Size Doesn't / Does Matter

By Ctein

The recent spate of camera announcements has brought forth many half-informed comments about sensor sizes and pixel counts. Let's clear some stuff up.

You can make great, as in fully professional quality, photographs with small sensors. I'm not talking about artistic subjectivity, I'm talking about objective technical quality. Now, just as with small-format film cameras (read: 35mm) they're not going to be the best choice under every single circumstance. Some subjects or working styles simply demand a camera with humongous real estate. No argument. My disagreement is with those who say it's not possible to do professional work with small sensors, stating it like some law of nature rather than their personal preference. Plainly and simply wrong.

My current camera, an Olympus Pen EP-1, has a half-scale sensor (scale relative to 35mm/full-frame—nobody can remember what those damn video-format names mean). My camera before that, based on 4-year-old technology, which is forever in Internet years, was a Fuji S100FS with a quarter-scale sensor. Almost all the photographs I'll be talking about were made with that camera.

As I wrote in my original review, photographs produced by the Fuji at low ISOs were almost a dead ringer for those produced by a Nikon D200. I was very happy using that camera at anything between ISO 100 and 400. In fact, most of the digital photographs made from 2008 through 2010 that I'm selling 17x 22" prints of were made with that camera. Seventeen by twenty-two inches is not a small print, and my standards are not modest.

This picture is a good example. This photograph depends on low noise, a long exposure range, and well-rendered edge detail to succeed, all things that small sensor cameras are supposed to be unacceptable for.

You don't necessarily need a lot of pixels, either. I'm not saying more pixels aren't nice, just not necessary. On that "newer work" webpage, all the digital prints from 2008 through 2010 came from 9–12 megapixel files. I wouldn't print them bigger than 17x22", but they're fine up to there.

This panoramic photograph doesn't really violate that rule. Yes, it's a stitch of three frames and has a total of about 24 megapixels, but it looks best as a 40-inch wide print. You can't really see any of the detail. This JPEG is one third the resolution of the original file and begins to give you a sense of what there is in that photograph. For compulsive pixel-peepers, below is a full-resolution section. Keep in mind that looking at this on your monitor is like looking at a section of a 7-foot-long print.

In summary, you don't need a large sensor and you don't need a lot of pixels. You can do serious photography of demanding subjects with a double handful of megapixels and a sensor the size of your pinky nail. As the title reads, size doesn't matter.

On the other hand...

...Those who say that there's no point to having more megapixels and/or a larger sensor are equally wrong. Even in small prints (these days that's an 8x10), side-by-side comparisons will show that more pixels are better, up to some phenomenally large number. Think nine figures. Again, I'm not saying you need this. But there is a palpable, visible improvement, the same way that a contact print from 8x10 film would beat the pants off of anything enlarged from a smaller format.

Now, before people start posting comments to the effect that lenses aren't good enough to take advantage of lots of pixels in small sensors or that diffraction prevents you from getting good quality out of small pixels, save your breath. I've covered this in excruciating technical detail in previous columns. I don't need to flog that horse again. It's wrong, okay? You can go up to insanely high pixel counts and insanely small pixels before more pixels stop meaning more detail in the photograph.

Sometimes, there is no substitute for more pixels. Take this picture, also viewable here at about 50% scale. You might be wondering how Photoshop did such an incredible job of stitching the surf together. Well, it didn't. This isn't a stitched photograph, it's a cropped one! I started with a 5K by 6K file from a Phase One back and cropped out the top and the bottom, leaving me with a "mere" 18 megapixel photograph.

Not so incidentally, I can see a visible difference between an 8x10 print of this photograph and one made from a file with half the resolution. Eight megapixels will look plenty good filling an 8x10 inch sheet; 32 megapixels will look even better.

This panorama didn't require a large sensor, just a lot of pixels. Sometimes, though, one needs the acreage. The picture below would look okay if made with a (relatively) small-pixel-count, small-sensor camera, but it looks positively great coming from a Phase One back. Thirty megapixels, huge exposure range, and a lovely signal-to-noise ratio. What's not to like? (Well, for one thing, a camera heavy enough to anchor a modest-sized yacht.) Like the title says, size does matter.

As I said at the beginning, this isn't about your particular needs and working style. If you're one of the minority who's passionately in love with ultra-shallow depths of field, you need larger sensors and faster lenses. If life for you starts at ISO 1600 and goes up from there, your sweet spot is somewhere in the 2/3 to full-frame sensor range; smaller sensors—and larger ones—will usually not be as well-optimized for high ISO work.

But, for most of us, size and pixel count don't matter anywhere as much as some people claim.

Ctein

Ctein's regular weekly column appears on TOP on Wednesdays. This one was written before today's CX-format announcement, although it surely pertains. —MJ

Send this post to a friend

Please help support TOP by patronizing our sponsors B&H Photo and Amazon

Note: Links in this post may be to our affiliates; sales through affiliate links may benefit this site. More...
Original contents copyright 2011 by Michael C. Johnston and/or the bylined author. All Rights Reserved.

Comments

[Chuck Kimmerle]

How dare you introduce knowledged reason into the irrational mix.

[JohnMFlores]

You've got your work cut out for you - we live in a country still drunk on SUVs and 60" televisions.

[Bill Mitchell]

Any thoughts about the relative pixel requirements for color images, and those converted to B&W before printing?
I have some apparently very sharp 16x20 prints from my 8 MP Sony T100 (shot as B&W with the conversion in camera).

[Ivan]

I don't see pros doing any crying about the Pentax Q; in fact, here's one who used a Q for a fashion shoot (mostly successfully, I would say):

http://www.adorama.com/alc/blogarticle/The-Pentax-Q-Stress-Test-Day-1

(Note that he doesn't expect the Q to be anything more than a digital Auto 110, which has been stated by the Q design team as the project goal - not to compete explicitly against m4/3 or NEX. To use an automotive comparison, it should fill the same niche that the Figaro did for Nissan.)

[Ivan]

Addendum... it seems the article I linked to has only days 1-3; the shoot actually went for nine days. Here's the rest of it, with the photographer's verdict:

http://www.adorama.com/alc/blogarticle/13285

[Steve Jacob]

Er, quite. It's rather hard to find anything to add to this appeal to common sense so I won't. Well maybe a bit...but only to support it.

The trouble with gear comparisons is absolutism. The trouble with the gear itself is compromise. You will never build a perfect camera, even for an individual. I know, I have three! My favourite? What day is it?

Why isn't my D700 perfect? Size. Why is my compact not perfect? Noise. Why is my D90 not perfect? Because it's bigger than my compact and noisier than my D700. Why don't they put a D700 sensor in a compact? Er....yeah, right.

Personally an MFT camera is too close in size to my D90, and loses all the compatability benefits which make it a great backup (and second camera on shoots). My compact however really is too far removed, quality wise, from the APSC body.

I think Nikon have a point with 2.7 crop. I may even buy one. If they just come out with some proper controls (a la P7100) and a couple of fast primes. Not yet Nikon, not yet, but I like the way you think.

[Mike Anderson]

Ctein,

If you print a 12 meg file at 17x22 inches that's about 200 DPI. Do you uprez the file, and if so how, or do you print it at 200 DPI?

Thanks,
Mike

[KeithB]

I just wish that they would be more creative with high Mpixel cameras! Why not a mode where I average 4 pixels to get a smaller file? (My wife's S-95 does this in its super high iso mode.) As far as I know, when my SLR creates a lower resolution image, it simply decimates - it does not average.

[Will Frostmill]

If life for you starts at ISO 1600 and goes up from there, your sweet spot is somewhere in the 2/3 to full-frame sensor range;
The first person I thought of was David Dyer-Bennett :)

A little more seriously, kid-tographers, like me when I'm in dad-mode, are very interested in ISO 1600-6400. Lots of life is lived indoors, under fairly dim (red!) incandescent light. Right now, the sensor in the E-PL1 is a pretty good fit at 1600, but I need another stop or two to get the shutter speeds I really want at a middling aperture(f/3.5-4). Bigger cameras need not apply, and shadow noise in hair textures is the enemy.

Will

[Clive Evans]

My buddy says his 7x17!just stomps all over my 4x5

[Dave Kee]

Three and a half trillion photographs have now been taken and only a tiny fraction of the new ones are being printed. Size no longer matters. See http://1000memories.com/blog/94-number-of-photos-ever-taken-digital-and-analog-in-shoebox

[Semilog]

Right on, Ctein.

I'm staying with small-format cameras (35mm film and 4/3 digital) because the quality is as good as I require and I actually carry the gear with me.

The major benefit of bigger sensors is (and always has been) that they are more forgiving of technical sloppiness.

[Ken Sky]

Ctein
I consider myself among the unwashed to whom you are preaching. The situation for most of us enthusiast photographers is we have neither your knowledge or discipline when we are in the heat of photographing. So we expect our cameras to provide us with the elasticity in our files to compensate for our mistakes in PP. For all the criticism, the marketing people at Sony know their potential customers' needs. I want loads of pixels to crop. I want a great EVF. I do photograph in low light without a flash & would like low noise and large dynamic range when in bright light. Why else would I get a new camera and have to hide it from my wife for 6 months;)

[C. Lund]

Interesting

A question, that don't a such has anything to do resolution. We have discussion at Dyxum about raw files bits size. Where there are many who seems to think a that 12 bit raw file, can only contain 12 dephts. I am not really sure that is right. Is that something you, or anybody else could help with more information or a perhaps a link or two.

[swr]

Yes. My Panasonic LX3 is capable of taking good photos.

You can make great, as in fully professional quality, photographs with small sensors.

But why use an interchangeable lens mount. Even the lens cap on the LX3 annoys me. I'm forever thinking about exchanging it for a Canon S95. But I never quite get around to it because it wouldn't be enough of a change to warrant the effort.

IF Nikon had made a camera with a small sensor the size of the LX3 with a built in 24-60mm zoom AND an electronic viewfinder, I would have put myself on the waiting list.

The new Nikon has a built in viewfinder. That recommends it over the Panasonic or Olympus 4/3s cameras.

But the new Sony compact has a built in viewfinder, isn't much larger, and has an APS sized sensor. It's not going to be much bigger than the NIkons AND Sony, unlike Nikon, will probably be able to keep them in stock in the United States.

[David Luttmann]

Well, let's see how that Pentax Q holds up to a Nikon D700 at 17x22....if the sensor size doesn't matter ;-)

[Bruno Masset]

KeithB wrote:
> As far as I know, when my SLR creates a lower resolution image, it
> simply decimates - it does not average.

Your word choice is a bit unfortunate, as "decimation", in the signal or image processing field, refers to a process whereby the original signal is first low-pass filtered — e.g. via simple averaging, or more likely, using proper methods like Fourier transforms — and then downsampled.

To test if your SLR just dumbly downsamples — a somewhat unlikely scenario, as digicams have hardware-based JPEG processors that can compute discrete cosine transforms that are akin to Fourier spectral methods — you might check if the lower-resolution image contains moiré.

Moiré is a tell-tale sign that dumb downsampling without proper low-pass filtering is taking place. DSLRs with movie mode, for example, do not have an image sensor with a bandwidth sufficient to deliver, say, 24 full-resolution pictures per second to the image processing engine. The image processing chips — in charge of elaborating e.g. the JPEG-, MJPEG- or H.264-compressed pictures, — in turn, typically don't have enough processing power to process a 24fps full-res image stream from the sensor.

On a Canon 5D MarkII, for example, 24fps would equate to 24x21Megapixels ~= 500 Megapixels/sec to be digested by the image processor, which is too much.
The Canon 5D2, in movie mode, thus just performs a dumb downsampling of the sensor data by skipping pixels and downloading just a limited number of them from the sensor. The result is typically a movie mode full of insightly moiré — i.e. aliasing, — as the sensor data contains frequency components that are above the movie mode's Nyquist limit.

In still picture mode, however, the image rate is low enough for the image processor to digest all the pixels the sensor is able to deliver. Thus, it's likely that with most DSLRs, the lower resolution images — even in raw mode — are created via a proper LP filtering that takes advantage of the JPEG compressor hardware, eliminating the unneeded high-frequency terms of the discrete cosine transforms that would just cause moiré in the low-res versions of the image.

I thus suggest you take a picture of a repetitive pattern — e.g. a grid displayed on your PC's LCD monitor — with a resolution such that it nearly causes moiré in full-resolution mode. If you then set your camera to produce pictures of lower resolution, and if your camera performs a dumb downsampling without proper LP filtering, then moiré should certainly appear.

[ctein]

Dear Mike,

No, I'm not upsampling; I'm printing out at about 200 PPI (*NOT* DPI). Surprisingly (well, to me anyway) that looks sufficiently good in those large prints.

I did examine various upsampling message and software several years ago and reported on it in a three-part article, here:

It's Bigger, But Is It Better? Part I - III

http://tinyurl.com/28wkub

http://tinyurl.com/26mybt

http://tinyurl.com/66rgwp

My broad conclusion, and nothing released since has changed that, is that there are subtle differences between the different approaches to upsampling but the key word here is SUBTLE. Claims of amazingly better image quality and detail for certain software packages are either the result of looking at a small subset of all images or wishful thinking.

(Anyone who has a particular interest in the subject or feels the need to argue about this should really, really, REALLY read all three of the aforementioned articles, plus the associated comments, before posting. The odds are strong that whatever you want to say has already been discussed.)

~~~~~~

Dear Keith,

I'm not clear on what that gets you over doing the averaging in Photoshop. A bit more convenient, to be sure, but does it actually get you a better image doing it in-camera?

~~~~~~

Dear Will,

Heheh, DDB was precisely who I was thinking of when I was writing that passage.

~~~~~~

Dear Clive,

He's right.

Do you care?

pax \ Ctein
[ Please excuse any word-salad. MacSpeech in training! ]
======================================
-- Ctein's Online Gallery http://ctein.com 
-- Digital Restorations http://photo-repair.com 
======================================

[Dennis Allshouse]

Bill trust your eyes. And your own goals.

[David Dyer-Bennet]

Steve, an EPL-2 with Panasonic 20/1.7 pancake on it is only a hair thicker than an LX3 -- I consider it very compact, a far cry from my D700. And it produces usable files from those dark indoor places where my LX3 consistently disappointed me (I've sold 20x30 prints of LX3 shots I took outdoors).

But it's not a backup or second camera much, no.

A D90 would be several times too large to fill the role my EPL-2 does for me.

[Timprov]

I don't think it's fair to say basically "if it's dark and your subject isn't still, it's your problem." That's a rather real situation. Similarly, sensor size doesn't really matter for image quality as long as you're in the normal range. Outside that suddenly it does, because wide lenses are harder to make the smaller you make the capture surface, and tele lenses are harder (or at least way more expensive) when it gets larger.

[Kostas]

I have been following quite a few conversations on various sites and I haven't seen any comment that the small sensor in the new Nikon cameras is unable to create good quality photos at low ISOs, so I think that's a bit of a strawman. I agree with you regarding the pixel count being more than adequate, but again I have seen only few people complaining about that.

The choice of sensor, and the new cameras in general have a number of real shortcomings. If Nikon had gone for a DX size sensor like Sony did with their mirrorless cameras, not only you would have gained 2 stops of ISO/DR. More importantly you would be able to use all your existing lenses with the usual 1.5 mutliplier that we're used to. Now we (well actually not me since I've ruled out this camera, but whichever Nikon user decides to get this) will end up with some really wierd focal length equivalents from their lenses.

Another problem, is that as the sensor size decreases, so do the benefits of having interchangeable lenses. Since I'm not going to get a shallow DoF or good low light performance even with the 10mm (27mm equiv) f/2.8 lens which is the fastest lens for the new format, why shouldn't I get something like the XZ-1 (which actually has a 28mm f/1.8) for half the price instead?

So what are the benefits of using a smaller sensor compared to the other mirrorless cameras (m4/3, NEX)?
- Price: Since large sensors cost a lot to make, a camera with susch a smaller sensor is going to cost a lot less, right? Only it won't. At $900 the V1 costs more that nearly all m4/3 cameras, about the same as the Sony NEX5 with an APS sensor, and more than the D3100 DSLR
- Size: Again, contrary to expectations, the Nikon cameras are heavier and bigger than the compact m4/3s (i.e. the Olympus ones, and the GF series) and the NEX5 with the APS sensor.Even including the lens, the Nikons are heavier and about the same size as the m4/3s.

To conclude, the new Nikon cameras:
- are not meant to be a 2nd camera for existing Nikon SLR owners, and Nikon are making it quite clear that this is not the target market
- are worse than their mirrorless competitors in every possible way (size, price, and performance)
- are probably closest in performance to the premium compacts, but cost twice as much.
So who would buy this camera and why?

[mark lacey]

I'm certain you're right, and that's why I really can't find anything to replace my SONY R1 that will make better photographs for ME.

[Kostas]

Why don't they put a D700 sensor in a compact? Er....yeah, right.

That's actually a question that a lot of Nikon users have been asking. Leica has proven that it's possible, and Nikon could do it at a much more reasonable price (no need for a RF, an EVF would do just fine).

[Richard]

Of course you are right the size of the sensor doesn't matter to actual image quality, heck I have a 30x30 inch print from a 6mp file sitting beside another 30x30 off a 6x7 negative. The thing that does matter (and maybe I will have trouble explaining what I mean)is the physical chip size in the way it sees the "world" or how much it sees of the world. An image taken with a 4x5 camera with a standard lens (for that format) will see so much more of the scene then would a tiny sensor with its standard lens. To get the same amount of the scene to fit on the small chip it needs to be bent, warped, to get everything in. That changes the "look" of the image between the two camera sizes, not a difference in quality just in the look and feel of the image. So perhaps the statement could be made that the size of the chip doesn't matter to quality but it does matter to the aesthetics of the image.

[ctein]

Dear C. Lund,

Exposure range is how tall the staircase is. Bit depth is how many steps there are in the staircase.

More is always nice, but it isn't always necessary. And, at what cost?

pax / Ctein

[Dale Baskin]

I often have friends ask me if one or another camera is a "professional" camera before buying one. (Lord knows, you wouldn't want to get caught toting around one of those amateur ones...)

I patiently explain that if you are a professional photographer and a particular camera proves to be a good tool for the type of work you do, and you can make money with it, then yes - it is a professional camera.

Nobody ever seems satisfied with that answer, but I can't think of one that's any more true. I have seen people do amazing work with an iPhone camera. In their hands it's a professional tool.

[Eric]

Ctein, I'm curious if your photos that you're printing as 17x22 prints were shot hand held or on a tripod. That is, how careful were you when you exposed those images?

[Crabby Umbo]

I kick myself a few years ago for letting an 'expert' talk me out of a 4/3rd's camera, when I far more appreciated that format (after years of shooting 6X7 and 4X5), than the 35mm long, thin, format, which has always yielded the least usable size for me, and always gets cropped. Now that I'm far more versed in digital mayhem, I realize that the 4/3rd's size output would be of no particular difference in my work as produced by magazines and catalogs, vs. the APS-C size I got talked into. I'm willing to bet that there might not be much of a difference either between anything I'm using and the new Fuji X10, whenever it comes along. The ability to set film preferences like Astia, Provia, Velvia, is something I've been looking for since day one, and I'll bet the zoom lens on it comprises about 99% of the subjects and situations I do...

...I might end up being a pro that just uses that thing for all my work!

[ian]

I got some nice results from a Fuji 3800, back in the day when I was still mostly using film: the secret was it locked at 100 asa equivalent, and that was your lot. The sensor never really got hot, if you like.

[ctein]

Dear Kostas,

I wrote my column before the Nikon announcement. I don't care anything about the new Nikon camera, I've barely read the article on it. My first paragraph was referring to any number of other camera announcements and the ensuing comments over the past month or so.

I have no opinion about the Nikon one way or the other, directly or by inference.

~~~~~~~~

Dear Eric,

A mixed bag. Some tripod, some not. And I have no aversion to pixel-peeping like mad in the field to make sure the resulting image is really rock-steady and deleting it on the spot if it's not.

~~~~~~~~

Dear Tim,

I don't think I said anything "basically" like that at all. And, you know I do more than my fair share of available-darkness photography.

Wide lenses are not harder to make for smaller formats. What's harder is making shorter-focal length lenses when the back-element-to-image-plane distance doesn't shrink. But small format cameras shrink that distance, so it stops being a problem. Anyway, it's no biggie making, e.g., a 5-6mm lens for a 1/4-scale sensor.

You're right, though, about long getting a lot more expensive as format goes up, 'cause of the huge size of the optics.

And that *is* one of those "it's your problem" things. If you favor wide, larger format isn't as big a technical problem (although the glass can still get breathtakingly costly, as 8x10 Super-Angulon owners will attest).

But, if you like long lenses, liking large format is a distinct handicap. It certainly was for me. I do prefer long, but with the Pentax 67 I could never afford to own, nor was muscular enough to lug around, a prime longer than 150mm (35-equiv). Now I'm at half-scale digital and a 400mm (35-equiv) is a barely noticeable lump in the bottom of my shoulder bag. I am happy.

But it's not a pro or con argument. It's just me.

pax / Ctein

[Vincent Ismail]

I see many people have problem with explanation with framework like this:

"A is A, but A is not A. Considering the assumption to each conclusion depends on which approach of time-space method is being used"

To have a lethal debate over technical stuff at least some science and mathematical thinking is involved. Not just like religious mindset which only recognizes black or white, good or evil :).

Keep going on Ctein!

[Tom V]

I sold a pro photographer a Canon Elph SD1200 to replace a SD1100. She was using it to cover a pro bicycle racing tour. She raved about the color rendition.

A professional camera is a camera you use to make money with.

[William]

Larger sensors mean more photons collected. Photons generate signal. Given a fixed amount of light, why not collect the best signal possible?

I do not understand how increasing the sensor area by twenty fold (CX vs FX) is of no practical importance in everyday photography.

Noise is essentially constant. That is: the read noise levels stay about the same as sensor size increases (for a given senor technology). The larger the sensor the higher the signal-to-noise ratio. Collecting data with the highest practical signal-to-noise ratio is important. Digital photography estimates the number of photons that interact with a given sensor site. Nothing beats signal-to-noise ratio when you have make parameter estimates.

Finally, sensor size is directly related to dynamic range. The lower the signal-to-noise ratio, the more bits are spent digitizing noise instead of signal. Wasting bits is not good.

However, when you have more signal than you know what to do with, the photon collecting advantage and superior signal levels with a larger sensor are of no value. There are situations in photography where there is an abundance of signal and throwing away the 20 X advantage from a FX sensor is irrelevant. This is especially true for subjects without shadow regions.

Otherwise, the larger the senor area, the larger the signal-to-noise ratio. Large signal-to-noise ratios mean better data to render an image or a print.

[ctein]

Dear William,

In a word-- no.

In many more words...

"why not collect the best signal possible?"

So, back when you were using film, you used an 11"x14" view camera, right? And, I would guess, poo-poohed those desperate enough to be using something smaller.

"Noise is essentially constant. "

Only when there are no differences between sensors, amplifiers and signal processing circuitry. No commercial camera yet performs at theoretical physical limits of counting statistics-- the best are 2-3 stops shy of that. Most are even further below that.

At low ISO's the Fuji performed identically, in every respect (noise, resolution, exposure range) to the Nikon, with 1/6th the sensor area. All cameras are not created equal.

And better is frequently the enemy of good.

pax / Ctein

[Bernd Margotte]

I absolutely agree with this post. Over the last days I have been looking into the quality difference between a compact size camera (a Ricoh GX200) and a larger format cameras (a Leica M8 and a Canon 5DMK2) and I was astonished on how well the Ricoh performs; actually, there was only marginal difference in sharpness between the Ricoh Zoom and a Leica Summicron lens. Don't believe me? Look at the examples of this report (scroll to the end):
http://www.berndmargotte.com/technical/sensor_de.html
The report is still in German, I had no time yet to translate it yet.
But one important thing which I think is missing here is the effect of sensor size to depth of field. Small compact cameras (and I talk here about really small, like a 1/1.7" sensor) have depth of field fully open which is eqquivalent to something like aperture 16 on a full frame sensor. There are also examples in the report which show this. I think this is one of the biggest draw backs of small sesnors (if you like to have small depth of field); it could also be a benefit, if you like infinite depth of field.

[Brian_F]

That's kind of a relief. Here I've been thinking the results I was getting from my Fuji s200exr were very good, in some cases better with my 645 scans. Yet there was that back of the head voice telling me they couldn't possibly be as good as I thought and I must be missing something. Now I can focus my anxieties elsewhere.

[Ed]

Now that is the first article under 15 Mwords that explains the lot in terms even the biggest git and pixel peeper can and will understand (unless there are even bigger gits then I would have thought possible). Basicly I use a rule of thumb for determining the pixelsize/sensor size......there is no rule of thumb. Sometimes I want a shalow depth of field so I use 1.4 50 on a GF1.....sometimes I want macro with some depth of field so I use a LX3.......all works fine at sizes till A3 plus. And 10 versus 12 Mp.....not a problem its only a 3.25% decrease in possible picture size. And under 200 ISO noise is not an issue anyway (for the GF1 I can go up to 800 and still be happy and that is deep in Tri-X territory). For larger prints I need a bigger sensor than (my) money can buy and so it's analog for me using a LF camera and a nice scanner.....Velvia 50 and 100 film for color and Delta 50 and Adox CMS for black and white (pixel peepers should try that film once in lifetime and be cured of their peeping habbits for ever :-)). And the rest is up to photographic skils......and these can't be bought in box, you know.

[Aaron Greenman]

All the new compact camera systems have really re-opened an interesting debate.

Ultimately, though, from the comments above, isn't the best-designed ("compromise") product to take care of size/quality needs the Leica M9? the only thing really being given up is video and autofocus. Otherwise, the specs are compelling (including the price): compact body, 18mp (enough), large sensor (full frame), good DR

If Leica would take the plunge and be able to produce them in sufficient quantity, lose the unnecessary snob appeal, and through volume lower the price significantly, there would be a lot of happy photographers out there.

Problem is once you have an M9, you're forced to focus on just taking good photos, as the commitment to the system precludes you from always going after the latest, greatest new "kit".

[Steve Smith]

Now, before people start posting comments to the effect that lenses aren't good enough to take advantage of lots of pixels in small sensors or that diffraction prevents you from getting good quality out of small pixels, save your breath. I've covered this in excruciating technical detail in previous columns. I don't need to flog that horse again. It's wrong, okay?

Thanks for posting that. I have often seen posts writing about a sensor out-resolving a lens and thought it was nonsense. I can't recall any lens which was ever out-resolved by any film (although I'm sure some do exist) so to suggest a lens can be out-resolved by a sensor does seem ridiculous to me.

[Mike Johnston]

Just thought I'd mention that I caught a lot of flak for saying essentially the same thing back in the Middle Ages of 2004, in an article I wrote for Black & White Photography magazine called "Eight Megapixel Image quality." Basically, the premise of the article was that you could get just as good image quality from an Olympus C-8080 as from a Canon 20D DSLR...until you started getting into the higher ISOs.

I had done the tests, but this didn't "seem reasonable" to readers, who disliked the column and its conclusion.

Mike

[Brad Nichol]

I learnt many years ago that posting such heretical content as Ctein has posted here on photography forums is something one does with great courage. My attempts met with abuse, name calling and worse. Some would have gladly burnt me at the stake.

I always tell my students " unless your using a pretty crook camera poor image quality is most likely your fault.....now let's see how we can fix that!

Once they see the prints and they find out what cameras were used the arguments stop dead!

A few points are worth making.

The higher the pixel count the more you need to pay attention to focus, camera movement and exposure if you want to liberate the power of those pixels. Many people pay too little attention to details and thus never get close to realizing the capability of their camera.

The whole argument about high ISO performance and small sensors is largely stupid, if your shooting for deep DOF you have about 3 stops or more advantage over FF ( ie f4 on a compact has about the same DOF as f13 on FF. In other words you don't need high ISO s on a compact normally.

Way too much time is spent pixel peeping at 100 percent and way too little actually making prints. The print is the thing!

A little noise of the right type never hurts an image, noise can be good, embrace it. Many cameras that produce utterly noise free images at 100 percent on screen look decidedly plastic in print.

I tell folks that cameras are like tools in the workshop, sometimes you need a saw, sometimes a chisel, sometimes hammer.....use the right tool for the job you have at hand and they will all perform their job perfectly.

[Semilog]

"Ultimately, though, from the comments above, isn't the best-designed ("compromise") product to take care of size/quality needs the Leica M9? the only thing really being given up is video and autofocus. Otherwise, the specs are compelling (including the price): compact body, 18mp (enough), large sensor (full frame), good DR"

You're also giving up accurate framing. And accurate focus with even moderately long lenses. And live view (which is one of the best tools available for achieving accurate framing and focus). And video. And modern exposure metering. And image stabilization. And high-ISO sensitivity. (Of those last two, wouldn't it be better to have at least one?) And sure, it only has three blinking lights, but it clutters the framing area with totally unnecessary extraneous frame lines (e.g., for the 135, a lens that it can't even focus accurately anyway).

Look, I realize that there are some freaks out there who are willing to make these compromises (you want compromise? I mainly shoot film and an M6), but it's simply absurd to say that video and autofocus are the only things the M9 is missing vs. other cameras. There are a lot of really good reasons why the Nikon F and its progeny nearly competed the Leica M into dust, and almost all of those reasons are just as compelling in 2011 as they were in 1959.

"If Leica would take the plunge and be able to produce them in sufficient quantity, lose the unnecessary snob appeal, and through volume lower the price significantly, there would be a lot of happy photographers out there."

This is like telling Ferrari that they really ought to ramp up production and lose the snob appeal. I'm sure they'll get right on it.

[Jon]

http://theonlinephotographer.typepad.com/the_online_photographer/2008/01/more-keychain-m.html

"Take my word for it: this is a must-have for any serious connoisseur of fine photography."

The current version (VQ-2005) now has a 2 Megapixels CMOS sensor, an optical viewfinder, takes SD cards and is around $21.

"And better is frequently the enemy of good."
:o)

[Patrick Perez]

@Ctein, re: William and getting maximum signal. I interpreted William's metaphorto film not as format, but as film choice in that a given film will have the same characteristics regardelss of format from 135 to 8x10 (obviously there is some change in actual formulas, but I'm talking principal here). So one would choose the best signal capture for the shooting conditions (Ektachrome 200, for example) and choose format for needed enlargement. Anyways, this is how I interpreted his idea.

Patrick

[Steve Jacob]

@David

You said "Steve, an EPL-2 with Panasonic 20/1.7 pancake on it is only a hair thicker than an LX3 -- I consider it very compact, a far cry from my D700. And it produces usable files from those dark indoor places where my LX3 consistently disappointed me (I've sold 20x30 prints of LX3 shots I took outdoors)."

Ah but you see, the rub is this. I dont use a 20mm pancake ;) Being restricted to a single pancake lens is simply another compromise. It just depends if its a compromise you don't mind making. For some people that can be weight, cost or performance. There is a place for compacts, MFT, 2.7 crops and FF.

[Mario]

I totally agree. It's not all about sensor size or pixel count. Provided the camera has the ability to do the specific task you're using it for. Good Article.

[Clive Evans]

Ctein
I know he's right, now I want one too!
Where does all this end?
Clive

[Doug Dolde]

I guess I'll have to stick with my Aptus 75S. It's sensor is twice the size of full frame at 36x48mm. It's paid for so until it dies I'll just keep making 24x32 prints at a whisker over 200 dpi.

[ctein]

Dear Bernd,

I did bring that up in the next-to-last paragraph. I only devoted one sentence to it, but that's all it deserves. I know it's hard for many shallow-DOF fans to believe, but they really are a very small minority of all photographers. It's one of those specialized niches, like folks who predominantly work at ISO 3200 or above, or folks who do pretty much only black and white photography. Nothing wrong with any of those, and it can affect your choice of camera, but it doesn't speak to broader issues.

The oxen I am goring in this post are those belonging to the folks who make blanket declarations about small or large sensor cameras as if they were the center of the photographic universe (or at least quoting absolute laws of nature).

~~~~~~

Dear Steve,

The resolution demands of typical sensors today are considerably higher than that of any commonly-used photographic film. It's from that that derives the mistaken notion that camera lenses can't live up to the sensor or benefit from a higher-resolution sensor.

~~~~~~

Dear Mike,

I also seem to recall an article published in some journal of ill repute written by some ne'er-do-well that demonstrated that with sufficient (read: extraordinary) care one could achieve 4 x 5 quality results from a 35mm rig. Obviously insane.

~~~~~~

Dear Jon,

I was so wondering if any readers would remember those articles! Oh, if only all readers had your memory; Mike and I would be obligated to repeat ourselves far less often.

~~~~~~

Dear Patrick,

If that's what William meant, then that is even less congruent with reality. The vast majority of photographers, both amateur and professional, settled on one or two formats and chose their work accordingly. 8 x 10 photographers did not suddenly run out and buy 35mm kits when someone approached them with a job that demanded 35mm; conversely 35mm photographers did not usually become view camera users when approached about a studio shoot. If they didn't think they could do it with the equipment they had, they would pass the job to someone more qualified.

There were a few exceptions to that. But this is overwhelmingly the norm. Folks settle on a format range and work with its pluses and minuses.

Similarly , few photographers used an extremely wide range of films. They settled on a few types they understood well.

There probably are some photographers out there who were well-versed in a dozen different films and owned (or at least regularly rented) everything from 35mm through 8 x 10 view camera equipment. With 30,000 readers here, I bet we'll even hear from one or two of them. It's not even close to typical, though.

~~~~~~

Dear Clive,

It ends with the Enlightenment born of understanding that even though he's right, you don't care. [Guru-like smile]

pax \ Ctein
[ Please excuse any word-salad. MacSpeech in training! ]
======================================
-- Ctein's Online Gallery http://ctein.com 
-- Digital Restorations http://photo-repair.com 
======================================

[Oren Grad]

Ctein
I know he's right, now I want one too!
Where does all this end?
Clive

It ends with you getting a 7x17. An 8x20 or 12x20 would also be OK.

:-)

[Steve Smith]

I had done the tests, but this didn't "seem reasonable" to readers, who disliked the column and its conclusion.

People don't like hearing the facts when they have spent money on the more expensive option.

[David Dyer-Bennet]

Steve, since the LX3 was a 24-60mm-e lens, a fixed 40mm-e is not terribly different. Even less difference now than back in the film era -- I can stitch to get wider FOV on static scenes much more easily than I could with film.

[Marcin Wuu]

I dare to disagree with at least one point -- as a portraitist (par excellence ;)), I don't really feel like I'm in some sort of very smal photographic minority or specialized niche. On the contrary - there's lots and lots of people taking pictures of other people, and using the DOF as their main composing tool. My point being, with all due respect - there are quite mainstream applications of photography where size does matter and fuji f100s just wouldn't cut the deal.