Camera Labs

Hi folks,

With Photokina behind us and comparatively few surprises (Sigma DP1 and Fujifilm X100 were highlights for me) I thought I'd take stock and have a think about where we might go from here.

My premise is that with the new breed of mirrorless cameras we've seen the last major innovation in the large (APS to APS-C) sensor interchangeable lens camera market. There'll be refinements and new players will enter the market but no more revolutions. The problem, if you can call it that, is that

1. Interchangeable lens camera bodies are already as small as they can comfortably go before holding them securely becomes an issue
   
   and in any event
   
2. Regardless of sensor size, if you want to achieve shallow depth of field then lens diameters can't go smaller

Here are three of my favourite mirrorless cameras, the Leica M9, Olympus E-P2 and Sony NEX-5, displayed approximately to scale:

The standout camera in terms of size is, in my personal opinion, the Leica because it's the only one to support a full-frame sensor but, of course, it's a rangefinder and doesn't do autofocus. While it is possible to argue that the space currently used by the rangefinder mechanism could be used by another manufacturer to house an electronic viewfinder (EVF) there's still a gotcha in terms of lens size. The Leica M-series lenses are pretty compact in terms of their maximum diameters because they don't house complicated electronics and motors to drive the iris and focus mechanisms. As an example the Leica SUMMILUX-M 50 mm f/1.4 ASPH has a length of 52.5mm and a maximum diameter of 53.5mm. By comparison the Canon EF 50mm f/1.4 USM has a length of 50.5mm and a much larger maximum diameter of 73.8mm. Looking at cutaway drawings of both lenses leaves a definite impression that the Canon's extra size has nothing to do with glass.

I'll return to lens size in a moment but first there are the Olympus E-P2 and Sony NEX-5 to consider. The E-P2 was used for illustration but size wise it's pretty much the same as the Panasonic GF1 and the Samsung NX100. These are all APS-C cameras though, surprisingly perhaps, the NEX-5, with the smallest body, also sports the largest sensor (same size as that in the NX100). I'd argue that the NEX-5 probably marks an end-point in terms of miniaturisation of the body as anything smaller may be difficult to hold securely, especially with a large lens stuck on it.

So back to lens size. It's well established that the larger the sensor the shallower the depth of field is for a given f-number but, as I tried to explore in my thread Full-frame Sensors Are Better - Fact or Fiction?, once one factors in the different focal length needed to achieve a given field of view it's a fact that Depth of Field (DoF) is, to a first approximation, determined by the diameter of the entrance pupil into the lens for a given subject. Some quick numbers can illustrate this. Using the DoF calculator here we can see that a camera with a crop factor of 2 such as the E-P2 has a DoF of 2.42m when used with a lens of 50mm at f/2 and a subject 10m away. But with a full-framer such as the M9 we'd have to use a lens of 100mm focal length for the same field of view and to get the same DoF for a subject at 10m we would use f/4. The entrance pupil of a 50mm lens at f/2 is 25mm and, oddly enough, the entrance pupil of a 100mm lens at f/4 is also 25mm. So DoF is dependant on the size of the entrance pupil regardless of sensor size.

Sorry to labour the point (especially for those of you who are already very familiar with this) but it is important in this context because it says that if you want to achieve a shallow depth of field for those artistic shots then the laws of physics dictate that lenses can't go any smaller and so, by implication, neither can the cameras they are attached to.

So we really have reached a limit now. The Sony NEX-5 is a marvel but there's absolutely no point in making a smaller interchangeable lens camera unless one is prepared to compromise artistically. Indeed if one wants a few extra knobs and dials the body could usefully be a little larger, especially if one wanted to incorporate an EVF.

That's my take on this but views differ so I'd be interested to hear what you think. Particularly if you think I've got it wrong.

Bob.

_________________
OM-D E-M5 + ED 12-50mm, H-PS14042E, H-F007014E, L-RS014150E, M.ZUIKO DIGITAL 45mm 1:1.8.
Gitzo GT1541T, Arca-Swiss Z1 DP ball-head.
Astrophotography: TEC 140 'scope, FLI ML16803 camera, ASA DDM60 Pro mount.

Thanks for that Bob, I enjoyed reading your views. I agree in that I like the depth of field effects that can be achieved with larger sensors/film.

Check out if you havent already Thomas' comparison of the Nikon 18-200mm and 28-300mm lenses on cropped sensors and full frame sensors to see in practice how much extra depth of field is gained with a cropped sensor with an "equivalent" sized lens.

_________________
Nikon D5000 and D300 with 12-24mm, 35mm f1.8, 50mm f1.4, 60mm f2.8 Macro, 18-200mm, 17-55mm f2.8 (all Nikon)

Hi Jeremy,

That would be Thomas' thread Comparing D700 + 28-300 vs D300 + 18-200 to save other members searching.

Just to be clear, I'm specifically arguing that in the context of this thread sensor size has nothing at all to do with how shallow depth of field can go. That parameter is, to a first approximation, determined solely by the maximum entrance pupil of a lens. The reason that shallow DoF is commonly associated with large sensors has to do with the fact that the smallest f-numbers are pretty standardised across a range of focal lengths and prices. A 100mm f/2 lens is something many can aspire to and it will produce a certain depth of field on a full-framer. But for the same field of view on a 2x crop factor APS-C camera you aren't going to find many 50mm f/1 lenses around at the same price point as the 100mm f/2 and that's what you'd need to get the same smallest DoF. Hence the perception that large sensors offer shallower DoF because in practice that's what you see with the range of affordable glass available.

However you work it for a particular scene and the same angle of view then a particular depth of field requires a lens of, effectively, the same diameter regardless of the sensor size behind it. The only way to make camera systems smaller now is to use smaller lenses and if you do that then shallow DoF has to be compromised.

Bob.

_________________
OM-D E-M5 + ED 12-50mm, H-PS14042E, H-F007014E, L-RS014150E, M.ZUIKO DIGITAL 45mm 1:1.8.
Gitzo GT1541T, Arca-Swiss Z1 DP ball-head.
Astrophotography: TEC 140 'scope, FLI ML16803 camera, ASA DDM60 Pro mount.

Over two years ago I suggested 'synthetic DOF' to become in use as a realistic alternative/addition to small(er) sensor cameras. I know that at least the Fuji S200EXR and F70EXR have implemented a synthetic DOF by taking multiple shots and combining them in-camera. This requires the subject to be more or less stationary for a short time and limits its use but as technology progresses more advanced forms of synthetic DOF, both in-camera and in PP, will quite likely be developed.

Ben
_________________
When in doubt..... Press the shutter.

I had to do a search to find it again but I was aware the needed technology basically exists already. One at least is the Omni-focus camera (photos here) and I think there is also another technique that I haven't yet been able to find again which allows focus in a selectable vertical plane after the image is shot. Once all the image and distance information is there the rest is up to the processing. This was discussed in the thread from 2008 which is here.

Ben
_________________
When in doubt..... Press the shutter.

Hi Ben,

Thanks for the links. If I've read it right "pro focus mode" is nothing more than a selective Gaussian blur and that sort of thing can be achieved in PhotoShop. Never as good a result as the real thing, though, as it's a constant blur effect regardless of the distance of the background objects, something that popo referred to in this post in the thread you linked to.

Whether contrast based AF would be up to the job of providing enough distance mapping data for the whole scene in a timely manner is above my pay grade, I'm afraid. Nice if it could, though, as it would seem to be a way around the minimum lens diameter issue.

Bob.

_________________
OM-D E-M5 + ED 12-50mm, H-PS14042E, H-F007014E, L-RS014150E, M.ZUIKO DIGITAL 45mm 1:1.8.
Gitzo GT1541T, Arca-Swiss Z1 DP ball-head.
Astrophotography: TEC 140 'scope, FLI ML16803 camera, ASA DDM60 Pro mount.

Hi Bob.

It's one of many interesting subjects about digital photography.

Obviously Fuji's 'trick' is a comparatively low grade form of synthetic DOF but it's a start. Digital photography for the masses is basically still in its infancy and no doubt there's still a lot to come but Fuji shows it's something they are looking into.

The Omni-focus camera is kind of a proof-of-concept that is capable of gathering both sharp image information from a sufficient amount of planes to allow a sharp image in every plane and distance information of every pixel. Once you have that the rest is up to the software. Once the software is sophisticated enough you can emulate every lens' characteristics because these characteristics can be measured/calculated and therefore copied through data manipulation by software. The bokeh resulting from different shapes and number of aperture blades and lens elements and whatever can be calculated and an image can be rendered that will ultimately be identical to the image actually shot with the lens you emulate once the technology has reached the necessary level. As I said in the 2008 thread. Once you're dealing with digital information you can manipulate it in ANY way you want. No limits! This will not happen overnight of course but think of the size and cost and capabilities of the early sensor based TV cameras compared to current technology. What differences will another 50 years bring us? I don't doubt for a moment that difference will be impressive.

The question is not when the level of technology will have reached that theoretical perfect level but when the level of technology will be able to render an image that for most of us will be 'close enough'.

Ben
_________________
When in doubt..... Press the shutter.

On Bob's original questions, I guess I agree existing EVILs are getting to the point where there isn't much scope for getting smaller. The NEX are pretty small now I've seen them in the flesh.

The 2nd point... arguably we're getting to the point of "good enough" synthetic bokeh. I believe not only Fuji are doing it, but Sony also. Maybe others too but I've not paid too much attention. They do something akin to reverse focus stacking, varying the focus to determine which areas are of interest or not and selectively applying blur. Arguably, if the scene is static enough, you could use the change of blur with focus to estimate the distance at a point. Compare it to contrast AF.

I can't remember the name but didn't Adobe demo a sensor that essentially gathers depth as well as spatial information, and with processing you can choose what's in focus or not. However, it doesn't come free. Putting aside the processing, it was very low output resolution. So for a given sensor size, you're trading output resolution for depth information.

I do think the long term future of imaging will store depth information as well, but it isn't something that exists nicely yet.

On a tangent, I have been wondering for a long time if there is sufficient value in an ultra-compact EVIL camera system. The further size reduction will come from using a smaller sensor, perhaps 4x crop factor. That will allow lens sizes to go down even further at the cost of reduced high ISO performance and natural bokeh ability. But it will be genuinely pocketable unlike the current breed. There's many shots I've missed because the E-P1 is big enough I don't carry it in summer. It fits my winter clothing fine, but it needs to be far smaller for summer wear, when on balance the weather should be better too... mitigating the smaller sensor to a degree.

_________________
Canon DSLRs: 7D, 5D2, 1D, 600D, 450D full spectrum, 300D IR mod
Lenses: EF 35/2, 50/1.8, 85/1.8, 135/2+SF, 28-80 V, 70-300L, 100-400L, TS-E 24/3.5L, MP-E 65, EF-S 15-85 IS
3rd party: Zeiss 2/50 makro, Samyang 8mm fisheye, Sigma 150 macro, 120-300 f/2.8 OS, Celestron 1325/13
Tinies: Sony HX9V.

Hi Bob,

that is a very interesting discussion you have begun there. Certainly I will not argue the logic you employ here - but I would say that it is constrained by the paradigm you operate within.

If we start with DOF and lens diameter, there is of course the possibility that an identical effect - 100% identical - could be developed using software and processor. From a logical standpoint, it would be a question of computerized application of a rule set that adequately encapsulate the what happens optically with a given lens. Not suggesting it's easy by any means, but from the perspective of the potential of the technology, I do not see it as a far stretch.

Secondly: you articulate that DOF is a function of the relation between the field-of-view and the aperture diameter, but isn't it true that if you could reduce the sensor-size and not sacrifice quality, you could maintain the same relation between them in a smaller form factor? If this is true, the miniaturization becomes dependent on sensor-development, would it not?

Thirdly: looking at the trend where cameras double as video-cameras as well, one could imagine that the form-factor would begin to resemble mini-cams more and more. Essentially where the lens is oriented differently on the body.Considering this along with improvements in both power-consumptions ratios and battery-capacities, this could lead to smaller form factor as well. Paradigmatically speaking, this could lead to a whole new way of interfacing with your camera..maybe voice-control or something else...thus saving space as well.

Lastly: the re-emergence of "3D" imaging has seen Fuji develop dual-lens compacts. And there are developments underway with 3D TVs (that doesn't require glasses) and picture frames (that also doesn't require glasses. 3D imaging is a new way of approaching DOF altogether. Intuitively one might argue that dual-lenses by definition will lead to larger bodies, but from the perspective of the potential of this technology, one could imagine that interpolation of the light captured by two small lenses could lead to a more acute DOF than one larger lens.

Obviously this is all speculation on my part and who knows where it will lead - but in consideration of the excellent topic you have opened Bob, I thought I'd be as creative as I could..lol.

Cheers

_________________
Nikon stuff and other stuff, but not nearly as much stuff as I really want!

I prefer the Panasonic mirrorless camera version better,sexier & better featured. (&...more expensive)

I would love to see in the future a full-frame camera in a mirorless body...Panasonic or Sony could make this step,both of them introduced many important innovations already. And Leica would still keep most of their fans because the mirrorless camera is very different from a rangefinder.

_________________
Gear: Nikon D80, 18-105mm, 35mm F/1.8, 85mm F/1.8, SB-600
Visit my work here

Hi LahLahSr,

I'll certainly agree that my paradigm limits my conclusions, particularly in that I've chosen DoF as the arbiter of artistic merit. And DoF is a right minefield as it depends on so many factors and assumptions. However I'm content that if one is considering all the variables as constant except sensor size and lens entrance pupil then only entrance pupil affects DoF. But I'll certainly concede, as I did in my reply to Jeremy, that in the real world where one is concerned with the affordability and practicality of lens production one can't ignore sensor size.

I really like your idea of using a stereoscopic pair of lenses and, by implication, sensors as it solves the issue of depth mapping that Ben was talking about without complicated measurements during focussing. The only drawback would be that to get a reasonably good measurement of depth I think you'd need two lenses and two sensors if the sensor size goes really small and that becomes a nuisance for normal lenses and impractical for telephotos. But it releases the DoF constraint on lens size completely so long as the lenses and sensors can be positioned a reasonable distance apart and one is back to constraints we are quite familiar with in compact cameras, namely diffraction limiting and high ISO noise - maybe it's actually a technique more suitable for compact cameras rather than the current mirrorless crop? And they can be marketed as 3D capable. Brilliant stuff!

Bob.

_________________
OM-D E-M5 + ED 12-50mm, H-PS14042E, H-F007014E, L-RS014150E, M.ZUIKO DIGITAL 45mm 1:1.8.
Gitzo GT1541T, Arca-Swiss Z1 DP ball-head.
Astrophotography: TEC 140 'scope, FLI ML16803 camera, ASA DDM60 Pro mount.

The stereoscopic view provided by a second offset image could indeed quite possibly also be a source for distance metadata to be used for calculating synthetic DOF. I guess it would simply be triangulation of every pixel (if sensors would both be the same size). Why not. Provided it's accurate enough of course. It would be interesting to see whether developments on the 3D camera front will show any indication of (also) going in that direction.

Ben
_________________
When in doubt..... Press the shutter.