What Is 'Global Shutter'? Part II (Sony A9III)

The new Sony A9III

So it's not like we've never had global shutters before. Most CCD (charge-coupled device) sensors worked that way, and most early digital cameras had CCD sensors. Today CCDs are still widely used, but mostly for practical uses like microscopes and other medical equipment, telescopes for astronomy, scanners, bar code readers, and so forth. CMOS (complementary metal oxide semiconductor) sensors offered many advantages for cameras—they have higher dynamic range (remember the DR limitations in early digital cameras?), they're much more energy efficient and generate less heat, and they offer size and manufacturing advantages because the readout logic could be integrated on the same chip. And they have far less blooming than CCDs. "Blooming" is when one photosite (pixel) becomes saturated and begins to spill over into the photosites around it. Canon was an early pioneer of CMOS sensors in cameras, beginning development in the 1990s. The first major pictorial camera to have a CMOS sensor was the landmark three-megapixel Canon D30 (not 30D!) of 2000.

At Photo Techniques magazine, I started a series of print sales to provide examples of techniques for readers called The Collector Print Program, and we offered several Canon D30 prints made by a guy I had just met named Michael Reichmann. CMOS sensors have undergone continuous development and improvement since then, and almost all pictorial digital cameras now use them. I'm not aware of any exceptions, although one or a few may exist. 

And there have been CMOS-sensor cameras with global shutter before: B&H Photo lists the Sony PMW-F55; Blackmagic Design Production Camera 4K, URSA 4K, and URSA Mini 4K; and the AJA CION. All of these are or were expensive video cameras: the Sony PMW-F55 CineAlta for instance has been described as a "super camcorder," and the first announced price was $34,900.

Enter the Sony A9III
The new Sony A9III (which isn't out yet, mind you, and might not be for a while) might not interest you personally for your immediate needs. But it's probably going to prove to be a landmark camera. Depending of course on future implementation. The 24x36 ("full-frame") CMOS global-shutter sensor demands a massive amount of processing power: Sony says eight times as much as the A9II, which was no slouch in that department. The highest shutter speed is a fantastical 1/80,000th of a second, and the sync speeds are said to go up to that (I still haven't worked out what's going on with the sync speed). The maximum frame rate is up to 120 per second, with new controls to set and access that. Rolling-shutter artifacts and banding will be completely eliminated.

As has the physical mechanical shutter: in the A9III, there isn't one.

On the disadvantage side, base ISO is 250, which might result in a measurable diminishing of best-case DR. But note that word "measurable." I think it would be unwise to make too much of that, at least until practical testing starts to get real-world effects sorted out. Until we know better, I would doubt it will be detectable in ordinary shooting.

The camera is a true tour-de-force, and a technical showpiece. As a product, it's a niche camera meant for very serious sports, action, and wildlife shooters, and for a subset of professional and semi-pro videographers. As such it will pass by in the parade of products as all digital products do. However, as a demonstration of future possibilities and of Sony's technical prowess, it's astonishing. Therein lies its real meaning: to remind us what Sony can do, and to bring Sony full-frame mirrorless back into the conversation—if not front and center, at least shoulder-to-shoulder with the rest of the players.

Mike

Original contents copyright 2023 by Michael C. Johnston and/or the bylined author. All Rights Reserved. Links in this post may be to our affiliates; sales through affiliate links may benefit this site. As an Amazon Associate I earn from qualifying purchases. (To see all the comments, click on the "Comments" link below or on the title of this post.)

Featured Comments from:

Peter Jeffrey Croft: "Re 'The camera is a true tour-de-force, and a technical showpiece.' What an incredible camera, and as you say, a technical marvel. I'm reading Akio Morita's [co-founder of Sony —Ed.] book Made in Japan at the moment (a bit late, yes) and it's no wonder Japan is way out in front in so many ways. Sony treats its employees so well, as a family, with a great deal of respect, encouraging them in so many ways to be their best. My work was in the electronics field (TV broadcast) and I used to say it felt as if our management read the books about managerial best practice, then took an oath to deny and forget everything they read and do the opposite. I felt discouraged and disheartened at the lack of respect (when our chief engineer retired, he wasn't replaced), and the lack of care about employees who felt un-valued (I was the union rep so I heard their tales of woe). Again and again, I had to wave goodbye to great guys who left because they felt un-valued and underpaid. We had secretaries who were paid more than techs! The company cared almost nothing for academic qualifications—not their business, they said. Training is not our business. It's no wonder Japan raced ahead and is so advanced. This camera is just one example; there are many, many more."

Jayanand Govindaraj: "My guess is that, for now, in practical terms, the 120 fps speed is meaningless and a marketing gimmick, as the camera uses slower CFE(A) and SD UHS II cards, so no amount of buffer can compensate for the slowness of the cards. I doubt you would get more than two seconds shooting at that speed before the buffer overflows—unless, of course, the RAW that is mentioned is a highly 'compressed' version, bordering on JPEG. In that case, the photographer will get an animal/bird/sportsman/whatever twitching, at best, and have 240 frames of that twitch to check out. If you are forced to cut the fps to get decent shooting time, I would guess there are many other cameras that will be better for the purpose, and at a lower cost.

"This is a technological tour-de-force which could well be the future, but I doubt whether it is the present."

Comments

[Dave]

I think one major "trickle down" feature that will help future lower level cameras is the increased processing power.

Most cameras can't process images quickly enough to have a high sustained rate before the buffer has to slow down everything.

I rarely take bursts of photos, but many would probably benefit from faster processing. Plus, camera makers would have a second tier of "non-global" shutters with lesser specs at a lower price.

Of course, taking indoor sports shots at reasonable ISO speeds and high shutter speeds would require fast lenses. Another opportunity for camera makers to keep the cell phones from totally taking over.

The RAW sample photos don't look all that great at an ISO speed of 1,600: (https://www.photographyblog.com/reviews/sony_a9_iii_review#sample_images)

So, if you need a large photo, it looks like ISO 400 may be the limit (although there were no photos at ISO 400 or 800 at the above link).

Maybe the release date of Spring 2024 is also necessary to make software adjustments to lessen the noise in the files.

[KeithB]

I know what you mean about shutter sync speed, at 12.5 microseconds, the actual response time of the flash comes into play. The shutter could be over and done before the flash begins to fire.

Bob Atkins measured the duration of a flash at about 1000 microseconds. But I don't know the time between when the flash is triggered and it fires.

https://bobatkins.com/photography/technical/canon_speedite_high_speed_sync.html

[robert e]

As K. Tanaka noted in the previous "Global Shutter" post, that flash sync speed and other specs may have more application for sci/tech research than for everyday photography.

I digress, but the term "global shutter" keeps reminding me of those proposals to cool our warming planet with some kind of artificial shade, whether space-based parasols or artificial clouds or something else. (It's a crazy notion for a number of reasons, but then, so is a 1/80,000 second flash sync speed.)

[Kye Wood]

If memory serves, the tech to build a sensor with practically unlimited dynamic range is fairly straight forward, but will never come to be.

This is because there is a company (who's name I won't mention for fear of legal action), that routinely patents ideas it has no resources to bring to fruition. It's business model is to register these patents and they sell them for truly epic amounts of money.

In this case, the patent covers resetting the wells once they hit saturation, then counting the number of times the wells have been saturated, to determine how much light they measured in total. So, no burnt out pixels in your image. Ever.

This model was working well enough, until there was a directive from tech companies from 'on high' to never engage with the company. I'd like to say, on principle. But I think it was just the outrageous amount of money and the percentage of royalties that they demand that did the trick.

I know it sounds very conspiracy minded. But it's not. And it's a shame. I'd love to have images and movies with zero unrecoverable pixels.

[Robert Roaldi]

The comment by Kye Wood is quite a condemnation of current patent law. To my mind, writing down that idea should not be patentable. It probably occurred to every engineer at every camera manufacturer. The idea itself is of practically no value, implementing it so it actually works has real value.

Good thing no one ever patented the idea of space travel. How about patenting the idea of heating a house using fire. Maybe someone should patent the idea that humans have ideas, then collect on every thought.

Allowing patent trolling is a bug in the culture. No one sane would want this. Why is no one fixing this bug? It's as if we have all forgotten why we're here.

[Tippler]

Patents don't last forever. They generally have a 20 year term, which in some cases can be extended. So a sensor with unlimited dynamic range can occur without patent induced inhibitions, if we wait a bit.

[David Dyer-Bennet]

This is precisely why, for so many years, "ideas" couldn't be patented, just actual implementations. It leaked in sort-of via computer implementations, which arguably ought to be patentable (the arguments against that I know center around how new and fast-moving the field is, rather than any claim that the algorithms aren't significant inventions).

[Jon Erickson]

Nobody "patents ideas" and then sells them "for truly epic amounts of money."

Ideas are not patentable, and there's no reason Kye Wood or anyone should fear "legal action" for citing patents just because the assignee is a non-practicing entity, sometimes referred to as a "patent troll."

Patents are public documents.