50-1/60th isn't that rare (from Leica and the original F). You get the same thing from faster shutters too, it's just that the subject has to be moving faster for it to be relevant.

 

I'm now mildly curious whether the EFCS could be mildly beneficial to shutter lag...

 

Andrew,

Shun already has the Z6 why don't you have the Z7 yet?

 

All donations gratefully received, BeBu!

[ben_hutcherson]

I doubt the electronic shutter exposes a whole strip at a time, just as the mechanical shutter doesn't judder. The line is reset (in the same way the whole sensor is prior to any exposure), then read back however much later is needed for the nominal exposure duration. Without a true global shutter, you can't do this for the whole frame at once (you might be able to clear it, but the read-out time is finite), but you can ensure that you only start exposing a line 1/16000th of a second before you'd be able to read it. If you're taking 1/15th of a second to pull the data off the sensor, each line would get reset 1/16000th of a second before it was going to be read, effectively one line at a time. The reset and read follow each other across the sensor like the first and second curtains of the shutter.

 

FWIW, the D1 series is often claimed to be able to sync at any speed. I've personally used one to test the flash duration on my Norman strobes, and know that depending on the power setting used on the pack I really shouldn't exceed 1/500(honestly, under my typical conditions any reasonable shutter speed isn't an issue because I'm often at f/22 or smaller and ISO 100 with the only other significant ambient light being 250W tungsten-halogen modeling lights that meter at several seconds long at the aperture I'm using). I think the maximum shutter speed is 1/16,000, and it's possible to sync at that speed, although you're going to see significant light loss from most any strobe.

 

In any case, I THINK the key difference there is in the use of CCDs rather than CMOS sensors. I've been under the impression that CCDs can be "pulsed" and actually acquire the image in a way that a CMOS can't. I don't fully understand the quantum mechanics of it, but I recall talking about using CCDs in one of my graduate school instrumental analysis courses as a substitute for PMTs in some applications. Of course, that's a very, very specialized thing(and it's not uncommon for them to be liquid nitrogen cooled to increase the sensitivity...and even when that's done they still can't approach that of a PMT) but the response time is definitely good enough for certain applications. I don't THINK CMOS shares that characteristic...and of course often times in instrumental applications we're talking about CCDs with a few hundred pixels at most(perhaps larger versions of autofocus sensors) and not millions of pixels.

[Andrew Garrard]

Thanks, Ben. That matches my hazy recollections: the CCD lets you "store" the sensor pixel values prior to read-out, meaning that further light (within reason) doesn't affect them after that point and they can be read out at leisure (and they can be reset collectively too). That would allow a true global shutter.

 

This suggests two separate modes of operation: either "digitally exposing" the whole sensor and relying on the mechanical shutter to expose the frame (with different exposure times at different points, as for a normal mechanical focal-plane shutter), or locking the shutter blades open and exposing the sensor only at the desired time. I assume the D1 does the latter only during flash sync, although I've not checked the manual.

[ben_hutcherson]

Andrew,

 

Anecdotally, it's been my understanding that the D1 does this on all shutter speeds above 1/250(the shutter travel time).

 

I don't know of any other Nikon at least(possibly not any other "common" camera) with a maximum shutter speed of 1/16,000-even the D5 and Z7 max at 1/8,000. It's been explained to me that this "retreat" in speeds was because the D1 fast speeds were entirely electronic. The D2 series of course does not use CCDs-the D2H uses Nikon's proprietary LBCAST sensor(which I understand is more like a CMOS in principle) and the D2X uses a Sony CMOS.

 

In any case, I've not tried it up to crazy high speeds, but with my D1 series cameras when I sync via the PC socket, I don't see shutter curtain artifacts-I just see a decrease in exposure across the frame since the light duration is longer than the sensor is "on."

 

Of course, the value of having 1/16,000 shutter speeds is questionable, especially in a digital camera. I can probably count on one hand the number of times I've even used 1/4000, especially since in my personal outdoor photography I'm often doing my best to work at f/5.6 or f/8(generally a safe place unless I've specifically tested the lens) and I'm either using ISO 50 or 100 slide film, or keeping the ISO of the digital camera as low as possible. Indoors, getting a high enough shutter speed is usually the issue. I could see the value of a shutter speed that fast if you were shooing ISO 400 film in full sun and wanted to have your lens as wide open as possible, but even 1/16,000 won't get you to f/1.4(there comes a time where you should need to start stacking NDs, but back in the film days if I really wanted to use an aperture larger than the maximum shutter speed of the camera would allow and I was shooting color negative film, I'd just overexpose it-most any color negative film will handle a stop or two gracefully, and more than that if you're daring). I can't think of much action that 1/8000 won't stop(perhaps if I stuck a camera inside a mass spec and wanted to photograph one of the turbo pumps spinning at 10K rpms, but if I really wanted a photo of that I'd just photograph one outside the instrument, and I also don't know how my cameras and lenses would handle high vacuum), although admittedly the lack of concern about distortion from the shutter is a big plus. At least we're not still using Speed Graphics to cover auto races and getting the consequent oval wheels from them!

[Andrew Garrard]

Thanks, Ben. There were consumer bodies with CCDs and faster-than-their-mechanical-shutter-speed sync (the D70 and D40 can do 1/500s, for example), so I assume they do the same but artificially don't stretch as far as the D1. The D4x apparently can't do it at all, and hits the now consumer standard 1/200s. At least, I assume the older consumer bodies didn't have extremely fast mechanical shutters.

 

I remember a sample image from the D1 introduction that showed an internal scene with bubbling water running from a tap. It was completely frozen and undistorted, because of a very short shutter speed in a global shutter; to do it with a modern camera you'd need to match that output from a flash with a very small and discrete output duration. I've been known to try to freeze things in flight, and sometimes run into the limit of my flash duration. No doubt it's a specific use case, though!

[Ed_Ingold]

Battery life is certainly an issue with mirrorless cameras. This is due to the limited size of the battery compartment and (more important) the high power overhead of maintaining the EVF and/or back screen. The Sony A9 and model 3 A7's have made huge strides in this regard. The FZ100 batteries have twice the capacity of the earlie FZ50 batteries, and the newer cameras use 40% less power. By powering down during idle time, I seldom have to change batteries in a day of serious shooting (1000+/day in Ireland last spring). Taking single frames, it's not unusual to get 600 images or 5 hours per charge. whichever comes first. Shooting bursts, the count can exceed 6000/charge. I can shoot 2.5 hours of HD video on a single charge, using an external recorder (Ninja V), and over 1 hour recording internally. The camera gets barely warm in this time.

 

Regarding auto focusing, the Sony A9 has nearly 700 phase detectors, compared to 150 for the Nikon D5, covering over 90% of the field of view, instead of about 25% for the Nikon. Furthermore, there is no need for calibration, because they're located in the sensor, not the viewfinder, and they work all the time except at the moment of exposure. Having owned an A9 for nearly a year (and still learning), I'm convinced negative reviews about AF are due to inexperience with the camera and the many options affecting AF. Tracking not only works flawlessly, it is almost too sticky. The designated subject can leave the FOV momentarily, and be picked up again when visible.

 

We haven't even touched on exposure control, including a "Face Priority" mode.

 

It is very early to be talking about a Nikon D6. According to rumors, Canon has already postponed (or abandoned) releasing a new DSLR. Canikon mirrorless cameras have some catching up to do, and the target is moving at full speed. In any event, the die is cast for the future

[Ed_Ingold]

Digital cameras do not suffer from reciprocity failure, unlike film, for long or very shot exposures. The problem with operation in high vacuum is heat dissipation, not functionality.

[Andrew Garrard]

Ed - I suspect it came up because there have already been rumours of both a D6 and a D760. Not detailed ones, and for all I know rumours generated by someone thinking "it's about time Nikon updated these bodies", but not out of the blue. Whether they're "due replacing" depends as ever on what the competition does - the D5 is still the best tool for the job some of the time, but in other cases the A9 or 1DxII will be. For the D750, I don't know whether the 6DII is a massively high seller, I don't know how much Nikon feel the squeeze from different versions of the A7, and I don't know whether they feel the Z 6 is "the answer". They may not know themselves yet.

[ben_hutcherson]

e problem with operation in high vacuum is heat dissipation, not functionality.

 

If I stuck a DSLR in a mass spectrometer at 10^-5 torr or better, I'd be concerned about things like the lubricants starting to offgas and go places that they shouldn't.

 

Not planning on doing, both for the health of my cameras and my mass specs :)

[heimbrandt]

So, no one longer believes in a D5s before the D6?

[Andrew Garrard]

FWIW, I just took a peek in the D1 manual. Although the fastest shutter speed is 1/16000s ("combined electronic and mechanical"), the fastest flash sync speed seems to be 1/500s. Still better than the current cameras with purely mechanical shutters, but not by an order of magnitude. Given the CCD, I'm actually a little surprised that it couldn't go faster if it wanted to. Apologies for conflating specs. Given that the D5 (and D850) have the same 1/250s flash sync speed as the D500 but the D500's shutter only has to move 2/3 as far, I'd almost expect the high-end DX bodies to have a faster flash sync purely mechanically, but I assume an electronic shutter is involved in the D1 getting to 1/500s. Or it's just possible that I still don't understand Nikon cameras. :-) (In my defence, this was before my time.)

 

So, no one longer believes in a D5s before the D6?

 

I guess we'd have expected to see a minor D5 refresh by now? Not that I'm quite sure what would go in it. (This doesn't mean I don't have a list, just that I don't know which parts Nikon would obviously prioritise.) The D850 doesn't add all that much which would be appropriate for a D5s body. Nikon could choose to call a replacement whatever they like, of course.

[Dieter Schaefer]

IIRC correctly, then the D1 has only one shutter curtain - so under all circumstances, the shutter is electronic.