Camera Sensor Size

One of many good illustrations from the article at Gizmag

The best, most thorough, and best-illustrated explanation of sensor sizes on the Web that I've come across is an article at gizmag.com by Simon Crisp from 2013.

The only thing the article might be said to lack is a short section making it more clear how big a "full frame" sensor actually is. Those of us who developed tens of thousands of frames of 35mm (yo!) have a pretty good grasp, but many people trying to sort out the bewildering welter of digital sizes might need some more intuitive illustrations in addition to being given the size in millimeters. Most illustrations showing relative sensor sizes enlarge the full-frame rectangle so that the smaller sensors tend to seem bigger than they really are.

Otherwise, this struck me as a very good "primer" type article. (Know anything better?)

Balance?
The Fuji press conference got me to thinking. Reader Doug Thacker reported that a Fujifilm executive in Tokyo responded to his question about full-frame by saying "X-Pro already big enough," and Fujifilm claimed in the press conference that its new 24-MP sensor out-performs 36-MP Bayer sensors (a claim that begs to be tested—I hope our friends at Imaging-Resource and dpreview will be on this). So apparently Fujifilm has decided to stick with APS-C for the time being. Pentax, meanwhile, must feel constrained by being APS-C only (as Nikon was, before it went full-frame in a big way starting in 2007), if the continuing rumors of an impending FF Pentax are anything to go by.

One of the most beautiful tall ships ever built, the British tea clipper Cutty Sark was designed to balance speed and cargo capacity. Photo by Cutty Sark Captain Richard Woodget, from the Allan C. Green collection of glass negatives at the State Library of Victoria in Melbourne, Australia.

Sensor size is always going to be about balance. It's a classic case of "opposing parameters," an old and familiar problem in design. The quintessential illustration of an opposing parameter is a sailboat hull. The more slender you make a sailboat hull, the faster it can go, but the less cargo it can hold. So if you maximize cargo capacity, you're going to take a hit in speed; and if you maximize speed, you're going to take a hit in cargo capacity. You can't optimize both at once. There are lots of examples of opposing parameters among design of many kinds.

Where sensors are concerned there are cases where you might want to depart from balance...for instance, very small sensors will fit in cellphones, and very large ones might be best for image quality in large prints (viz. the Leica S). But the ideal balance for all-purpose applications is naturally going to be a compromise between extremes.

I'm a fan of sensors in the 1" to APS-C range. Mainly, those are:

• APS-C (23.6 x 15.6 mm),
• 4/3 (17.3 x 13 mm), and
• 1" (12.8 x 9.6 mm).

Those seem to me to be the sweet spot, affording the best overall balance of good depth of field (with shallow depth of field still easily achievable if desired), good low light performance, excellent overall image quality, reasonably economical fabrication cost, associated camera and lens size, print size potential, and cropping potential.

In photography there are a number of technical "ranges," and in many of them there's a useful concept called the JND or "just noticeable difference," most notably applied to photography by C.E.K. Mees and Loyd Jones in their seminal perception experiments at Kodak starting the 1930s. To notice a difference between angles of view, for instance, the rule of thumb is that there should be at least a 15% difference in focal length; that is, you can't tell the difference just by looking at pictures between a 50mm-e lens and a 45mm-e lens.

My belief is that the "JND" in sensor size is about four places on the above chart (if you eliminate the double entry for APS-C). That is, you're not generally going to be able to tell the difference, in pictures, between two sizes that are next to each other on the chart; you have to make a bigger jump than that. A possible exception is the jump from APS-C to FF; some people certainly claim to be able to tell the difference from pictures. I'd like to see that tested perceptually, myself.

Another thing I'd like to see tested perceptually is whether people can really tell the difference between APS-C and Micro 4/3 in prints, "all else being equal." I doubt they can, however much sway the idea of the difference may have on their minds. APS-C and Micro 4/3 aren't sufficiently separated that they cross the "noticeable difference" threshold.

Mind you, all such statements are with "all else being equal," and, as Ctein often points out, all things are never equal. But there are advantages in that. At each generation, for example, sensors in the APS-C to 1" range lag behind full-frame and larger sensors, globally speaking. But, as I explored for myself recently, with a few years (say 5–8), the smaller sensors surpass their larger neighbors. It's ironic that as full-frame continues to find great popularity with dedicated photographers, APS-C sensors have already surpassed the earliest FF sensors. So at what point do we simply stop minding being a few years behind? Probably pretty soon, as sensors have gotten so good. There just won't be a clear enough real-world advantage to push people into larger sizes.

Two of the nicest 1"-sensor cameras, the hugely popular RX100 IV pocket camera (top), and the do-it-all RX10 II (bottom)

I personally prefer APS-C to full-frame—right, all else being equal!—but my guess is that as sensor development proceeds, eventually 1" sensors will be enough for "serious" interchangeable-lens cameras with single sensors. It's large enough to get focus separation with longer focal lengths, and small enough to optimize efficiencies in fabrication and lens design. That's just a guess, of course. Already, however, 1" sensors are fine for all Web work and for smaller prints.

Two interesting questions, both beyond my capacity to answer: what stage of APS-C development do today's best 1" sensors equal or exceed, and, how long before 1" sensors get as good as today's best FF sensors? Might happen sooner than you think.

Mike

P.S. Didn't someone once make a keychain with the sensor sizes marked on it, so you can carry a real-life size guide in your pocket? I thought so, but my memory fails me, and Google was no help.

Ctein adds: Nice summary. A couple of useful technical points:

1. Almost all metrics for image quality scale with the linear sensor size, not area. Comparing area gives a vastly inflated impression of the difference in quality you'll see moving between formats.
2. The JND for resolution is 12–15%, so the JND for pixel count is 25–30%.

Both of these presume "all other things being equal," which they never are. But it's a starting point.

Original contents copyright 2016 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.

TOP's links!

(To see all the comments, click on the "Comments" link below.)
Featured Comments from:

Phil G: "The size comparison is indeed interesting, but whilst it is true that traditionally a 1" sensor was 12.8 x 9.6 mm with 4:3 aspect ratio, the current Sony, Panasonic and Nikon 1" cameras have a 3:2 aspect ratio and thus a 13.2 x 8.8 mm image size."

Phil Service: "According to data that I extracted from DxOMark, the 1" sensor of the the RX100 III is very similar to the APS-C sensors of the Nikon D70 and D50 in terms of noise and 'low-light ISO.' The R100 III lags a bit behind the D300, which was released in 2007. Of course, the RX100 is a 20 megapixel camera, whereas the D70/D50 are only 6 MP, and the D300 is 12 MP. So, if the noise and ISO data are normalized for photosite area, the sensor of the Sony far exceeds that of the Nikons. The data for these and some other cameras can be seen here—see Table 2. Just my two-cents-worth."

Chuck Albertson: "Cool! They named that boat after my favorite Scotch."

That's very much in error.

It's "ship," not "boat." —Mike the Ed. :-)

Thomas Paul McCann adds: "Lord Mountbatten was off Cowes on an exercise with the British Fleet. Within his hearing someone asked his mother what was he doing at the time. 'Oh, he's off playing with his boats,' she said. To which an unamused Lord Mountbatten said, 'they're not boats, Mother, they are ships.' 'What's the difference?' was his mothers response. 'You can put a boat on a ship, Mother, but you can't put a ship on a boat,' replied Mountbatten."

Kevin Purcell: "The obligatory 'type notation' explanation: it's the diameter of the vidicon tube's glass envelope that would contain a sensor of that size. It's about 30% larger than the sensor diagonal dimension. It's a standard in the business even if obsolete. I'm sure camera marketers love it.

"A type 1" sensor is the same size sensor that would be in a 1 inch diameter vidicon tube. It's about the same size as a 16mm film frame.

"The 4/3 sensor is so named because it's type 4/3" (and it's also 4:3 aspect ratio).

"The APS-C sensor could be called type 3/2" or type 1.5". That doesn't sound much bigger than type 1". It's about the same size as cinematographic 35mm film frame or a still photographer's 'half frame' (but those are both 4:3 aspect ratio).

"The 'full frame' 36mmx24mm sensor would be a type 2.15". Not as impressive as 'full frame.' It is the same size as two adjacent cinematographic 35mm film frames."

Sylvain G.: " I remember taking this test, months ago: Probably not an authority, but still funny."

Mike replies: I got 70% right. Maybe it's easier to tell the difference between sensors than I thought. Most of the ones I got wrong were Micro 4/3 vs. APS-C, and as I say I really don't think there's a lot of difference between those.

Moose: "Some months ago, all the enthusiastic praise of the Fuji X cameras and their wonderful IQ started to nag at me, with my puny Micro 4/3 bodies.

"So I grabbed some Raw files of the standardized studio shots from DPR and IR reviews, converted them and stacked them on top of each other in PS.

"My conclusion, after way too much pixel peeping, trying various deconvolution settings, and so on, was that, at base ISO, it was an absolute toss-up between the X-T1 and the Oly E-M1, the latest of each line at the time.

"There were parts of each test image where one or the other had a tiny edge, to my eye. I'd give the slightest of edges to the Oly, but that may be my priorities and prejudices and most certainly wouldn't be visible in any but gigantic prints, if then.

"I didn't check higher ISOs; that gets amazingly tricky and starts to depend a lot on the Raw converter, NR and sharpening/deconvolution tools, etc.

"Nor did I check DR, although both easily handled the step charts.

"Then again, for some uses the HR mode of the E-M5 II transcends its sensor size in both resolution and color accuracy. In my limited testing so far, it also handles highlights better than standard mode, so may have better DR?

"Just another data point...."

Comments

[Kurt Kramer]

YES! Thank you. Finally a visual display of the plethora of sensor sizes.

[Thom Hogan]

Technically, full frame, APS/DX, m4/3, and 1" are all very close to one stop apart from the neighboring size. So the question is this: can you see one stop difference?

The one stop difference can come up in practice, particularly where DOF is concerned (assuming you're controlling perspective, angle of view, aperture, etc., the same).

Thing is, the "good enough" bar keeps going down in terms of sensor size. As I've noted, the current Nikon 1 cameras (1") are as good as the older Nikon D80 (APS/DX).

[Jack]

It would be good to get desensitized to sensor size preconceptions and prejudices, and focus on the final image output (e.g., phone screen or billboard). The end use justifies the means. (ATB=)

Like the old saying goes, once you have what you NEED, the rest is just showing off.

[Carsten Bockermann]

>>but my guess is that as sensor development proceeds, eventually 1" sensors will be enough for "serious" interchangeable-lens cameras with single sensors<<

But as manufacturing sensors gets cheaper all the time, there might be no real reason to use one that small. Of course, smaller cameras and lenses are always nice, but the size of our hands poses a limit to further miniaturization.

[[email protected]]

Printing is an amazing image quality equalizer when done well. I was admiring some prints a guy was selling at an art fair and he pointed out several he had taken with his iPhone 4. They were hanging right next to his full frame prints and moving well. I would never have guessed, even with that puny sensor and such limited dynamic range and horrific on screen appearance at 100%.

[Glenn Brown]

All very confusing,,,for me APS-C is 35mm film, full frame sensor is medium format and medium format sensor is 4x5 sheet film. With in between sizes available.

[Geoff Wittig]

Just my two cents, but it seems intuitively obvious that the real determinant of sensor size 'sufficiency' is the largest print you want to make. As of 2002, the initial full frame digital SLR generation (for example Canon's Eos-1Ds) could just about match the image quality of a really well scanned 35 mm slide. The digital image was smoother in terms of tonality and grain though a bit lower in resolution, for a net wash in image quality. So you could print (depending on your personal standards for print quality) to maybe 12x18" with no compromises, and maybe up to 20x30" with some acceptance of minor artifacts.
Each generation of sensors has gotten better, to the point where I find with a 50 mp full frame sensor I can print a no-apologies 24x36" image. That's as big as I'll ever need to go, so we're at the point of sufficiency for me. The equivalent generation APS-C sensor, to my eyes at least, can't go past about 20x30", again depending on your tolerance for resolution limits and artifacts.
If you're never going to print bigger than 20x30", then APS-C seems the sweet spot. And the micro 4/3rds sensors, per the 'JND' concept, are likely to be close enough, and more than big enough for smaller prints.
Another generation or two, and every sensor will move up another print size or two, making the larger sensors potentially less relevant, except for bean-counters who continue to focus on the numbers rather than the actual photograph.

[Miserere]

I long ago gave up on sensor size discussions, either online or in person, because sizes have become a religion.

I'm of the opinion that I want a camera with the smallest sensor I can get away with in terms of image quality. The day I find a compact, fixed lens camera that can produce IQ at the level of today's Fuji X series (and full PASM control and RAW files), I might very well switch over.

I've not tried Sony's RX100, but that's along the lines of what I'm thinking of. Already I've been using a Canon S90 for a few years and made some lovely photos with it; which, I should point out, no other camera would've made for me because I wouldn't have been carrying it. And that's why portability is near the top of my list of needs.

So let me rephrase my statement above: The day my carry-everywhere pocket camera can take pictures like my Fuji X series, will be the day I stop needing a larger ILC.

That will also be the day the marketing department goes into overdrive hyping the superiority of the "huge" APS-C sensors. The irony will most likely be lost on them.

[Dennis]

A few thoughts on sensor size:

You mention "all else equal" and one place where all isn't equal is in lens selection. Legacy mounts (particularly Nikon & Canon) have extensive FF lens selections, while their APS-C lens selections are lacking in lenses that some people find important. Meanwhile, micro 4/3 offers a great lens selection, and one that (at least in many cases) does exploit the sensor size, offering reasonable compromises between speed and compactness (Oly's f/1.8 line of primes).
Stepping down further, I think 1" sensors are best suited to fixed lens cameras like Sony's RX10 and RX100. Nikon's '1' series offers f/3.5-5.6 zooms, which aren't terribly exciting on a 1" sensor, and a fast prime, like the 32/1.2, is expensive and doesn't quite offer the shallow DOF of an 85/2.8 on FF. It feels very much like trying to pound a square peg into a round hole. It's has great niche uses, like a lightweight birding kit with the sharp 70-300CX.

I've seen nice prints from Fuji 2/3" sensors (20x30"). Dynamic range is always the bugaboo with small sensors, even in good light where you can shoot base ISO to avoid noise. (Sony's 1" sensor is pretty good). I recently saw some samples shot with iPhones using the new Zeiss add on lenses. They were nice and sharp, showing waterfalls that were blown out blobs of white. Small sensors require work to deal with the compromises.

Meanwhile, I've never used a FF digital sensor or larger (though I have shot 6x6 film). So I can't comment on whether I'd appreciate something bigger than APS-C.
I like 2/3" or 1" for compacts (including relatively compact do-it-all like RX10). 4/3 or APS-C for an ILC. And FF, I consider the medium format of digital, only more affordable, so for all of us who used to dream about medium format systems we didn't need, FF has a more attainable allure. (Time will tell whether I resist). A lot of people practicing photography have no issue with using large systems.

As for the article, it seems decent, but focuses a little too much on the actual sizes and not enough on what those sizes mean to the target audience. It's okay to see how much bigger one sensor is than another, but how much "better dynamic range" does one sensor give ? In other words, it says "bigger is better" but offers no guidance when it comes to deciding how much bigger to buy. But I appreciate anything that gets the message out to consumers that sensor size is more important than resolution (and I'm not aware of a better article).

[JK]

I'm confident that the only difference one could see in prints from APS-C vs. m4/3 would be if you printed them, excuse the expression, full-frame and so you'd see the difference in aspect ratios.

When students ask I always tell them that if they like the 3:2 proportions, then go APS-C. If you prefer 4:3, you lose nothing significant in terms of image quality with m4/3. As someone who has always found 3:2 too oblong, I know which I'd pick.

[Torsten Bronger]

Crisp article contains the usual myths, especially with respect to noise and low-light performance. https://en.wikipedia.org/wiki/Image_sensor_format#Dynamic_range gets it right.

[Rick in CO]

Mike, you don't need a key chain, just take a pix with your cell phone. (Can we unlock and start our cars with our cell phones?)
The issue of full-frame vs APC is very interesting to me. The Leica M9/M 240 and the Fuji X-Pro 2 bodies are very similar in size, the Fuji being slightly lighter (more plastic?) The Fuji has autofocus with a "hybrid" finder, the Leica has manual focus with a, to me, very clear and accurately focusing viewfinder. Fuji lenses, from my experience with the 6x9 Pro cameras can be among the best I've used. Leica lenses, and the ZM Zeiss, are also among the best. A Fuji f1.4 APC lens would have the equivalent depth of field of an f2 lens on full-frame. I use the selective focus of my f1.4 Summilux for only about 25% of my pix, but its very nice to have.
Questions: Everything being the same size (travel photos are my thing, so a compact outfit is required) how do actual larger prints (say 13x19 to 16x20 inch) with the Fuji APC sensor compare to those taken with the M 240? How fast and accurate is the Fuji autofocus? Can one fix the point of focus, ie. by slightly depressing the shutter release or some other simple and fast way, for hyperfocal shots to maximize depth of field? Does the "hybrid" finder provide a clearer and more life-like view than the EVF finders? At 61 my eyesight is still very good, but who knows what will come to pass. The Sony alpha 7 cameras are a non-starter for me, as are any camera with an EVF finder which I find completely inadequate to visualizing my photographic subjects.
I guess I will need to get a loaner when/if supply catches up with demand. I'd be interested to hear you opinion though.
Thanks in advance,
Rick in CO

[Eduard de Kam]

When small sensors get better because of the technical advances, the bigger one's will get the same improvement. Unless of course, and that would be an interesting option, when special progress were to be made because of small sizes.

[BH]

You're right, sensors are becoming excellent across the board. I'll do you one better and say the 2/3 sensor in the Fuji X30 is really under appreciated. That camera sports a fast 28-105 (?) equivalent lens and in my opinion is much nicer to use than the RX100.

It really comes down to how you shoot and what kind of depth of field controls you really need. There are times when I wish I could get more depth of field out of a larger sensor, and times when I wish I could get shallower depth of field out of a little camera. But I'm finding it harder and harder to care about the actual image *quality* coming out of a sensor, no matter what the size.

[robert e]

"many people trying to sort out the bewildering welter of digital sizes might need some more intuitive illustrations"

I just realized that a standard SD card is about the size of a full frame. Most people interested in camera sensors have probably seen one. (Kind of surprised me, actually--I thought the card would be smaller than FF.)

That Gizmag article is certainly the prettiest one I've seen on this topic. I like orange, so I guess I'll stick with APS.

But, seriously, the chart seems to correlate with JND as well. APS-H resembles a full frame slide mount window; the two APS-C are nearly indistinguishable; and so on in pairs, until we get to the bottom row, which we camera junkies don't care about. (What about Foveon--is it the missing link?)

This older engadget article isn't as colorful, but takes a historical perspective, for those into that. I also prefer their version of the chart (it's down near the end) for its more neutral drabness.

http://www.engadget.com/2011/12/16/engadget-primed-why-your-cameras-sensor-size-matters/

[Bruce Bordner]

As a wide-angle freak, I felt constrained to use full-frame due to optical reasons. My most-used lenses are 14mm and 18-35mm. There are now some choices in APS-C, but the low end is just physically hard. Good for telephoto, but I don't do much of that.

And don't get me started on DOF...

There are optical/physical reasons why we need a range of sizes - but maybe not so many...

[Jim]

"a Fujifilm executive in Tokyo responded to his question about full-frame by saying "X-Pro already big enough"

Or not:

http://petapixel.com/2016/01/19/fujifilm-is-testing-medium-format-sensors/#more-199392

[marcin wuu]

I'm still getting a serious case of cognitive dissonance whenever someone calls 35mm frame "big".

[David Brown]

"... whether people can really tell the difference between APS-C and Micro 4/3 in prints ..."

I recently finished a project where I shot most of it with a FF sensor, albeit an older one Sony a850. I also used a newer Lumix M4/3 for some shots, including a few that wee jpegs, and not RAW to begin with. Printed big - in some cases - and published a book. I can't tell which camera shot which picture without looking at the source file, and it hasn't come up in conversation with anyone else, either.

I don't think I'll ever have another interchangeable lens digital camera. (Film is a whole other discussion) These Sonys and Lumix are getting just too good.

[Stephen Gillette]

Nice job. Bravo! A great clarification of not only sensor size realities, but also how this relates to real-world printing. (As in: prints. Remember those?)

Anybody considering getting in line to acquire FF gear should read this first.

[dd-b]

The drive to smaller formats has certainly been a success! In the film era things had stalled out at 35mm (with 126, 110, and disc cameras eventually being swept away by the auto-focus auto-loading 35mm cameras in the 1980s—the only example I've ever heard of where quality beat out both price and convenience in a mass-market product). But digital has gotten that train moving again.

And the 35mm still frame was originally created by combining two frames that were quite close to APS-C size (the motion picture frame of the era).

[Robert Roaldi]

Good questions, especially for amateurs with limited budgets who buy 2nd-hand exclusively. I bet an old R1 would do just fine for many uses. But so might an Oly E-620, which you can buy for a song these days.

It takes some digging on the web to compare older models to determine if they'll do. And you have to know what to look for. I tell friends and acquaintances who are looking for photo gear to buy second-hand equipment until they figure out for themselves what they really need. But the urge to buy new toys is very powerful.

For casual sports shooters whose only delivery is to the web, why bother with anything heavier than a Nikon V2 (and a plastic cover for when it rains).

[Ranjit Grover]

Mike, In the pre digital era one could create "every thing else being equal" situation while comparing films. All films suited all cameras!, well almost. But that is not easy, if not impossible in digital era. So considering the size of the sensors in isolation is not the right way to look at the situation. The sensor size, the electronics, the pixel density, the quality and structure of the raw file, the lenses, the built in noise reduction, the AA filter or the lack of it, pixel binnning, everything counts. In effect comparing the sensor size as a surrogate for comparison of cameras is like comparing oranges to apples.

[Eric Brody]

Agree with this excellent post. The issue for many of us is where is the intersection of image quality and camera/lens size. I own a Fuji X T-1 with many lenses and recently rented a Sony A7RII and 50mm f/1.8 Sonnar, "to get it out of my system," since I had been reading so many laudatory articles about the Sony. My careful look at the files AND THE PRINTS, in my usual size of 9x13" told me that the Fuji, with its small body size, and with the excellent distribution of lenses, was just fine for me. Yes, I could tell some differences, physics is physics, but the overall system works for me. Others are perfectly entitled to their $50,000 Phase 100MP systems. A corollary to this is the video by Michael Reichmann and Kevin Raber, https://luminous-landscape.com/is-good-good-enough/ ... and it is.

[MarkB]

Those two Sonys illustrating the article look like they have great Zeiss lenses. Will we be reading about their 1" sensors in 13 years?

[Kev Ford]

I would be very interested to see what the majority of enthusiasts actually ended up using if price wasn't a factor. I think the outcome would be surprising.

[Matt]

Smaller formats aren’t really smaller or lighter if you take into account the lenses they come with. A fuji 50-140mm f2.8 is just as big and heavy as a canon 70-200mm f2.8 or nikon 80-200mm f2.8.

[Hi Matt, That's not true--I looked 'em up, and the Fuji is .87 inch shorter, .24 inch smaller in diameter, and 1.09 lb. lighter than the Canon.

With teles and tele zooms, sensor size doesn't come into play as much because circle of coverage isn't a serious constraint with longer lenses. It is with shorter ones, where the smaller sensors have a big advantage, for instance:

http://camerasize.com/compact/#469.484,520.422,ha,t

--Mike]

Take into account the higher noise and deeper focus of smaller format, and you end up with bigger, faster and more expensive lenses. Take for instance said fuji 50-140mm f2.8. The nikon or canon 70-200mm f4 generates similar depth and is actually much lighter. Or nikon / canon 50mm f1.8 vs fuji 35mm f1.4, or fuji 14mm vs nikon or canon 20mm…

Another thing that’s really bothersome with apsc vs FF is the small viewfinder. Small viewfinder = poor control = poor pictures.

[Also not necessarily the case. I just compared the Fuji X-T1 to the Sony A900, which has a nicer, brighter viewfinder than the Nikon D750 I just rented, and the Fuji's viewfinder is both larger and has better eye relief than the Sony's. Granted, there are situations where I would prefer the A900's OVF to the X-T1's EVF, but just as often it's the other way around. --Mike]

[Rob]

Fuji made similar claims about their 16MP X-Trans sensor outperforming higher MP Bayer sensors. Those claims turned out to be false, as I expect will be the case for such claims about their 24MP sensor, especially when you consider the inherent IQ advantages of FF over APS-C.

[Roger Bradbury]

Mike, it isn't just a rumour about a full frame Pentax. It's due out in the spring. There's some photos by a Pentax fan of the camera here, I think in a glass cabinet at a photo show.

[jim]

As far as i know and understand, its not only about sensor iq that determines iq. Even if a 1'' small sensor has the color depth, MP, s/n ratio etc of a FF sensor, the following "problems" will arise:

1. Diffraction will hit at as small apertures as f4 or f2,8 or smaller ( i dont know exactly)

2. This means that if you need to utilize what dof/speed combination can give you, you need lenses below f1.

3. Although you can get sensor iq with smaller pixels, you cant shift the aperture range downwords to get equally useful range:you cannot get an f0.7 lens at f0.7 for 1'' sensor to work as well as a f2 lens at f2 for a big sensor because optics dont work like that (may be unless you build a huge and expensive lens f0.7 for 1'' sensor)

4. But even then, you have problems on the shutter speed side of things: to have a sharp photo (not of the runner but of the surroundings) of a runner blurred because of shutter speed of 1/20 secs, you may need an aperture of f16 or much more, which will be awful on the 12MP 1''sensor because it will offer a way blurred iq due to extreme diffraction.

5. Im not sure if in such small sizes of sensors, to get lenses assembled with precision, you need much more precision technology, so as not to get decentered or misaligned or offset lenses.

For me, the combination of aperture range, shutter speed, MP, size so as to help us be creative ( i hate bokeh) at an artistic level, is best delivered with m43 to APSC sizes, unless technology changes everything (digital dof, or i dont know what other it could change)

For me FF is too big since f1,4 or f2 do not add much as a creative asset, since sensitivity of smaller sensors is good enough, and will be even better.

[Burdette Parks]

Re: Relating sensor size to something we can connect with. In the classes I teach I say that full frame is the equivalent of a 35 mm frame. Then I reinforce that by holding up an SD card and saying that it is only slightly smaller than a 35 mm frame (FF = 36x24 mm vs SD = 31x23 mm) and therefore gives a pretty good physical demonstration of the actual size of a full frame sensor.

[Michael Perini]

A very good summary especially the reminder about linear size vs Area. Bigger tends to be better but there is a huge amount of overlap. And to Ctein's point about all things are never equal, it especially shows up in subject matter and exposure and brightnes range. Sensors have sweet spots that when hit can make the sensor sing. I was recently asked if I had a picture of a friend who had passed away. I did have one picture, from a long time ago Sony Macica. It was a 1mp jpeg and it produced a really beautiful 10x10 print. I was amazed. Everything came together by luck in that picture.
---Not at all how I remember the AVERAGE picture quality of that camera.

[Bill Mitchell]

I love my Sony RX100. It has replaced my Minox, which is far smaller and lighter.("sensor" size 8x11mm, almost the same). I carried one in my pockets for 50 years. Too bad that film and processing is virtually unavailable.

[psu]

I stopped minding be a bit behind two years ago.

m4/3 is good enough even at ISO 3200 for me to simply not think about it that much.

Even the phones are not bad (almost as good as APS-C from 8-10 years ago), as I say in every comment I make on pieces like this.

[Beuler]

Putting together Pentax, the topic on system selection and the current discussion, I would argue that user interface is as much part of the system as lenses or cameras. In that sense Pentax is the only system that gives you 645 and Aps-c with the a same interface. It is most distracting having to recombobulate your finger memory in order to change sensor size.
Regarding image quality, remember that in order to get 645 quality from FF sensors, lenses must be huge and outrageously expensive, as in Otus and "Art". But don't quote me on that.

[GKFroehlich]

I've always found this article--indeed, the entire site--useful and informative <http://www.cambridgeincolour.com/tutorials/digital-camera-sensor-size.htm>. It's just a bit dated with regard to current camera models, but the information is solid, and well-conveyed, in my opinion.

[aaronL]

I'm not so sure about the gains being made over the last few years on sensors.
I've been looking at M4/3s, waiting for a 'JND' but I'm not seeing it.
I have a 5 year old panasonic m4/3s which uses the same 16 mgpxl sensor as almost every M4/3 camera since. The recent Panny GX8 has the newer 20 Mgpxl sensor which seems to have very little advantage over the previous generations..
So where is this improvement?

AaronL

[David Kieltyka]

My only beef with the sensor size illustration is that it stops with "full frame." Where's 44x33mm or Leica's 45x30mm? Or say, for historical reference, 4x5" sheet film? Think there might be a *marketing intent* at work here?

-Dave-

[Thomas Rink]

If you print with a desktop printer (say, an 3880 as an upper limit), then the sensor size and legacy nowadays is largely irrelevant - an "ancient" 12 MP camera is sufficient (according to my experience). The more interesting question is which sensor format or mount type gives me the system I want, at a reasonable price point? If one wants to shoot primes (e.g. a 35/50/85 mm kit), in my opinion, one can rule out any APS system on the market. Canon, Nikon and lately Sony neglect those systems lens-wise, and Fuji is rather expensive. This leaves M43 and 35 mm full frame as options. As of 35 mm FF, one can currently choose from a lot of low-mileage second hand bodies - starting from an A7 or a D700 for about 700 Euro, a D800 is about 1600 Euro. In the EU, camera shops even have to offer one year warranty for those bodies. Lenses? From Ebay. Many manual focus lenses from the 80s work well on these bodies (provided they're stopped down a bit). IMHO, the Sony FE bodies are nice options for this - their short register distance means that a lot of legacy lenses can be adapted. In addition, they usually suffer huge value depreciation. If you want AF, a second-hand FX Nikon with a kit of the new Nikkor f/1,8 primes might even be cheaper than a comparable kit based e.g. on the Fuji X system. I could imagine that an M43 system is even more of a bargain, since the lenses seem to be quite affordable.

[Chris Stump]

May be a bit OT from this particular post, but as a relatively new and appreciative reader [since, apparently the 70's in (your) various guises], here's a post from today on the Leica forum. I'd love reader feedback:

Hello all,

I’m new to the forum but not to photography or Leica. This seems to be a wonderfully supportive community, and I’d love to bounce a few thoughts off of you all.

Title: Looking for my ‘Mini-M’

I think it was in a David Farkas review where I first heard this term, and it fits me perfectly...and dare I suspect, a few of you?

I wish to find a digital replacement for my beloved M3, but with some modern conveniences made necessary by changing habits and aging eyes.

1) The primary requirement is that the camera have a Leica [or Zeiss T*] lens.

2) Second is a viewfinder. I prefer [surprisingly enough to me] an EVF at this point in my eyes’ age [I’m 57]. A ground glass or hybrid would be cool. I think. Not sure to be honest about the hybrid.

3) Third is a wide to portrait zoom, and of some speed. I just don’t care to swap around between lenses anymore. In one short outing I’ve used both the short and long ends of my new D-Lux Typ 109 to good effect, and sadly, as I see what Jacob Aue Sobol and David Alan Harvey et al accomplish, cannot imagine shooting with 2 prime lenses all day, or swapping any number of them out in a squall [or family picnic].

4) You’ll notice I’ve left physical dimensions and price out of the equation. One cannot have everything, can one?

5) However, to be fair, portability does enter into it. If physical size [and cost] were truly of no consideration I’d have an S Type 007 and be done with it, right [David]?

6) But that leaves cost, and I have to let that one go. My D-Lux Typ 109 was $1,000, and the Q Typ 116 that I SO want to love [but can't] is the better part of $5k. So, $10k for my dream camera?….I’d have to call that a bargain. More? Hmmmm……Let’s talk.

What I appear to be looking for is a combination of the DL 109 [EVT, fast zoom, compact], Vario-X [better lens (but fuck slow, and only 28mm), can add EVT], in the form of the Q 116 [Full Frame sensor, classic form factor, what looks like a prime lens stuck on the front. Gorgeous], with a decent sensor [4/3 min], [min] 24mm on the wide end and as long as possible on the other…85mm? [105?!?]. Oh, and f2~2.8’ish at most points. f4 if you give me the 105mm. Too much to ask?

Have I just described some Lumix monster I’m not aware of?...that wants to be press fit into an M body? [perish the thought]

So, what has the group got to offer…I'm truly ready to buy. Or are we just waiting for the engineers in Wetzlar to finally give us what we want?

[Moose]

Some months ago, all the enthusiastic praise of the Fuji X cameras and their wonderful IQ started to nag at me, with my puny µ4/3 bodies.

So I grabbed some Raw files of the standardized studio shots from DPR and IR reviews, converted them and stacked them on top of each other in PS.

My conclusion, after way too much pixel peeping, trying various deconvolution settings, and so on, was that, at base ISO, it was an absolute toss-up between the X-T1 and the Oly E-M1, the latest of each line at the time.

There were parts of each test image where one or the other had a tiny edge, to my eye. I'd give the slightest of edges to the Oly, but that may be my priorities and prejudices and most certainly wouldn't be visible in any but gigantic prints, if then.

I didn't check higher ISOs; that gets amazingly tricky and starts to depend a lot on the Raw converter, NR and sharpening/deconvolution tools, etc.

Nor did I check DR, although both easily handled the step charts.

Then again, for some uses the HR mode of the E-M5 II transcends its sensor size in both resolution and color accuracy. In my limited testing so far, it also handles highlights better than standard mode, so may have better DR?

Just another data point ...

[Kefyn Moss]

At one stage I owned both the Oly E-M5 and the Fuji X-E1, both 16MP although different aspect ratios. I did a few comparisons with the same subjects, same lighting, same settings (as close as possible) and found it hard to notice much difference. If I looked hard the APS sensor was slightly "better" to my eye but I had to look hard, which is not how I would normally look at a photograph anyway. I have printed A2 size from both of these cameras with good results, as long as the file is good enough to start with, but preferred the haptics of the Fuji and the "look" of the files too, so went with that. Both out performed older files from my D700 in comparison.

As far as the parameters that count to me, I think that Fuji has made all other things even less equal for the same or similar sized sensors which has become more of a differentiator than sensor size or pixel count. I am however, looking forward to a jump from 16 to 24MP with the new Fuji sensors. Definitely more than a JND.

[John McMillin]

Like Mike, I shot full frame with a Sony DSLR, back when there was such a thing. Pentax's new K-1 (what else could they possibly call it?) will go straight to the top of my wish list when it's introduced next month. Not because I need higher sharpness or more pixels than my K5IIs, but because I want another full-sized optical view finder. As long as I remain a diehard fan of the OVF, sensor size is related to mirror size and the resulting view I get. Better images are just icing on the cake.

[Richard Alan Fox]

My printer is an Epson 3880, a 17 x 22 inch machine. 4/3 makes a perfect 15 x 20 image size with an inch around border.
I wish my Nikon D800E included a 4/3 crop.

[Darlene]

I currently shoot Foveon APS-C, Nikon FF, Hasselblad 44x33 CMOS and Phase One 37x49 CCD. Each format produce a unique look. I do not know if "all other things being equal" can really exist without stripping away these unique characteristic features or qualities.

I enjoy shooting the APS-C system the most because it is simply easier for me to handle. I have been working with glass all of my life and like so many other readers here, I have seen a remarkable amount of technology changes in the past few years, but I can see the differences between some formats and some lenses. If those differences went away, we could loose the unique creativity it has also generated and I do not know if in the end that would be a good thing.

[Dan]

Depth of field control and smooth tonality far outweigh "sharpness"...so all I want is the largest possible sensor in the smallest possible body.

The digital replacement formy dear departed Contax T3 has yet to arrive.

Migration to traditional sensor sizes is the natural trend of the market. We are finally seeing true medium format sensor sizes and that may just save the mf market. Bigger is not always "better" but its always better selling.

[Kevin Purcell]

This is a geeky post but to understand why things are sometimes needs geekiness.

@Carstern: "But as manufacturing sensors gets cheaper all the time, there might be no real reason to use one that small."

This isn't true for image sensors.

Image sensors don't gain much from Moore's law. They gain some by going to smaller processes (lower power, better QE) but 65nm is state of the art for commercial CIS (at Sony and Samsung) but pretty old for processors.

The base cost for an image sensor is proportional to how much time it uses on the fab to make (and how old the fab is -- has the capital cost been paid off). That's proportional to the area of the sensor and the yield (which is also related to area) though yields for even large sensors are now very good. The consumable cost (the wafer) is proportional to sensor area too.

Big sensors will always cost more than smaller sensors (in proportion to the area of the sensor plus a bit more for lower yields).

Fab costs are almost entirely capital costs (a fab will cost in the order of a billion dollars or more to build) with some operating and consumable costs. So sensors get cheaper the older the fab. This (and being "good enough") I think explains Canon staying on the 180nm process for a long time.

Sensor size drives other trade offs in camera design: a larger sensor means a larger lens (lens dimensions are proportional to sensor size) and the weight of the lens goes up with something between the square and cube of the sensor size. If you want to design smaller cameras then the bias will be to "smaller" sensors especially if that camera needs a zoom lens.

Smaller sensors will be around for a while. I think cameras with zoom lenses will settle on type 1" as the best trade off between camera size and sensor size (we already see this with the RX100 and similar cameras). For prime lens cameras APS-C and FourThirds both are in the sweet spot. I think we'll see 35mm and larger sensors as "specialist" cameras as time moves on.

@Dennis: "but how much "better dynamic range" does one sensor give "

Dynamic range is a pixel property not a sensor size property. You can put different pixel designs on the same sized sensor and get different dynamic ranges for the same sized sensors.

In general, the larger a pixel is the larger the floating diffusion capacitance to store the charge transferred from the photodiode (usually the limit on the full well capacity of the pixel). The larger the FD capacitance the larger the full well capacity the pixel can measure and the higher the dynamic range for a fixed read noise.

But (here's another trade off) the size of the FD capacitance contributes a component to the read noise (so called kTC noise). That limits your low ISO performance and with low noise pixels and low ADC noise you can see the effect of the larger FD capacitance.

The Aptina type 1" sensors in the Nikon 1 cameras had a feature that could switch the size of the FD capacitance between two values for the whole sensor. At low ISO (up to ISO 400) they would use a larger FD capacitance to optimize for larger full well capacity in bright light. At high ISO they'ed optimize for lower read noise with fewer photoelectrons. This gave excellent low read noise at high ISO when you need it. The side effect is the dynamic range across the ISO range was improved overall especially at the high end (because of lower read noise). You might see this idea reappear in future sensors with larger pixels.

The recurring fact here is image sensor design, along with camera system design and pretty much the whole of photography is all about trade offs.

[MarkR]

My own brief foray into FF taught me that it is a lot harder to get a lens that is edge to edge sharp. Which means mushy corners, which means the sensor's theoretical advantages often stayed theoretical.

[Dave in NM]

Let's not forget that image quality, maximum print size, and noise characteristics aren't the only factors that might lead someone to choose full-frame over the other available sizes.

In my case, after shooting 35mm for nearly thirty-five years, I held off getting a DSLR until I could afford a full frame Kodak SLR/n in 2006. My primary reason, equally important to image quality and pixel count, was that I was already familiar with the focal length/angle-of-view relationship of lenses made for the 135 film format. I wanted to use the lenses I had, at the angle of view I was familiar with, and frankly, I wasn't interested in retraining my brain with conversion factors.

Having said that, if lenses were originally labeled by angle of view for the format that they were meant to be used with, rather than by focal length, things might have turned out very differently for me...

[Earl Dunbar]

Do the camera makers actually conspire, by producing so many sizes, to actually confuse the non-cogniscenti shooting public, thus driving sales and churn? No, it couldn't be so...

But to be totally off-topic, I am glad you brought up Cutty Sark. As a lover of Scottish whisky but with somewhat modest budget, I often select a blend rather than a single malt -- at least for my "everyday" dram.

One day at the libation store it dawned on me that I had never had Cutty. So I bought a bottle.

Cutty Sark may have been (be?) a really nicely balanced ship, but the whisky? Aargh, matey, not so good.

[Jeff Smith]

For the last couple of years i have settled on the micro 4/3 camera system ad it represents a good balance between image quality and camera and lens size for me. I have had recent APS-C cameras (Ricoh and Fuji) but sold them both becuase while the image quality was pretty good i really could not see much difference in image quality from micro 4/3. I also have a Sony RX100 and continue to be amazed at the image quality that camera can deliver, particularly in good light. Looking at photos in good light from the RX100 look pretty darn good and not really different from micro 4/3 to me. Where i have noticd a difference with the RX100 is in situations where the lighting is less than ideal.

So if the lighting is good i personally find little difference between APS-C, 4/3 and 1" sensors.

I think these observations underscore the broader point that the image quality of many cameras today us pretty darn good so jst shoot with a sensor size/camera system that you are happy to carry around.

[Neely Fallon]

Want to see an actual difference in sensor size? 645. Medium format. The ratio of FF to APS-C is negible compared to 645. So in that respect
Fuji"film" is correct. Hopefully Fuji"film" will produce a digital 645 offering. Moore's law suggests that sooner rather than later another digital 645
could be quite viable.

[Michael Bearman]

Mike the ed. - She's a ship.

Bill Mitchell - try http://www.bluemooncamera.com for minox film in cassettes and also processing. I used to love using both a minox and also a Minolta 16 mm (don't remember what it was called). I still have a developing spool for a 35 mm tank somewhere, I think. I also still have a couple of Olympus Pen F half frame SLRs and some lens or two, which were a favourite of mine in the 70s. So I should be more than happy with the RX100. But just can't like it. I will try the new Mk IV.

[Ross]

I was working on a digital sound lay for a feature film and had need to work with the 35mm work print for one step. Fresh out of college assistant had a sudden revelation of what those "frames" counted in SMPTE code really were.

[Gary]

I'm wondering (really; this is not a rhetorical question)why some people seem to be so interested in the coming Pentax FF camera. Is that ground not well covered by Nikon and Canon? What advantage do these folks expect Pentax to bring to the table?

[Doug Thacker]

For serious practitioners of 35mm film photography, the 36 x 24 legacy is hard to shake. You spend years internalizing the correlation between focal length and angle of view, between f-number and light gathering and depth of field, so that you know it by heart and don't have to think about it much; you devote much time, perhaps, months or years, to working with a single focal length: and then your Nikon D70 arrives and everything changes. Suddenly nothing is plumb, level, or square - what you know, or thought you knew, is at variance with reality.

This digital thing is great, you think, because one more thing that no longer holds true is the old Kodak slogan, You press the button and we take the money. I can press this button as many times as I want, haha! - and for a while that fact trumps everything else - a little revenge against Rochester. But this soon leads to shooting around the shot instead of making the shot; and the world is still off kilter.

Oh, if only my sensor were the same size as film, you think. It would be the best of both worlds. I could forget about the substrate and get back to shooting again. And then salvation arrives with announcement of the 5D/D700. Thank God.

But wait, does this camera seem a little large and heavy to you? And what about those lenses? They're huge. Didn't they used to be smaller? What happened to my F3HP and 50mm AIS lens? Why does it feel, kind of, like I'm walking around with a medium format rig? I never wanted to do that. What? Nikon just announced some new zooms? Can't somebody just wedge a goddamn 36 x 24 sensor into a Nikon F3 body? What's the problem? Laws of physics, that's what - notoriously intransigent.

Being part of the physical world, humans can be intransigent, too: form factor, you remember belatedly, is just as important to your photography as angle of view, depth of field, etc. Maybe more important, in fact. Form factor and interface. And that's when Fujifilm steps in with the X-series cameras, because someone there - a photographer, obviously - remembered, too.

APS-C for 35mm film photographers means having to re-wire some circuits, to un-learn much of what you'd internalized, and then to learn it anew. For what is nominally called 35mm - the erstwhile small format - APS-C is the new normal.

Once we legacy laggards die off, the field is open, or may be. "Normal" could include anything and everything, including possibly a chip implanted in your brain, or stills selected by algorithm from a video file. Google only knows. But for now, for many of us, and for some time to come, APS-C is digital film.

[Hugh]

Nice theory. Didn't work that way for me.
Baseline is a 5D3 with L primes and A2 prints (17" x 23"). Sometimes print 24" x 36".

Tried Sigma DP2M and DP3M. Brilliant prints when everything was perfect. As good as the "full frame". Terrible software, limited to good light.

Tried an OMD EM5 with 17mm/1.8 and 45/1.8 lenses. Lovely camera and lenses. Also tried a Canon 100D ( US version = SL1 ) Lovely camera.

Both more modern sensors than the 5d3. Should have been comparable results, but for me they were just a bit disappointing. Sold them all.

I'm sure it would be possible for a great printer to make prints as good as those I get from the 5d3 with some effort, but I would rather carry an extra kilo of camera and lens than spend an extra hour in front of the computer working on a print. Printing isn't my hobby.

Pity, because I'd have loved the tiny, light Olympus kit for travel.

[Matt]

With teles and tele zooms, sensor size doesn't come into play as much because circle of coverage isn't a serious constraint with longer lenses. It is with shorter ones, where the smaller sensors have a big advantage, for instance:

http://camerasize.com/compact/#469.484,520.422,ha,t

Hi Mike,

In order to achieve the same noise and depth, you should be comparing the fuji 23/1.4 to canon 35/2.0, not to the much bigger 35L. This principle is fundamental, else you're comparing apples to oranges.

And that's where smaller format turns out bigger, heavier or more expensive than its full frame equivalent.

300g (fuji) vs 200g (canon non-IS)
€800 (fuji) vs €300 (canon non-IS)

Same holds true for 14mm vs 20mm, 35mm f1.4 vs 50mm f1.8, 50-140mm f2.8 vs 70-200mm f4... You get the picture.

Point is: what you win in terms of bulk / price of smaller format bodies, you have to make up with faster (and thus bigger, heavier or more expensive) lenses.

[Michael Farley]

Mike, your article is notable for mentioning a former member of my camera club, Kenneth Mees, who was behind many of the developments in film technology during the first half of the 20th century. He started his career at Wratten and Wainwright, based in Croydon, England, but refused to leave when approached by Kodak as he wanted to stay loyal to the company. Instead Kodak bought the firm, which is how they also acquired the Wratten name.

You can find out more about Mees and see some of his photography in the Club Archives section at Croydon Camera Club's website - http://www.croydoncameraclub.org.uk

[Actually that misstates the relationship a bit Michael--Dr. Mees agreed to come work for Kodak when invited to do so and he was the one who asked George Eastman to buy Wrattan & Wainright, which Mr. Eastman did. The situation was cordial and cooperative.

And thanks for that nice link! I'd never seen "the Mees book" before. --Mike]

[s.wolters]

Teaching some friends the basics of photography I also used this Gizmag illustration to explain the different sensor sizes. I don’t think anybody understood it. The millimeter sizes are clear and precise but hard to get into your brain. The inch designations are not even the real measures and full frame, APS-C and Micro Four Thirds are abracadabra to most mortals as well. And why should they bother anyway, especially since I told them we that the differences are hardly visible.

Of course, once you are getting more serious you need to know the relation of your sensor size to your focal length. Why not name the sensors after the diameter of the image circle? In millimeters of course, like the focal lengths. Could be something like S50 for full frame, S35 for APS-C, S25 for Micro Four Thirds and S18 for 1-inch.

Looking back I wish the sensor industry would have taken off with some kind of DIN-standard where all the paper sizes have the same height-width ratio. 1:√2, approximately 1:1,4. When you cut an A4 in half you will get two A5’s with that same 1:√2 height-width rate. Cutting an A5 in half will give you two A6’s and so on. This practical shape is slightly less wide than APS-C and Full Frame, and bit wider than Micro Four Thirds. Very usable, not only for paper but also for all kinds of displays, horizontal as well as vertical.

In the world of the screen sizes of computers, tv’s, phones and pads it’s even a bigger mess. Nowadays we are all forced to use very wide horizontal screens. Some might get the feeling that this makes the world larger, but I often get the feeling of peeping through a letterbox. Fortunately there is a strong opposition coming up. All those smartphone witness videos we see in the news programs are vertical!

[Christer Almqvist]

I worry less about sensor size than about camera body size.

The EVF M4/3 mirrorless bodies are much too big for the sensor size, but they are fine to hold in my hand. So I upgraded my first Sony A7 to the new huge MP version. The files can be cropped to a substantial degree without quality loss so I don't need to carry a tele lens.

And if I downsize FF pictures shot at high ISO, then the quality I get is much better than what I would get with a small sensor, final picture size being equal for the two.

[John Lehet]

The thing is, as a photographer you might want to have a big palette, a large range of possible effects and flavors, and the effects of sensors goes beyond simple resolution or other qualities. The sensors are paired with lenses and so yield different "flavors."

I carry both a FF bag and a MFT bag. (Yes, in fact I am going to physical therapy for back problems). It's not a matter of quality as an ultimate characteristic of either system -- it's the quality of the particular thing I'm trying to do.

The FF excels at two things: If I want a delicious degree of control over depth of field, or if the subject is far away and I want ultimate detail in the (too rare) case that there is enough light or enough distance to get everything in focus. The depth of field control on FF with a decent prime is really exquisite. Though I am a huge MFT booster and user, it really can't compare in this regard unless the subject is close and the lens is long (which for me does happen a reasonable amount with MFT).

I find with the FF and a prime that I have a very broad range of bokeh or blur effects that are likely to be pleasing, if that's what I'm after, and if there is enough light to use a range of stops. The MFT on the other hand has a little range at the very wide open side of things, and then the blur starts to get less pleasing generally. This is even true with the Olympus 60 macro, which is known for it's beautiful bokeh and which is a long lens on MFT.

On the other hand in dim light having the DOF control at the wide open side of things tips the scales to the smaller sensor, and in the case of wanting a lot of DOF the small sensor wins this case too.

Besides the edge cases where the FF shines, I use the Olympus EM5ii and a bag of lenses for everything else. Fairly often I do want a lot of depth of field, and it is much more reasonable to get it with the MFT. You can only stop down so far before diffraction starts degrade quality. You can only stop down so far period. It's also true that sometimes I want some but not razor thin depth of field in limited light. For example in the following shot, I tried to do it with the FF camera, but the MFT won the day. In this case the the 45mm lens on the MFT gave good depth of field and pleasing blur. The light was dim. I tried to do this same shot with the FF but much less DOF did not yield a keeper.

http://www.lehet.com/photo/details/two_frogs_foliage_reflections_a103888.html

The smaller sensor can deliver enough quality for an excellent 16 x 20 print and the system gives a lot of versatility. It's often the camera I grab first if I want full versatility. For me it's not always so much a question of the resolving power of the sensor but of its characteristics in relationship to lenses in front of it.

The FF does offer more "ultimate" quality, but only for certain things. If things fall into place so that the resolution of the sensor can be an advantage, I will go for that. In bright light with a distant subject and the framing such that I could use the 85 prime on the FF or the 45 prime on the MFT, I'll use the FF and get a better photo.

[Alan Carmody]

For those who have difficulty with metric millimeter measurements, it is easier to visualize full frame in inches as 1" x 1.5".

About the size of two SD cards, side by side.

[Dr Tom Bell]

Whilst I love my Pentax K5 and all the lenses I have collected over forty years ...and for no logical reason enjoy my X100 ... the sweet spot for me AS A SYSTEM camera is M43

Because if you pack an

OM10 11
Olympus 12/2
Olympus 17/1.7
Olympus 25/1.8
Olympus 45/1.8
Olympus 60/2 Macro

Then you really really notice the difference in size of your bag...

I have the Oly 12-40 attached and then I glue two lens rear caps together and then two lenses go in each slot in my bag ...

It is when you see the cameras with their lenses together that the size advantage of the M43 system magnifies ...I fit all these plus the Pan 100-300 in a Lowepro Slinshot 100 ...incredible

[Speed]

Mike wrote, " Those of us who developed tens of thousands of frames of 35mm (yo!) have a pretty good grasp ... "

As do those of us who have loaded tens of thousands of frames into slide carousels.

[Sarge]

Mike,

The size differences in terms of area and sensor diagonal (linear) as between: 1" vs. Four Thirds, FT vs. APS-C, and APS-C vs. Full Frame; are all well above Ctein's JND threshold estimates as illustrated in Figure 1, below.

Figure 1: SENSOR AREA/DIAGONAL SIZE DIFFERENCE

Notes, Source: The sensor areas and sensor diagonals are all drawn to scale. The blue outline is that of a digital MF sensor (44 X 33), included for reference only. Please see this Google spreadsheet for my computations of percentage differences.

The least difference is that between the FT and APS-C formats, as follows: 48-64% areawise and 27-31% linearly.

I arrived at the lower percentage figure by using the average of the areas or sensor diagonal lengths between formats as the base (divisor). The higher figure is arrived at when using the area/diagonal of the smaller sensor as the base/divisor.

Given the rapid advances in sensor technology, which make differences in image quality between adjacent sensor sizes almost imperceptible, shouldn't the JND thresholds be raised? Say, from 12-15% to 25-30% linearly (sensor diagonal); and from 25-30% to 45-60% in terms of area. This means a doubling of the current thresholds.

As you suggested, the difference in IQ between late generation FT and APS-C cameras, ceteris paribus, is almost imperceptible. The higher thresholds address this. As Ctein noted, Figure 1 also shows that the sensor diagonal comparison "scales better" than the sensor area comparison. (I'm assuming that sensor diagonal is a valid proxy for resolution.)

[Steve Jacob]

The current generation of sensors have reached a level that's hard to improve on using Silicon and a colour filter array. Maybe a 1/3stop improvement here or there, but not the 1+ stop improvement we used to see every 4 years up to around 2009.

But at some point we all reach a 'level of personal sufficiency' or LOPS. Philosophically, I think that knowing your LOPS is the key to contentment in all things.

When I bought a D90 (Nikon) in 2009, the performance of the D700 I also bought was clearly superior. My LOPS was very much in the FF camp for all but daylight photography and casual shooting.

When I bought the Xpro1, the performance was comparable with the D700. Stupidly, I later bought a D800, thinking it would be better again. It was, but as I found out it now exceeded my LOPS and was largely a paperweight. Too much hassle for too little gain.

I also found out where my LOPS was.

The 16MP XE2 allows me to make 24X16" prints at ISO400 which, at 2ft viewing distance, have an angular resolution of 90 pixels per degree. At ISO 400 the noise level is still extremely low.

By way of perspective, a magazine sized photo at 300ppi is about 70. The human viewing limit is about 120-150, but that assumes there is sufficient contrast (ie black and white lines - we are generally much less sensitive to colour detail unless it contains a large green component).

The same print has a linear resolution of 200ppi or thereabouts. This means it doesn't fall apart when you peer closely at it either. 12MP is marginal.

In real images the difference between D800 and XE2 images is moot. The acutance of easily visible detail is generally more important than the resolution of very small detail. Your threshold depends rather on what you shoot, but with 16MP+ good PP skills are the decider, not the camera.

Bearing my LOPS in mind, there are 2 sweet spots currently in operation.

24MP is a good compromise with some crop space. I crop a large number of my images to a moderate degree. The storage requirements and processing speeds are reasonable (given current tech).

APSC is a good compromise for SNR related quality up to about ISO 3200, while remaining relatively small, portable and cost effective (I would dare to say it's not far off 6X7 MF in many ways, though noise vs. detail comparisons are fraught).

So, the Xpro2 is very attractive to me.

Of course, it may not be to anyone else. Sports shooters want better high ISO and fast cameras with good predictive AF, but there is always the Nikon D500. Also APSC. Compared to the D5, it's a good compromise.

If you are a serious landscape shooter, then full frame and 42MP+ is a probably a good idea. You can push out 36X24" prints with high levels of critical detail all day, but it also makes lens aberrations and sloppy technique a bit more obvious when you start to look closer...

So, once you find your LOPS, you just need to find a camera you LIKE USING. For me, that's Fuji. YMMV of course - and no I don't think that MFT is significantly worse in any way, or that the A6000 is an inferior camera.

I just prefer 3:2 aspect ratios and I like using Fuji cameras. That point is not objective in any way. Just a preference.

[Phil Service]

Mike,

This comment is late to the party, and you're probably tired of this thread by now, anyway. But,... The principal reason why many comments note that it is very difficult to see any differences between images made with sensors of very different sizes is that, for the most part, we are dealing with approximately "equivalent cameras". In particular, photosite size is generally larger for larger sensors and total resolution stays the same -- think 24 MP APS-C and FF and 20 MP 1". If we use equivalent apertures and equivalent focal lengths, nothing really changes. Diffraction blur and photosite spacing as fractions of image field dimensions remain constant across formats, and the NUMBER OF SAMPLES of the image field remains the same. Larger sensors will yield greater image resolution only if we put aside camera equivalence. In particular, if we produce large sensors with small photosites. I'm not sure we need to go as far as putting 2.4 micron photosites (Sony RX100) on a FF sensor -- to get 150 MP. But if we did, we'd get about 2.7 times more samples (linearly) of the image field -- comparing FF to 1". Of course, to realize the full resolution advantage of the larger FF sensor, it would be necessary to use a larger aperture to control diffraction blur, thus sacrificing DoF equivalence. Obviously, I'm assuming that lenses are not a limiting factor, and that ISO, noise, etc. are not issues. The film analog is that emulsion grain size is invariant across formats -- at least so I assume. Therefore, for equivalent focal length, larger format film produces sharper, higher resolution images.

Phil

[Dave]

I moved from a Nex 6 to a A7Rii and still have a number of apsc lenses.

Compared to the Nex 6 my A7Rii is a superior apsc camera in every way except that it is a little bigger and heavier.

I think is in part due the the in-camera IBIS.

As the price of full frame sensors drop the cost of buying one and having the choice of using a mixture of crop and full frame lenses depending on the needs of the situation will be very appealing.

I think this is the way things will end up going.

[ctein]

Dear Sarge,

It is important to understand that a JND is a JUST NOTICEABLE Difference. It is not a large difference. It is the smallest difference you are likely to be able to see.

Anything less than a JND is basically ignorable. But a single JND's worth of improvement is not, usually, of particular importance.

pax / Ctein

[Right, and some people can't notice certain JNDs. Depends what you're good at. --Mike]

[Sarge]

@Ctein, Mike:

Got it. Thanks!

It's just as well that I'm stuck with my aging ILC. My ND threshold is so high I can't tell the difference across formats viewing images on my PC screen. And I'm a Philistine when it comes to print. Just maybe, if I channel my GAS to TOP's print sales or photo books, I can hone my threshold from ND to J~ :)

[s.wolters]

After reading the comments on this more than interesting topic it strikes me that nobody, unless I missed it, mentions the quality of the color as a reason to choose one camera or another. Quality of light, color, contrast and tone are much more important to the results than a few pixels less or more.
I love the standard colors of my Olympus. If I had to describe them I would say that they are close to Kodak Portra. Not neutral, a bit oversaturated, with great skin tones and clean whites. Fuji film has an impressive range of film simulations, and I adore the Nikon 1 colors too. But they are all different and probably a better reason to go for one system or another than the JND of resolutions.

Another noticeable thing is that where the whole world is mainly using electronic monitors, most readers still use print as their main reference. Eight years ago I also was worried about how to get all my digital files printed. I don’t know when there was a tipping point, but at the moment I’m more worried about how I can make all my prints and transparencies digital.
Oh well, if T.O.P. is slowly turning into the Jurassic Park of photography I don’t mind. Can be a pretty exiting.

[Doug Thacker]

As a very late follow up to my comment above, where I idly speculate that the future may bring photography via implanted brain chip, I offer this, just published in The Daily Beast:

The Pentagon Wants to Put This in Your Brain

Quoting the article,

"The U.S. military is beginning work on a new 'implantable neural interface' that it hopes will allow wearers to transmit data back and forth from their brains to external digital devices [. . .] Imagine controlling your tank, car or microwave oven with your mind."

Imagine your mind being controlled by the Pentagon. Now that's what I call a "backdoor". I'll stick with regular cameras, thanks. :)

[Doug Thacker]

Dpreview has an excellent interview with two Fuji execs here. (These might very well be the two I spoke with in Tokyo in early-mid 2014 but I can't be certain.) They aren't asked about and don't mention medium format, but they do go into some detail about their overall philosophy and their choice of APS-C. Their openness and candor are impressive.

[Dean Johnston]

Hiya!

> That's very much in error.
>
> It's "ship," not "boat." —Mike the Ed. :-)

You almost had to buy me a new computer, cos I almost spat coffee all over mine.