The Sony a7S III’s Missing Feature

The Sony a7S III’s Missing Feature

Between the Sony a7S III’s announced specs satisfying the five-year hype and its praise given in early reviews, it appears that the new camera is off to a great launch. However, I noticed a feature was missing.

It’s the same feature that’s also missing from the Sony a7 III and a9 II: Pixel Shift Multi Shooting. In this article I want to talk about why I think Sony’s strategy is wrong and why all these cameras should have it, especially the a7S III.

Pixel Shift Multi Shooting is currently only found on the Sony a7R III and a7R IV. Looking at the newest a7R IV with a 61-megapixel sensor as an example, Pixel Shift allows users of that camera to capture a 16-exposure series that can later be compiled on the computer to create a raw image that is about four times the pixel resolution: 240.8 megapixels (19,008 x 12,672 px). There’s also a 4-exposure Pixel Shift mode which doesn’t increase the pixel resolution of the image, rather it resolves much more fine detail out of the 61 megapixels. This latter 4-exposure mode is the only Pixel Shift setting on the a7R III.

If this feature was made available on the 12-megapixel a7S III, we could expect being able to capture final images around 48 megapixels of resolution. For the a7 III and a9, around 96 megapixels.

Before we get too carried away in this fantasy, I should point out the obvious drawbacks of Pixel Shift. For one, the camera needs to be locked down on a tripod or similar steady platform in order to fire off the 16 identical exposures without movement. Secondly, the scene in front of the camera needs to be still. Lastly, compiling the photos needs to be done off camera using computer software and takes extra time (although PixelShift2DNG speeds this up tremendously over using Sony’s Imaging Edge Desktop). For some situations, Pixel Shift isn’t a very helpful feature. But, I’d argue that being able to leverage its high-resolution capability some of the time (or all the time depending on your photography genre) rather than never makes it a very important addition and is sorely missed with a 12-megapixel camera.

Now, I know what you’re thinking: Pixel Shift Multi Shooting surely sounds like an a7-“R”esolution type feature. We’re talking about megapixels and fine detail and all that goodness; to the a7R it goes! And I have to disagree. Between the a7, a7S, a7R, and a9 cameras, Pixel Shift’s value in the a7R is really the least beneficial. Nothing makes that more obvious than the lower resolution a7S III.

For the a7R series, Pixel Shift is more of a novelty than a necessity. Now up to 61 megapixels, the a7R IV’s 16-shot Pixel Shift creates 240.8-megapixel files that are plain unwieldy to process and edit on my computer. The file size of an outputted Pixel Shift composite is about 1.45 GB before any editing takes place. They are greatly detailed and fun to comb through all the way zoomed in, but unnecessary. After trying it a few times, the novelty wears off and the ease of shooting still very high resolution 61-megapixel images wins out. The feature is wasted inside the a7R.

The point of the a7R series should be that you don’t even need Pixel Shift to create these photos. You just simply snap the picture with none of the drawbacks. The resolution is built in, whereas with other Sony cameras we would have to use Pixel Shift to achieve that level of resolution; It’s possible, but it’s more complicated.

When Sony develops new features, we often see them spread to all of their cameras. The a7S III should be able to shoot 48-megapixel Pixel Shift photos because it makes it a more viable and salable hybrid camera while still allowing the a7R to have undisputed purpose in the portfolio. There is no cannibalization, just better products.

Log in or register to post comments

29 Comments

Matt Williams's picture

Agreed that all of them - a7s, a7, and a9 series - should have it. Really hope Nikon implements it in the Z cameras. They could add it via firmware to the current Z6/Z7 even. I mentioned it to them among my long list of things I want either via firmware or future bodies via hardware.

It's a limited application feature, so I'm not sure why it would be reserved for only the most expensive body (well, I guess the a9 is the most). I don't really see people deciding between an a7III or a7RIV based on that one feature, so what could it hurt? Olympus has had it in all their OM-D cameras (and the PEN-F) except the consumer E-M10 series since the original E-M1 in 2013.

Just me's picture

First missing feature is 4KDCI mode.
Second is the fact that you can't really consider it as a "photo camera"...

Matt Williams's picture

Yeah I have zero idea why this wasn't included from the get. They said they "may" add it later, but like.... why not just do it now? Fujifilm, Panasonic, Leica, Canon... they all have DCI modes in a number of their models. For what is supposed to compete with the S1H and Canon R5 it's a very silly oversight (or intentional omission).

Just me's picture

You can't add pixel to a sensor.
The 12M size forbid you to go for resolution higher than 4K.
So DCI is not possible.
As report shows same overheating than with the R5.
suddenly, it's important to hold for ore real life comparison.

Karim Hosein's picture

My understanding is that the sensor is 4240×2832 px. 4k DCI is 4096×2160, (8.8Mpx apprx), IIRC. Therefore, the sensor is quite capable of 4k DCI.

The additional space is to allow for the possibility of Digital image stabilisation, and other image correction algorithms. Strictly speaking, they really only need a few extra pixels for tripod PixelShift, and improved BTA algorithm.

E.g., the Pentax K-3/K-3 II says, “Total: 24.71 Mpx, Effective: 24.35 Mpx,” yet, the largest file from the K-3 is 6016×4000 px, or 24.06 Mpx. This is because the image actually captured by the sensor is 14 pixels wider on each side, (6044*4028 = 24.35 Mpx approx.), to allow for edge adjustments to the BTA for more accurate edge colours, AND, the sensor also tilts to compensate for auto level, rotates & shifts for AstroTracer, vibrates for AA Simulation, and rotates & shifts for 5-axis IBIS, thus requiring additional imaging real-estate, to bring the total to 24.71 MPx.

To summarise, TV4k images are 8.3 Mpx approx., 4k DCI are 8.8 Mpx approx., and the camera is 12 Mpx, approx.

4k DCI is possible.

L C's picture

The read recently that if one outputs raw to an external recorder the output may end up capturing 4.2k.

Rick Knight's picture

Less pixels means capturing more light and more pixels means captures less light. But you can't do both.

barry cash's picture

That’s cause it doesn’t with Sony’s firmware

Alex R's picture

Ryan, first off, thank you for reporting on this. Was holding off on my pre-order till I knew this for sure. I agree it really should be included, but I'll take it one step further. I believe you've reviewed the EM1 III and/or EM1X, and it would seem Olympus has found a way to combine them in-camera. I'd think sony's new processor **should** be able to do the combination in-body, and it really would make the A7SIII a much better hybrid camera.

Matt Williams's picture

I thought most companies combined them in camera, so I was surprised to read that. Granted, I've mostly just used it on Olympus cameras, but I think many of the others do it in-camera. My five year old E-M5 Mark II can do it in-camera, so can the 2013 E-M1.

Jerome Brill's picture

I'm going to have to disagree. Pixel Shift seems great until you try to use it. Then you realize everything needs to be dead still to even take advantage of it. And the extra steps to process it. It's uses are very thin. Adding that to a dedicated video body doesn't make sense. Maybe it seems like it could use it at 12mp but again it's not something that would be missed for the 90% of people buying it. I understand why it's not there.

Now, if Sony added the ability to process it all in camera then that would be awesome. But I would want the final photo to highlight areas that had movement based on the first reference photo. That way I can take a series of photos until I get something usable. Less culling when I goto process later. If it actually worked like that, then yeah, it should be in every one of their cameras.

T Scarb's picture

That is a stupid feature to expect on this camera.

David T's picture

I don't think it's stupid, but let's say 10% of the buyers would actually want or use it.

Maybe they should just say it's a 50 dollar DLC (or free in the kit or a bonus for purchasing certain lenses) to satisfy demands while keeping cost low for those who don't want it.

John Adams's picture

You have no idea how they create those features and what resources it would take on the camera hardware so it's not smart to make such comparisons like DLC..

Herco le Fevre's picture

Apart from the fact that pixel shift in practice is far less useful than on a specs sheet, I think we're missing the point here. Sony has been working on a purpose-build line of cameras. The A9 is the sports-action camera for journalists and the A7S is the hybrid-but mostly video camera. Pixel shift on an A9 or on the A7S doesn't make sense. On the A7R's (and possibly the A7) it might make sense provided there are absolutely no moving objects in the picture. The whole debate about the 12Mp of the A7S is getting silly anyway. It was not designed for high-res stills. Horses for courses...

Some Fella's picture

This is a question I posted twice on Sony Alpha rumours website regarding the a7Siii it seems like a great option to cover the 10% of the time you want to have a high resolution photo. Olympus and Pentax both have this option in handheld shooting as well surely with the fast read out speed of the new sensor and chips one would think it would be possible and add great value to the camera.

Karim Hosein's picture

One has to think about the intended purpose of the camera. That being said, with only 12Mpx, there is a great deal of movement needed for tripod PixelShift shooting, to line up the four images. (…But possibly with less precision needed???)

As for hand-held pixel-shift, that is not the same thing. It was never the same thing. Hand-held pixel-shift takes four images, from four slightly different positions in space-time, then aligning them, (using data from the pixel-shift algorithm), to create greater spacial separation of objects in the image, creating a greater resolving power of the lens. (Same resolution of the sensor). [CLARITY] Hand-held pixel-shift does NOT align the RGBG pixels to create a non-CFA colour image, the way that tripod PixelShift does. [/CLARITY]

At 12Mpx, the lens is probably more than good enough, anyway. Not much improvement would be expected by hand-held pixel-shift.

Josh Leavitt's picture

I think the one missing feature of the a7S III is a built-in variable electronic ND filter. Give it one of those and it has 95% of the capabilities of the FX-9 at 1/3 the price.

Keith Mullin's picture

My understanding from the FX9 engineers is that the choice is either ND or IBIS, there isn't room between the sensor and the flange for both systems.

Josh Leavitt's picture

Makes sense. Sony's 18mm flange doesn't offer a whole lot of real estate for added features between the lens and focal plane. IBIS was probably the right choice between the two for a MILC camera body. Even without the built-in ND, it's still one hell of a video camera for $3,499

Karim Hosein's picture

Do you know what the one missing feature from almost all photoblogging websites? The knowledge on how to type, α on a keyboard. As in, “Sony α7R IV, α9 II, α7S III,” etc.

It is not that hard. I hardly ever talk about the Sony α cameras, and I can do it. They talk about them all the time, and I am still yet to see even one of them do it.

Even if one does not know how to do it on a normal keyboard with Windows 10, is it that hard to go to the Sony website, copy the stupid character, then paste it?

I don't know, man.

Yeah, that was my only takeaway from the article. 😉😆😁😀😄 (Being that I do not use Sony gear).

[EDIT] It would also be nice if photography YouTubers would learn to say, “alpha,” as in, “The Sony alpha seven ‘R’ three,” etc., especially the ones who (correctly) say, “It is ‘eye-so,’ not, ‘eye-es-oh’!” If one insist on saying one thing correctly, one ought to say it all correctly.

P.s., it is usually ‘exposure index,’ they mean, or ‘sensitivity value.’ Rarely do they actually mean, ‘ISO.’
[/EDIT]

Keith Mullin's picture

Literally the only people who use "alpha" even at Sony, are the project engineers, everyone else says "A".

Karim Hosein's picture

I find that the only people at Sony who say, “A,” are the salesmen, and they only say it if you say it first. One will never see “A” on the website, (or hear it from an exec†), except in reference to the lens mount.

†Well, maybe an American exec. Never from Sony, Asia. It has been “alpha” from the days of Minolta and the first α7000, α9000, & α5000, and that has never changed, (save for the brief, Maxxum, & Dynax, names in NA, & Europe, respectively). The mount, however, was always called the “A-mount.”

Keith Mullin's picture

But its E mount, not A mount (unless you are talking about the A77 etc).

Karim Hosein's picture

When the α series first came out, Minolta went from the SR-mount to the A-mount. The E-mount came during the Sony era, when they introduced the mirror-less system.

So now Sony has two mounts going; the A-mount, and the E-mount. I was simply saying that when the α-system was released, the name of the new mount created for it was, in fact, “A-mount,” and never, “α-mount.”

Bert Nase's picture

I not even use PixelShift in my A7RIII. But I like that they finally steal the great menu structure from Nikon! But nevertheless, it's totally overprized with 4200 EUR in Europe!

Gabriel Gonzalez's picture

This man does what you want without needing the function you are asking for
https://www.facebook.com/ezcurrabayerupscale
-
For those of us who make video and love the Sony S series for just keeping the 12MP and also keeping the 8.4 photosites the same size, the camera turned out as it should.

John Adams's picture

I don't think anybody ever said they want pixel-shift feature on A7S-III.. hence the reason it's not there.

L C's picture

It would be useful if the s3 could leverage the small files sizes and increased processing power to do it in-camera.