The Megapixels Poll (Follow-Up)

There's not a lot of point to polls like yesterday's. The bell curve is right where I thought it would be, right where it should be. (There's even a little shelf in the bell curve at 22–24 MP, representing D3x/A900 aspirations.) All I was wondering was whether 18 MP in a Rebel is really going to be a significant draw. That premise looks a little soft, but still safe.

Polls like this one basically represent people saying "What I've got right now is perfect for me...right now." Thom Hogan uses polls for projections—that is, he knows that the curve would have peaked at a lower number if the poll had been taken five years ago, and he knows it will peak at a higher number five years from now. He looks at the rate of change. Because really, pixel counts (given today's technology continuing unchanged) are almost arbitrary numbers—like computing power and drive space, the numbers are normalized only for now, only in terms of this point in the march of progress. At one point, six megapixels was the bleeding-edge aspirational number, and now, only 2.7% of poll respondents would pick that level. (And I'm kind of surprised it's that high.) The number has moved all the way through the bell curve.

Complicating the projections for cameramakers, though, is that the bell curves don't move infinitely. At some point every sensor will have enough resolution, and people will stop caring. Then, the manufacturer who invests in a bet that more megapixels will continue to move product down the road (just because more megapixels have moved product up till then) will be out of luck. Predicting that point of sufficiency and not investing too much in overshooting it is one of the dicier tricks of product design. (An egregious example of that kind of overshoot on the part of manufacturers: APS. Pronounced "oops.")

Looks like most people are pretty happy with what they have now, though.

For now.

Mike

Send this post to a friend

Note: Links in this post may be to our affiliates; sales through affiliate links may benefit this site. More...

Comments

[Rubén Osuna Guerrero]

Mike's comments are intelligent and insightful.

I think the limit will be determined by the practical use of photographies and, mostly, by domestic use. Professional cameras will adapt to clients' requirements.

Domestic printers (print size and resolution) and visual acuity may determine the number of pixels needed/wanted/significant for most cameras. Ctein wrote about visual diferences in prints at optimal viewing distance, and the largest printer for domestic use is A3+. It is easy to do a calculation: 600 ppi and 13 x 19 inches... 88 megapixels! 360 ppi and 13 x 19 inches... 32 Megapixels!

[Steve]

I have no idea what poll you're talking about. Maybe a link would help?

[Mike Johnston]

Steve,
Just scroll down a bit. From yesterday.

Mike

[David Lykes Keenan]

I have an M9.

Seriously, I think that this last major digital camera I will ever buy.

Seriously.

An "M10" with Nikon D700 high-ISO capability might be nice but I really don't think I'd need it (or be compelled to ditch the M9 for it). I have a D700 for those situations anyway.

Okay, I'll probably buy a second M9 as a backup at some point.

But when it comes down to a camera to take the kind of pictures I love to take, I can see using the M9 for the rest of my life. I honestly can't see how it can get any better then this.

[Mike Johnston]

"I have an M9. Seriously, I think that this last major digital camera I will ever buy. Seriously."

Want to bet a hundred bucks?

(The humor in that is that I could only win such a bet, never lose.)

I have a friend who said the same thing about the M8. He's moved on. And I remember a man standing in a camera shop saying the exact same thing, in tones of joy and rejoicing. The camera: a Contax 167MT, the first year it came out. Bet he's moved on, too.

On the other hand, I did meet a guy in the '90s who'd been using the same screwmount Leica since the late '40s. So I guess it is possible.

Mike

[Richard Man]

You polled for the wrong thing :-) You should ask "what digital camera would you like to see that is 'doable' given the current technology."

Now that there is a digital M and that various cameras can almost fulfill the dream of a DMD, wouldn't it be just fantastic to have a digital XPan?

[Michael C]

I really think the megapixels I have now (Pentax K20D) are more than enough - but I don't do or intend to do any large-size prints.

Even if I did, I understand that you need _less_ dpi not more, unless you want something that people can view at a fairly large (and comfortable) distance _plus_ has an enormous amount of detail _if_ a viewer whats to get nose-close to the print.

In that case, I'd say most subject matter really doesn't warrant very-high dpi anyways...
What? A very detailed landscape, a la Ansel Adams? A very detailed cityscape? Exquisitely delicate patterns?

I am not at the level of doing these anyways :-) so I'd rather have a few great lenses than more megapixels.

mike c

[Martin Doonan]

I have always been narked by the way developers have used advances in technology to add geegaws rather than true function or speed. Computers an example - my current whiz-bang machine is no faster in some respects than my ancient DOS computer was. Broadband internet has meant more flash and graphics rather than genuinely fast content. Cameras has meant more megapixels rather than core function, battey life etc.
In many ways I am disappointed in all the cameras newer than the one I have and I see no move to improve the areas I want fixed.

[hugh crawford]

"Now that there is a digital M and that various cameras can almost fulfill the dream of a DMD, wouldn't it be just fantastic to have a digital XPan?"

I'm waiting for my digital Speed Graphic, although I'd settle for a digital Mamiya C220.

I was voting for the gazillion megapixels , but really the digital part of digital cameras is not so much of a problem as the camera part. The pretend sports/news photographer wanabe market segment is well served, but the folks who were perfectly happy with late 1970s gear and would like to replace it with something digital have to do a lot of improvising.

[Kurt]

I think the higher the pixel count the better because it might force photographers to more closely edit their shots.

I'd be happy with enough pixels to output a 13 X 19 at 360 ppi. Of course there's cropping to account for too.

[Scott]

If my mysterious benefactor really wanted to help my photography they would just give me the cash so I could quit my job. The time would be much more valuable then a new camera.

[Andre]

I'm pretty happy with the now mediocre 10MP my Pentax K10D is capable of capturing and since there are several cameras that can pass as a DMD I don't have anything practical to long for anymore. I answered 32-56 just because I think that the digital equivalent of a Fuji GW-690 would be fun to mess around with. That pixel count is sufficient to blow anything I've used out of the water while still allowing me to refrain from upgrading my computer for a little while longer.

[Jeremy]

I thought I was a bit on the outer when I thought more megapixels were better (at least from reading the comments on this blog).

It seems I'm in the slight majority of people who want "more than 16 Megapixels" (50.1%) vs those who want less (49.9%)

[Iain Dawson]

How about a poll on sensor shape ? What proportions would people prefer? 3:2, 4:3, 1:sq.root 2 (A series paper proportions) etc. Personally I am quite satisfied with about 10 or 12 megapixels and regularly print up to A2, even from small sensor cameras like the Canon G series. What I would really like to see is a square sensor of say 16 megapixels with a viewfinder that can either shoot a square image (for the days when I miss my Hasselblad) or can be masked to produce either a horizontal or a vertical A series proportion image without turning the camera. This would improve handling, especially tripod use, and the rectangular images would be a better fit on A series paper.

[Robert Noble]

It often seems to come down to how manufacturers can make updated versions of their cameras appealing enough to compel buyers to trade in their perfectly peforming two-year-old models: should the new model have more megapixels or lower noise? Whichever choice they make will be criticized by those who wished they'd made the other choice.

For example, in the dpreview review of the Nikon D3S (published yesterday) they basically said that the updated D3 is as good a camera as any in the 12mp class-- but they kinda wish Nikon had put in more megapickles instead of bumping the top ISO from 25,600 to 102,000:

"My problem with the D3S isn't what it can do - which is in all respects exemplary - but what it can't. Even when the D3 was announced in 2007, some commentators expressed surprise that Nikon didn't go with a higher resolution sensor.

"Almost three years later, and 12 million pixels looks even more conservative alongside newly-minted competitors like the Canon EOS 1D Mark IV. Nikon's view is that in this section of the marketplace, better image quality at high ISO settings is worth a penalty in resolution. Anyone that shoots regularly in low available light would probably agree, but it can't be denied that the [D3S's] extra high-ISO boost answers a criticism of the D3 that no one really had. After all, it's tough to argue that the original D3 wasn't good enough at high ISOs..."

Of course, no one knows what Nikon has up its sleeve. The D3S may be a mainstay for the "low-noise" market and a forthcoming 24mp D700x-or-whatever-it's-called would be a more affordable version of the D3x for the "high-resolution" market.

[Mike Johnston]

Scott,
Good point.

Mike

[John Roias]

Of course, if you had asked whether I would bother upgrading if any given camera could produce high quality results without the need for more megs, the answer is no. Why would would I? This happened with 35 mm film cameras, that reached a point where only the type of films (to some degree) and the user mattered. 6, 8, 10, 12, or more megs, as long they produce top notch results that are comparable, who cares. At some point it is all hype. Maybe the Foveon sensors will produce top results with only about 8 megs. It's not the amount of megs, but the quality of the results that counts. In the mean time I will keep using my now lowly Pentax K10D. I only shoot jpegs anyway.

JMR

[Stan B.]

I just wish the good people at Ricoh would have invested their R&D in a B&W digital only, instead of their multi-aspect, multi-ratio, multi-format Rube Goldberg debacle. Cheaper to make, and anything will probably outsell that(those) thing(s).

[Evan]

I voted 8-10, not because I wouldn't be happy to have more, but because I'm aware of the trade-offs. I want better low-light performance more than I want a zillion pixels.

I currently shoot a 6.1 MP K110D, which is honestly plenty for me. I'd take more, but don't need them. I need better low-light performance, particularly chroma noise. And shake reduction - that's a purchase decision I regret.

[petavoxel]

Camera-makers have done an excellent job—in fact, a pathologically good job—at getting all of us to use megapixel counts as a proxy for the question that truly matters:

"Are my photos detailed enough to convey my creative intentions?"

But megapixel counts are demonstrably an incomplete way to evaluate that.

Now, some beautiful and evocative photos have been made with Holgas. It's hard to argue that the emotional impact of HCB's photography would have been improved if you quadrupled the lp/mm of his images. But there are certainly subjects where capturing every nuance of detail is essential.

I think one point that needs to be made is that once pixel pitch drops below 5 microns or so, diffraction and lens aberrations become a serious limitation. For those who do require the highest achievable detail, the answer isn't more, smaller pixels, but rather larger sensor area.

[Evan]

I should say - by need I mean "would really like". All I need is food, water, and shelter.

[Jeff Hartge]

I know I threw this tantrum before, but...

I really believe that the primary means of image display will be through large sized LCD panels by the end of this decade. I can see a world where most homes have multiple lcd frames of varying sizes. Further, I imagine the lobby of nearly every small to mid-sized business with multiple displays (many venues have them now).

I would be surprised if the consumer versions were more than 12-15 megapixel displays. Even the professional versions would probably not be much higher than 24 mega-pixels (a 6 foot by 4 foot display at 100 ppi would be about 35 mega-pixels). I think 100 ppi displays would represent a level of sufficiency for most people.

Its hard to imagine that the majority of people would not choose the high contrast of an lcd display over the high resolution of a print once the display pricing is in line with expectations. I believe that the end of the camera mega-pixel race will come via the widespread adoption of a new means of display and not by hitting the limits of sensor technology.

I see the transformation of roadside billboards as a harbinger of the trend above.

[Christian]

"What I would really like to see is a square sensor of say 16 megapixels with a viewfinder that can either shoot a square image (for the days when I miss my Hasselblad) or can be masked to produce either a horizontal or a vertical A series proportion image without turning the camera"

I love this idea!

[J Hayes]

While sensor pixel count has been growing, although slowing, the real changes have been behind the scenes in software for image processing, where much of the future development lies. HDR is obviously coming, noise reduction, distortion and CA correction are available already, but then what? I would like to see cameras with software options - allowing plug-ins like PS or even custom developed operating systems. Then we'll see some cool stuff!

[ctein]

Dear Stan,

I won't say a dedicated B&W camera will never happen again, but you have to understand the B&W market is truly miniscule. Even 20 years ago, color film outsold B&W by 100:1. I'm not exaggerating! And the steady trend, starting in the 1970's has been towards more and more color and less and less B&W.

B&W may be artistically important, but market share? Beyond insignificant. A company's gotta decide they want to do it just because they want to, not because it will outsell anything else they could build. It won't. And y'know, they might; companies still come out with new medium format film cameras, and that market is miles beyond moribund.

No matter how much you may hate the Ricoh camera, it'll still sell better than a B&W dedicated camera would.

pax / Ctein
==========================================
-- Ctein's Online Gallery http://ctein.com
-- Digital Restorations http://photo-repair.com
=========================================

[James McDermott]

The blogs yesterday were full of the sumo stand-off between Nikon and Canon flagships, and which of their squillion megapixel sensors resolved the best detail at a million ISO. The exciting news for me yesterday? Cosina are bringing out an SLII (chipped) version of their delectable 90mm f3.5 Apo-Lanthar - in K mount, among others.

I just can't get excited about megapixels any more. But a new, drop-dead gorgeous, metal, manual focusing lens that'll last forever? Oh Lord! It WILL be the last 90mm f3.5 I ever own.

[David Bostedo]

Stan or Ctein - Would you really gain anything significant by getting a real monochrome sensor? Are there really that many artifacts from the demosaicing and the conversion to black and white? Would you really gain any efficiency in light collection without the bayer filters?

My suspicion is that there would only be a small gain in resolution of a monochrome sensor over a color one, and a small gain in efficiency (lower noise) from not needing the color filter array. It would be interesting to know those answers for sure.

So in addition to the small market, there may be only a small advantage over just converting color images. Plus you'd have to go back to using colored lens filters, and lose a lot of the post-processing flexibility.

[John Banister]

I lust after square centimeters more than I do megapixels. I want to capture the image I see in the viewfinder, and I just enjoy looking into a bright 6x6 viewfinder so much more.

Well, it's not quite that simple. It started out that way, but now I want to capture the images formed by my lenses for 6x6. I don't think a TLR with a viewing lens from a Rolleiflex 2.8F and a taking lens from a Mini-Digi would work for me.

I have an old Imacon Flexframe 3020, and while it's not very convenient to use, I very much like the images formed by those 15 micron pixels. So, I'm with Ian and Christian on the square 16 meagapixel sensor, though if they made it three layers deep and called it 48, I'd hardly complain.

[Tim]

I also thought the poll was a bit confusing, so I answered >56MPel because my ideal camera would involve a *full-frame* 5x4" digital back - I'm sure it's possible, just prohibitively expensive.

Then again, my dream camera would also be made from Lewisian gneiss...

[ctein]

Dear Petavoxel,

You need to read this column:

"Why 80 Megapixels Just Won't Be Enough..."

http://tinyurl.com/c4dj5n

Matching the pixel size to the lens resolution is an utterly wrong-headed analysis, conducted by people who don't understand how this stuff really works. A case of a little knowledge, etc., etc.

pax / Ctein

[Zach]

Am I the only one who wanted a 56mp sensor that is able to pixel-bin* down to 12 exquisite mp on demand?

(* I don't /really/ know what pixel binning is)

[Robin Harrison]

We all know that not all pixels are created equal, but still we persist with a single numerical figure to abbreviate sensor 'quality'. It's the same situation for sensor 'ability', where we use a nominal and semi-ficticious high-ISO limit.

I'd hope that the emergence of DXO-mark and other analysis tools might lead to a widely-accepted metric of 'quality/ability'. Ideally this wouldn't be a single scalar value (which obfuscates so much, in the same way an arithmetic mean only tells you one piece of information about a distribution of data), but I guess anything richer would end up as a tool for the minority. After all, you don't see the average photographer choosing a kit lens based on MTF graphs.

I remember writing to you, Mike, that I had been convinced 'six megapixels is enough' upon recieving your excellent print of "Wisconsin #7". Now try and fit that sentiment into a coherent relationship with the following facts: I bought the A900 over the D700 on the grounds of nominal resolution, my best digital images come from a 10MP camera, I have most fun with a 3MP camera, the 'best' digital camera I've used had 6MP, I would love to try modern medium format digital one day, and I don't have the cofidence in my own work to make big prints.

Does not compute.

Robin.

PS - Congrats on the Pentax Sponsorship - a sure fillip.

[adrien]

Yes, what the world needs is a good 6MP SLR.

All other things being equal, the trade off would be less pixels per image for far, far, far *better* ones. Quality vs Quantity.

Low noise, high ISO and incredibly sharp.

Think about what the technology that goes into today's 20MP+ sensors is capable of in terms of sensitivity and color accuracy, and then give each photosite 4x more surface area. (!)

Compare the per-pixel quality of a Canon D60 (6MP) from 8 years ago to a Canon 5DmkII (21MP) from last year - then extrapolate back down.

ISO, through the roof, and *clean*.

With a low MP count, lenses are a lot more forgiving and thus more versatile (you can stop down more before diffraction becomes a problem, you notice less scheisse when pixel peeping, etc.)

You save a lot more space on your memory cards and on your computer, so you'd be able to shoot a lot more images without hesitation, and thus probably shoot more better images than if you worried about wasting 22Mp of space before pressing that shutter.

Given the same buffer size, burst rates would be better and longer.

Batteries would probably last longer due to a 75% reduction in processor load.

Dynamic range and color accuracy, boom.

etc. etc. etc.

I think we'd all be *amazed* at what a 2010 tech 6MP sensor would give us.

- - - - - - - - -
veering off topic into the megapixel discussion:

Looking at the (admittedly special) Foveon sensors (Sigma DP1), the actual (RGB) Megapixel count was 4.5MP, but those pixels were so freakin' good that they get interpolated up to 14MP image and *still* look good, perhaps better than what you'd get from a 14MP bayer sensor maybe, even, depending, apples, oranges.

As a anecdote/note (anecdenote?), I have seen 1.5 meter wide prints made from 6MP jpeg files from a Canon D60 (Iraq, 2003) hanging on a wall next to 1.5 meter wide prints made (on the same printer same paper same day) from a 21Mp MKii (Somalia, 2009) and I can really see the difference at about 50cm, a distance at which I can't see the whole image. Standing back far enough to see each/both images on the wall, I can only sense the difference because I know it's there.
(http://www.boring.ch/matt/)

[Mike Bailey]

Is it possible that after a certain point, despite the moving bell curve, why more megapixels are desired is because of the belief it will improve one's photography? And that this belief might be stronger in one less experienced?

Belief = hype = unreality, and some other math.

Mike

[Tim Parkin]

I just want a sensor to match what my lenses are capable of.. is that too much to ask?

[John Robison]

Feature wise what would have me 'upgrade' to a newer DSLR is not more MP but an articulated LCD. My little E-410 with it's (now paltry) 10 MP is more that enough for any print size I'd do but the ability to frame from ground level, overhead or sideways would be truly useful to me.

On another note; is it just the old timers like me who miss the Marvelous-Metal-Manual-Mechanical cameras of 30~40 years ago? On my daily walk occasionally a 18 to 20 something carrying one of those huge CaNikon rigs will stop to chat when they see a svelte OM-1 around my neck. You should see their eyes pop when they look through the viewfinder, a real Alice in Wonderland moment. Sometimes they quickly pull the camera away and look at the outside again, suspicious there must some trickery going on.

[Mike Johnston]

"I think we'd all be *amazed* at what a 2010 tech 6MP sensor would give us."

Adrien,
I do agree it would be interesting to find out.

Not holding my breath, of course.

Mike

[Paulo Rodrigues]

I was going to go for 12mp because I'm pretty happy with that level of quality and doubling the pixels will slow down my workflow, but having said that I would love to do huge prints. so I thought long and hard and decided that seeing as you will be giving me 5 of these perfect cameras I could sell three of them and use the money to upgrade my computer system and storage. The obvious choice then was to go for the highest megapixel count.

[Carl Mosher]

The longer I follow these conversations about the right number of megapixels, the more I come to an inescapable conclusion:

I do not understand what factors make up the image quality I desire.

[Matt Needham]

The new camera I want is an affordable Canon or Nikon, APC-S format, compact camera with about 6mp. I'm very happy with my old, obsolete DSLRs.

My New Year's resolution for 2009 was to never spend more than $2000 on a small format DSLR body ever again (unless I win the lottery). I can't wait until I can afford an old, obsolete D3x. It's looking like my next DSLR is going to be another 5D mk 1. Unless Canon brings something new out to drive 5D mk 2 prices down.

[Craig]

I'm pretty firmly in the "more pixels is better" camp, but not for the obvious reason. Comparing my 1Ds II and 1Ds III, I don't really see a significant improvement in 16 X 20 inch prints. For me, there are enough picture points.

But, if the engineers could pack more high-quality pixels on the sensor, then maybe we could explore new ways to use them. For instance, arrange the pixels in a pseudo-random arrangement and eliminate the anti-alias filter. And, while we're at it, arrange the groups with different ISOs to increase dynamic range. Instead of just three photosites to make a pixel, maybe have 6, 9 or 12 photosites, with red, green and blue photosites covering a plus and minus 2 stop (ideally user selectable) range of ISOs.

Also, look at Phase One's use of bundling photosites to reconfigure the camera for higher ISO, lower resolution shooting when needed. When you have more pixels, you can create alternative uses to fit a variety of needs.

If you could put the equivliant of 50MP or 100MP on a sensor, just imagine what you could do with them to improve the overall image quality.

[petavoxel]

Ctein,

"Dear Petavoxel,
You need to read this column:

"Why 80 Megapixels Just Won't Be Enough..."

http://tinyurl.com/c4dj5n "

I know that column well (and have just re-read it). I certainly respect what you do to test the limits of achievable detail.

I would love to read your analysis of how the limits from aberrations and diffraction scale up with respect to sensor area.

On the broader point, it's probably most useful to reply to your article in terms of "spatial frequency response" (by analogy to audio frequency response).

You are saying we still have a ways to go before we reach the highest perceivable spatial frequencies. I don't disagree.

But the "midrange and low treble" is what grabs us first. I suspect that once those reach sufficient quality to enjoy the tune, it may be enough for many.

[David Dyer-Bennet]

Robin: and I don't have the cofidence in my own work to make big prints

How much does cost factor in? Do you have a Costco in the area? One of the ones with an Epson 9800 printer? That will make you a 24x36 print for $15? Just make sure to tell them "no adjustments". Actually I think you can order on the web and they'll mail it from some store, even if there's none in your area.

I never expected my digital work to hold up at that kind of size nearly as well as it did.

[Michael Bernstein]

I cast my vote in the 'more is better' category, not because I expect to be making use of 60MP all that often, but because I've always envied the quality that 6x7 film was capable of, but never found a 6x7 camera quite to my liking (big, heavy or rangfinders...).

Regarding what a good APS-C 6MP DSLR would be capable of (image-wise) today, I think the Nikon D3s in DX crop offers a fairly good approximation. Sadly I can't see anybody following this path. As Robert N. pointed out upthread, Nikon is getting plenty of flack as is for releasing the D3s with 'only' 12MP...

[Dave Cattell]

I was one of the few that voted for Max Pixels, and it's because I want them for landscapes.

In my film days (well actually, personally I'm not sure I have quit, yet ;-)) I managed all the way up to 4x5. I have some 13x19 ink jet prints from Imacon scans of 4x5 negs and I can't wait to be able to afford a digital camera that has that level of detail.

The subtlety of tonality still seems to be missing from the digital I have used so far, and I'm sure that is related to pixel counts as well as pixel accuracy.

So my plan with that generous offer, was to sell all but two of the highest megapixel versions on offer and with the money get the fastest machines and biggest printers I can fit in my apartment. Then any left over would go on an M9 and set of 3 lenses.

Dave.

[CMS]

We already know that often there is a difference between what people need and what they want. Not taking money into account there can also be a difference between what people want and what they buy.

The number of megapixels is only one of the characteristics of a camera body. Moving up on the MP scale might not always be a priority of an upgrader. Here is an example: The owner of a Canon 50D who is of the opinion his camera has already enough or even too many pixels may well choose to switch to a 7D in order to get a superior metering system and better AF. But in the statistics he would help fatten the 18 MP buyer category and send wrong signals to the marketing people.

[Peter]

"I think we'd all be *amazed* at what a 2010 tech 6MP sensor would give us."

Sorry to be cheeky, but we already have a pretty good idea what that would be like: take a Sony A900/850 or a Nikon D3X and then downsize the picture by 50% in each dimension. There's your pixel-binned 6MP sensor. If that's too much work for you, I'm sure you can get some clever Photoshop/Lightroom guy to setup an action/macro/batch command/whatever to do that for you automatically.

[Peter]

"And, while we're at it, arrange the groups with different ISOs to increase dynamic range."

Fuji's S5 was pretty impressive in that regard.

[Peter]

"My suspicion is that there would only be a small gain in resolution of a monochrome sensor over a color one, and a small gain in efficiency (lower noise) from not needing the color filter array. It would be interesting to know those answers for sure."

See here: http://www.luminous-landscape.com/reviews/cameras/achromatic.shtml

One thing to note that is assumed by the article is that the monochrome sensor is 39MP while the SOTA color sensor used in the first comparison is 60MP. Also the physical size of the 39MP sensor is slightly smaller than the physical size of the 60MP chip, although the difference is in no way proportional to the approximate 50% difference in pixel counts.

[CMS]

I suspect VERY few photographers know how many megapixels they actually need.

Theoretical discussions won't help. Too many variables stand in the way. For a given amount of MP some cameras perform better than others. A camera will perform differently depending on the ISO dialed in. Shooting techniques differ from one photographer to another. Post processing skills differ even more. The printers' output varies from brand to brand, not to mention the effects of ink/paper combinations. The viewers' quality expectations are not always the same, with most of them not even able to formulate what their expectation is. An almost infinite number of factors could be mentioned here.

A minimum of methodology is necessary.

Statements concerning the 'right' number of pixels is would first require to tell how one defines the desired output, e.g. A3 prints to be viewed from five feet, or A4 prints to be viewed at normal reading distance, or anything else one likes. Then one would use a number of cameras covering a range of megapixels and print to the best of one's abilities, and compare the results in order to find the point above which additional pixels do not add to the visual quality of the print. Finally, one would publish this with as much explanation about the method used as possible and reach a conclusion that would still be very personal, but certainly more educated and satisfying than most we can read concerning this topic.

[charlie]

I must confess, I responded to the poll before reading the entire post. I responded as though you asked how many MP my camera had.

I must also confess, I own a D700 and was uncertain how many MP it actually had. I looked it up online. ;-)

[Dave Reichert]

I voted for high megapixels, but on second thought I'm not sure a new camera is the answer. I'm thinking that I might be happier with a Photoshop plug-in that works like the "enhance" feature I see in all those CSI programs on TV. You know, the feature that pulls infinite detail out of a still frame from a video security camera in a bodega...

Couple that with one of the lenses that David Hemmings used in "Blow Up" and I'd be in business.

[Eolake Stobblehouse]

'Polls like this one basically represent people saying "What I've got right now is perfect for me...right now."'

Damn, I was beginning to suspect that.

British mag Amateur Photographer had an interview with an old lady photographer, who still use one of those 6x8cm bellows cameras with the little mirror cube you look down into as a viewfinder. What are they, from the thirties? She actually said, and I think with a straight face: "I don't trust these newfangled things".

[adrien]

"Sorry to be cheeky, but we already have a pretty good idea what that would be like: take a Sony A900/850 or a Nikon D3X and then downsize the picture by 50% in each dimension. There's your pixel-binned 6MP sensor. If that's too much work for you, I'm sure you can get some clever Photoshop/Lightroom guy to setup an action/macro/batch command/whatever to do that for you automatically."

Peter,

Your argument/point is valid, and you may be entirely right. Or not. I'm not a DSP nerd by any means, but the it could equally *not* be the case...

In much the same way as a basketball team of 10 midgets wouldn't necessarily tie against a regular team of 5 giants, despite having the same total amount of height on each team.

There's some cheeky right back at'ya :)

[Forrest]

The 13 million pixels in my 5D have been more than enough for the several years I've had the camera. Almost always, the softness in my images, when there is any, is due to other factors ( exposure time, movement, optics ). It's only certain, highly detailed landscapes, when printed as large as possible, which could benefit from more pixels.

On the other hand, 3200 ASA is nice at f/2, but this has been limiting.

[Nicholas Fiduccia]

I responded >56MP and assumed that I can have a square format sensor of at least 55mm squared. If we use the "ideal" 6MP APS-C sensor and just expand the sensor size using the same pixel density, we will want a 44MP "medium-format" sized sensor.

[ObiJohn]

I voted for the 10 MP range, because I want a 1.4x crop factor Foveon with 8 MP. That would be nirvana for me....

[Stephen Scharf]

This is a great topic and the results show that TOP's readership really gets it. I've stated here before that the full-frame 12 megapixel sensor as exemplified by the original Canon 5D really hit the sweet spot for image quality and resolution, and my opinion in the intervening time hasn't changed. Unfortunately, the mfrs still don't seem to get it. A classic example is the Canon 7D, a really superb body hamstrung by putting an 18 megapixels on a 1.6X crop factor sensor. While the operation, body, focusing, and viewfinder of this camera is superb, the noise performance really lets it down. If only Canon had put the 5D MkII in the body of the 7D, and kept the sensor size at 12 megapixels, they would have had a real winner on their hands.