How small can and will system cameras go?

[BJL]

Discussion of this topic started in another thread, with these posts
http://www.luminous-landscape.com/forum/index.php?topic=65427.msg519427#msg519427
http://www.luminous-landscape.com/forum/index.php?topic=65427.msg519493#msg519493
http://www.luminous-landscape.com/forum/index.php?topic=65427.msg519499#msg519499
http://www.luminous-landscape.com/forum/index.php?topic=65427.msg519604#msg519604
http://www.luminous-landscape.com/forum/index.php?topic=65427.msg519710#msg519710
but maybe it belongs in a new thread.

My thinking is that the new mirrorless system cameras are far less tiled to backward compatibility with SLR lenses, instead being used primarily with the new lenses designed for the new systems, so there is less motivation to choose a formats size for "historical" reasons, and we are indeed seeing a variety of new, smaller formats becoming popular: 4/3" (Panasonic, Olympus), 1" (Nikon), 1/2.3" (Pentax Q). MAny guesses are that Canon will soon enough launch a mirrorless system, again with format smaller than in any of its SLRs.

My question then is what format sizes will be successful?

On one hand I suspect that the "APS-C" formats were chosen in part to work reasonably well with lenses from 35mm format SLR systems, since custom lenses for those systems came later, and the APS-C lens systems were not very extensive for some years. Without that constraint, and with these systems often marketed on god part for a size advantage relative to DSLRs, smaller formats seem likely to be popular. To me, the sizes of zoom lenses for the NEX system hamper its appeal to the great mainstream of photographers who want a couple of zoom lenses, not a bag of primes.

On the other hand, once sensors are about APS-C size or smaller, the need for adequate controls and a big enough LCD mean that even smaller sensors do not allow bodies to get smaller. Also, the costs of sensors of these sizes are a fairly small part of the total system cost, and there is little sign of a price advantage to smaller formats (Pentax Q, Nikon One) vs larger mirrorless formats (m4/3, NEX, NX). So lenses become the dominant size and cost factor. In principal, a larger format might be able to use about the same pixel size as a smaller one, and so through cropping, match the telephoto reach of a smaller format with a lens of the same maximum focal length, and so about the same size weight and cost, but with better IQ at the wider angle end of the zoom range.

Where will the balance fall?

[BernardLanguillier]

I believe that this question has to be considered in the context of how people use images, ie, how large they print if they print. Most customers of those systems do in fact never print. All they care about are well focused and correctly exposed images. Resolution is in fact not relevant.

Still, the Nikon 1 series we have today is good enough for excellent A3 prints and delivers an image quality in the same ballpark as the original 1Ds then referred to as a film MF killer, at least at lower ISOs.

So, assuming that we agree that the smaller the sensor, the smaller the lenses have the potential to become, I am not sure why we need larger. I agree than Nikon has not yet realized the potential of the system, but we have to think in terms of product release plan, and a very finely tuned one here. My guess is that Nikon could have released smaller lenses, but this is a system with all the implications in terms of attracting people, having them to start investing in the system and managing to have them spend more later.

I am not predicting here that the Nikon 1 series is the winner, I believe that 4:3 is a very appealing proposal thanks to the more or less "open" approach (more less than more unfortunately). I am just saying that they have, in my view, selected a very suitable sensor size that enables them both to go above the minimum acceptable quality for reviews today while having the potential to release smaller camara+lenses combo in the future compared to the competition.

They have also delivered a breakthrough product in terms of focusing speed, which again, hits right where customer have actual needs.

Cheers,
Bernard

[John Camp]

I'm not sure that there has to be a "winner" among the smaller formats with EVF, but I do think that the churning in the system could lead to some losers. I think APS-C could be one loser, and I'm pretty sure that the small Pentax system will be another, unless Pentax has enough sales in Japan to stay afloat. (IMHO, if Pentax-Ricoh had joined m4/3, that format would be *much* stronger than it is right now, and might have become the default mirrorless system; Pentax and Ricoh might also be much stronger. That could still happen if the Q system goes down without taking Pentax-Ricoh with it.)

I also suspect that the Nikon D800 isn't the end of the FF story. We'll eventually see 60 or 90mp FF cameras. I think MF may be toast.

(This post modified because of a typo in which I referred to the D800 as a D300. See the post below.)

I was one of the posters in the group referred to by the OP, and so will repeat only a small bit of that here: I see only two reasons for different camera-system sizes: (1) convenience, and (2) quality. FF systems are heavy and bulky, even the D800 and D700-sized cameras, but the IQ is brilliant. The m4/3 sized cameras will never match FF quality, because size will always count for something, but the quality will get better, and you can get a full multiple-body system in a briefcase, as I have done, and still have room for other things. FF will not be able to match these small mirrorless systems for convenience, even if the bodies shrink, because the lenses can't shrink much. So, you take your pick -- or, you buy both, as I have done. APS-C is in the inconvenient middle -- bulky bodies and lenses, quality inferior to FF.

[PhotoEcosse]

I also suspect that the Nikon D300 isn't the end of the FF story. We'll eventually see 60 or 90mp FF cameras. I think MF may be toast.

 

Ermm...  D300 is a DX camera. 

Going back to the OP question, technologically there is no reason to imagine that sensor size (and, hence, lens size) cannot continue to become miniaturised and, increasingly, produce high IQ from smaller and smaller rigs.

However, there is the downside of the handling qualities of very small cameras.

One of my reasons (not the only one) of "downgrading" from my D3s to my D800 was that the D3s was just too heavy and bulky for comfort.. On the other hand, however, I would not feel confortable trying to handle a very small camera. My PEN PL-3 is just about as small as I would want to go.

For me, eye-level shooting is a "must", which is why I added a VF-2 to the PEN. But I can't see any attraction to having a usable eye-level VF on a camera that only measured, say, 2cm x 1 cm.

If I could get D800 IQ from a camera the size of the PEN, would I go there? Don't really know until it happens. For the time being, it's horses for courses.

.

[AlfSollund]

Isnt the Nikon ! allreday below the limit where diffraction in lenses limits the potential IQ?

Seeing the welcome of the Nikon D800 and the "demand" in foras for Sony FF mirrorless I would say that FF is the way to go. I would imagine that Sony could squeeze a FF sensor into a body not much bigger than the NEX-7. On the other hand the APS-C is a good compromise, and one that can utilize lenses for FF.

Also a trend might be that consumers are becoming more wealthy on a global scale and demanding better quality (instead of using Apple products  ).

So putting together this means a slow, well earned and agonizing death for m43 and smaller, and a well earned triumph for APS_C, FF and larger.

[Petrus]

If we are talking about serious (landscape etc.) cameras, my view is that APS-C is just about the smallest sensor size possible. There is really no reason to go smaller, as when the size of the camera goes too small, it is not ergonomic to handle. Not enough space for the necessary dials and switches. Not everything is best done via touch screens and menus, we have seen a trend in pro cameras where more buttons and joysticks are added to the bodies, not less (Nikon D4). So if the body has to be certain size, there is no need to use too small a sensor with a loss of sensitivity and more problems with diffraction. So APS-C will be the small size winner, and FF the bigger size winner, as there is a long tradition of great lenses for that format and the size is quite optimal compromise between size and cost and IQ. MF will suffer, but still exist for diehard HQ photographers, and certainly we will see sensors crossing the 100 MP mark, maybe around 120 MP for 645 size backs in the next few years.

[Vivec]

I believe that this question has to be considered in the context of how people use images, ie, how large they print if they print. Most customers of those systems do in fact never print. All they care about are well focused and correctly exposed images. Resolution is in fact not relevant.

Still, the Nikon 1 series we have today is good enough for excellent A3 prints and delivers an image quality in the same ballpark as the original 1Ds then referred to as a film MF killer, at least at lower ISOs.

I think there is one aspect that we forget here that is important to most people: namely subject isolation. For family photos for example, most people love faster shutter speeds (ie. wide aperture) together with subject isolation (especially since the background is usually very cluttered in family situations).

A problem with anything smaller than m4/3 is that it becomes much harder to achieve good isolation. Of course, having more depth of field can also be an advantage in some cases. However, I feel that with anything smaller than m4/3 it becomes hard to get good isolation with a 35/50mm-eq lens, while anything larger than FF can make it hard to get enough DOF in other situations. So, m4/3, APS-C, or FF might all strike the right balance...  for anything smaller I think smartphones and Canon S100's will do just fine.

[BJL]

Thanks for the comments so far. A few responses:

Bernard: I agree that for the target market of "compact system cameras", pixel counts can be adequate in all formats, even if we all move to viewing on "4K retina displays". The IQ question then is noise in poorly lit padts ofnthe scene, whether that be shadow handling at low exposure index (aka dynamic range) or overall at hi exposure index. The latter ultimately comes down to effective aperture diameter, and none of the new CSCs is even pishing the limits there, instead going for zooms f/5.6 or slower at the long end. So I would say that already, the speed of these systems (thanks to the vastly greater sensitivity of these sensors compared to film) is satisfactory for a large proportion of mainstram "system camera" customers. And also to even many serious landscape photographers and such, who mostly work at low ISO speeds anyway.

So maybe dynamic range at base ISO speed will be the most visible weakness when the format gets too small, even for common scenes like a landscape with bright clouds above and deep shadows below. How does the Nikon One system handle that?


Vivec: I agree that shallow DOF options are important to many people, and the degree of need varies greatly, so for some photographers, this alone could decide how small a format is acceptable. Some feel the need for f/1.4 in 35mm format and so no smaller format will do; others like me are satisfied, just barely, with what a f/2.8-3.5 lens in 4/3 can do; yet others might not care in the slightest, just happy to get everything in focus and so content on that count even with Pentax Q.


Petrus: the smaller sensor format does not _force_ the cameras to be smaller; it just allows it, for those who want a small, simple camera like some of the NEX and m4/3 bodies. Bodies can be bigger than the format size might suggest, as with the Nikon V1 or Fujifilm X1 Pro. But it does mean that for most photography enthusiasts, the body size advantage does not improve below about APS-C ...

... And this leaves _zoom lens size_ as the biggest size advantage of a smaller format (or to be pedantic, smaller pixels and higher overall resolution in lines per mm), at least in the mainstream where the camera is usually carried with a wide-to-telephoto zoom lens attached, not some trendy little pancake prime.


PhotoEcosse: technology can only downsize lenses much more than they are now by reducing their (maximum) focal length, and for any significant telephoto reach, that means reducing the format size. Or at least, reducing the effective format size for narrow field of view (”telephoto") shots, maybe by cropping to a smaller format than that of the sensor.


John: though I eschew APS-C, and agree with most of your arguments, I expect that it will persist as the largest affordable format, due to the probably permanent and big price jump for the next step up to 36x24mm. But maybe only in SLRs, with mirrorless customers instead favoring more compact lenses, and thus smaller formats.


AlfSollund: no, Nikon's format is not close to being diffraction limited when using the lower end of its viable f-stop range. And when a smaller aperture must be used to get more DOF, the diffraction limits are the same as in any larger format when it is stopped down to get the same DOF, due to the larger format needing to use a higher f-stop to get that same DOF.

[hjulenissen]

AlfSollund: no, Nikon's format is not close to being diffraction limited when using the lower end of its viable f-stop range. And when a smaller aperture must be used to get more DOF, the diffraction limits are the same as in any larger format when it is stopped down to get the same DOF, due to the larger format needing to use a higher f-stop to get that same DOF.

Alternatively:
When system MTF is truly diffraction limited (as in trying to get a maximal DOF in a landscape image), I believe that sensor size does not matter in terms of MTF.

The practical caveat is that:
0. For many photographers, maximal DOF is a concern for only a fraction of their images.
1. "Image quality" is more than MTF
2. You ned a lense that can actually be dialed in to the aperture corresponding to the desired DOF without introducing significant other optical errors (a pinhole camera will always give lots of DOF, but never lots of details)
3. System MTF is to some degree always affected by sensel density and arrangement - stuffing 180MP into a cell-phone-sized sensor is yet not possible.

-h

[Petrus]

The practical problem with small sensor is diffraction. In that Nokia 808 phone-camera with 41 MP sensor the lens is f:2.4, which seems to be the diffraction limit with that size sensor at that pixel density. If I have understood correctly, there even is no adjustable f-stops in that camera. So there is a problem of not having usable f-stops for light control with small sensors.

Green light diffraction limit F-stops for 30 MP sensors of different sizes (interpolated from table #3 of http://www.luminous-landscape.com/tutorials/resolution.shtml ):

MF (36x48mm) F:11
FF (24x36mm) F:8
APS-C (16x24mm) F:5
4/3 (13.5x18m) F:4.5

If we take F:2 as a reasonable lens speed, different size sensors give 5 (MF) to 2.5 (4/3) stops of usable aperture adjustment before diffraction kicks in with same MP sensors. This makes small sensors (smaller than APS-C?) impractical for serious work. The there is the other side of the coin also, narrow DOF, which is often desirable also. Balancing all together I feel FF/35mm is ideal, APS-C usable sensor size, the later in the form of X-Pro1 and similar "street camera".

[BJL]

First, to clarify, my question is about how small will the smallest succesful interchangeable formats be, not any bizarre claim that all formats chosen by all photographers will shrink! I have no doubt that formats up to at least 36x24mm have many years of success ahead of them amongst sufficiently demanding photographers who are willing to pay several times more for a camera and lenses than most system camera buyers do.

Petrus:

On those diffraction limit numbers, f/4.5 is a completely viable value, so those numbers show that even 4/3" format can comfortably handle 30MP, which is far, far more than the great majority of mainstream system camera buyers care about. At the suggested limit of f/2, even Nikon's 1" format could go to about 80MP, way beyond mainstram needs. For most camera buyers, the current maximum of 24MP (NEX7) is already quite excessive, and for that, the limit is about f/5 in 4/3" and f/4 in 1" format; completely doable.

As to the talk of "range of f-stops", The large DOF options are essentially the same for any format (just with the large format having to use a higher f-stop, and so either a longer exposure time or a higher ISO speed), so that is just another way of saying that a smaller format is more limited in its shallow DOF options. The persistent popularity of f/3.5-5.6 zooms, and the fact that this same f-stop range has remained the mainstream favorite through the whole progression from 35mm film format down through APS-C to 4/3” to 1", despite the shrinking options for shallow DOF, indicates that the mainstream market for system cameras is far, far less affected by the craving for shallow DOF than are many participants in photographic forums. Instead, the market keeps voting for using technological advances mostly to allow smaller and lighter kits, with lens speed apparently valued mostly for what that nickname implies: an increase in the range of usable shutter speeds, not an increase in the blurriness of backgrounds.

This persistent preference for what more demanding photographic enthusiasts call "slow lenses" is probably due to the fact that any increase in shallow DOF options or low light performance comes through larger, heavier, and probably more expensive lenses, due to requiring larger effective aperture diameters and thus larger front elements (for example, through having the same minimum f-stop at a longer focal length as needed to get the same FOV on a larger sensor, since aperture diameter is roughly focal length divided by aperture ratio). A larger sensor with lenses of similar size and similarly sized front elements gains nothing in either shallow DOF options or improved low light performance, and the mainstream market seems to take account of that size/weight/cost/performance trade-off better than many forum contributors.


The bottom line is that as far as I can see, a very substantial proportion of mainsteam system camera buyers get all the resolution they need from a sensor of 10 to 20 MP, and seem satisfied with the "selective focus" options given by standard consumer level f/3.5-5.6 zooms along with a few faster options, and for them formats like 4/3" or 1" or maybe even a bit smaller are likely to offer the most appealing balance of size and weight against performance. On the other hand, there are valid arguments for the desirability of the larger formats, from APS-C to 35mm to MF, but probabably only relevant for smaller, more demanding sectors of the market, not for the mainstream of system camera buyers.

[AlfSollund]

The bottom line is that as far as I can see, a very substantial proportion of mainsteam system camera buyers get all the resolution they need from a sensor of 10 to 20 MP, and seem satisfied with the "selective focus" options given by standard consumer level f/3.5-5.6 zooms along with a few faster options, and for them formats like 4/3" or 1" or maybe even a bit smaller are likely to offer the most appealing balance of size and weight against performance. On the other hand, there are valid arguments for the desirability of the larger formats, from APS-C to 35mm to MF, but probabably only relevant for smaller, more demanding sectors of the market, not for the mainstream of system camera buyers.

Thanks for clarification on diffraction issues   

Your bottom line might be correct today, but I not convinced that this holds true for future consumers. I believe that mainstream will be driven towards even more resolution when they see their friends whooping 8 x 2 m 3-D screen at home displaying 200Mp images. And the formats like 4/3" or 1" will not be able to deliver. Jokes aside, my point is that some of these small format systems are not future proof taking into account developments. Abandoning these small camera bodies are one issue, but what about investments made on lenses and the pile of lenses not longer usable? Luckily I have not paid for this 

[BJL]

It is risky making predictions, but let me make one observation about the limits on resolution set by diffraction: they apply to our eyes too, and our eyes have a roughly 17mm lens with an aperture of about f/5 in good light such as when viewing an image on screen. With normal field of view covering about the same diagonal size as the focal length, that says that the central part of the retina, excluding peripheral vision, is about 1" format, suggesting that the resolution limits of the eye can be matched by 1" format with a f/5 lens, and comfortably exceeded by just moderately bright lenses in formats 1" or larger.

[ErikKaffehr]

Hi,

One way to see that is that there is a view that 12 MPixels is what is maximally needed for everyday photography. It is certainly possible to print A2 from an image with 12 million very good pixels. If we assume that the minimum pixel pitch that is reasonable is 3 microns we would than have sqrt(12e6) * 3e-6 > 10.3, say 12x8 mm.

I would imagine that for pixels to be very good diffraction would not exceed 1.5 pixel pitch (my experiment shows that we have visible decline for Airy ring diameter of 8.2 microns on  4.77 micron pitch sensor): http://echophoto.dnsalias.net/ekr/index.php/photoarticles/49-dof-in-digital-pictures?start=1

My guess is that the lens would be diffraction limited at f/3.5 which is certainly reasonable. A normal lens would have around 14 mm focal length, so I presume that the camera would be around 2 cm thick with such a lens.

Best regards
Erik

[aizan]

Pentax and Ricoh will probably go in opposite directions for mirrorless. Pentax's cameras will be toy-camera-esque, and they'll cede the making of utilitarian cameras to Ricoh.

I hope Canon makes a 2x sensor with a 2:3 aspect ratio. I wish the Nikon 1 system did that. What I like about the Nikons is the AF and the flat-block design. Those protruding lens mounts on Sony Nex and some m4/3 weird me out, though I like the deeper grips on the Nex-5 and 7. If Nikon makes a bona fide AdAm camera with a standard hotshoe, control dials, and rangefinder-style EVF placement, I'll seriously consider the system.

Back to Canon, I do not expect that their mirrorless system will be APS-C. Nikon and Sony are their two biggest competitors, and it makes sense that they will split the difference in sensor size. This way they will have a larger sensor than Nikon but smaller lenses than Sony.

On the other hand, I doubt that APS-C will continue to seem "too big" when Sony makes some collapsible zooms and a wideish pancake. I won't consider buying the Nex-7 until they either make a silver version or make all of their lenses in black. =)

Samsung already has some collapsible zooms, but they're like Rodney Dangerfield.

As for m4/3, I'd probably be a user if there was something that looked nice like the Panasonic L1. Did Panasonic fire their old designer or something? Don't get me started on the Olys.

[ErikKaffehr]

Hi,

Have you seen any decent quality pancake lens lately?

In my view it may make more sense to make a small camera with a small lens than a small body with a pancake lens. Very few of the present lenses for the NEX are or really good quality, and those that are come with a significant size. I don't really feel that 24 MPixels make a lot of sense without good lenses and good lenses are seldom small.

On the other hand, I passed over the NEX-7 and bought an Alpha 77SLT. After finding out that my 70-300/4.5-5.6G was underperforming on the Alpha 77 I upgraded to a 700-400/4-5.6G. So I guess I care not much about size.

Best regards
Erik

On the other hand, I doubt that APS-C will continue to seem "too big" when Sony makes some collapsible zooms and a wideish pancake. I won't consider buying the Nex-7 until they either make a silver version or make all of their lenses in black. =)

Samsung already has some collapsible zooms, but they're like Rodney Dangerfield.

As for m4/3, I'd probably be a user if there was something that looked nice like the Panasonic L1. Did Panasonic fire their old designer or something? Don't get me started on the Olys.

[aizan]

nope, not lately, unless the pentax 40/2.8 counts. it would be really cool if sony made something as good as the 40/2 zuiko, though it is unlikely.

i don't see the nex-7 as a "serious" camera, so i don't see much use for 24mp on a camera like this. if i need lots of megapixels, i would plan on using full frame dslrs. for casual purposes, i could shoot at medium resolution with a so-so pancake and not feel like i wasted money on all of those megapixels. to me, it's just a nice option, not an essential criteria.

i could also see a m4/3 or nikon 1 with a high quality prime, e.g., the 25/1.4 summilux, producing better results than the nex-7 and a new, so-so pancake. which just makes me pine that much more for a kickass panasonic L1-esque m4/3 or nikon 1. if it had a control interface like the leica s2, with the shutter speed dial and led screen on top, and a thumb dial on the back...now i'm just getting carried away.

[ErikKaffehr]

Hi,

I would see the NEX-7 as a serious camera, it's essentially the same camera as the Alpha 77. The Alpha 77 is pretty decent with good lenses. Also note that 7 in Sony/Minolta speak means below top of the line, so there is room for an NEX-9 in the naming scheme.

Michal Reichmann seems to use the NEX-7 mostly with a 16-200 zoom, which seems to be one of Sony's better E-lenses but is a bit on the big side.

By the way, the microlenses on the sensor are not optimized for the chief ray angle of the Leica lenses so using Leica ultrawides on the NEX may be a bit problematic (source Lloyd Chambers, mainly).

Best regards
Erik

nope, not lately, unless the pentax 40/2.8 counts. it would be really cool if sony made something as good as the 40/2 zuiko, though it is unlikely.

i don't see the nex-7 as a "serious" camera, so i don't see much use for 24mp on a camera like this. if i need lots of megapixels, i would plan on using full frame dslrs. for casual purposes, i could shoot at medium resolution with a so-so pancake and not feel like i wasted money on all of those megapixels. to me, it's just a nice option, not an essential criteria.

i could also see a m4/3 or nikon 1 with a high quality prime, e.g., the 25/1.4 summilux, producing better results than the nex-7 and a new, so-so pancake. which just makes me pine that much more for a kickass panasonic L1-esque m4/3 or nikon 1. if it had a control interface like the leica s2, with the shutter speed dial and led screen on top, and a thumb dial on the back...now i'm just getting carried away.