Do Megapixels Tell the Whole Story?

14 Megapixel SLR vs 14 Megapixel Digicam
Image of two 14 megapixel cameras: slr vs digicam Two 14 megapixel Sony digital cameras: A550 SLR and W320 digicam. Read on to discover the real story behind a camera's true resolution.

The answer is a resounding no, the megapixel rating on a digital camera is a poor measure of a cameras ability to resolve detail and therefore is a poor spec to compare. Unfortunately most consumers don't know this, no surprise: this is not how cameras are marketed to the public. That's why we calculate true resolution so you can understand how many megapixels a camera really has. True resolution provides an estimate of the maximum resolution a camera can resolve based on average shooting conditions and some fancy physics.

Digital Cameras with High True Resolution

These digital cameras excel against all others in their ability to resolve detail. If resolution is important to you should consider these but should also read our study: How much resolution do I really need to fully understand how much resolution is necessary.

Sony Cyber-shot DSC-TX30 The Sony DSC-TX30 is one of the highest resolution ultra compacts.
Sony Cyber-shot DSC-RX100 III The Sony DSC-RX100 III is one of the highest resolution pro digicams.
Alpha A6000 - 24.3 MP
Sony Alpha A6000 The Sony A6000 is one of the highest resolution mirrorless interchangeable-lenss.
K-3 - 24.1 MP
Pentax K-3 The Pentax K-3 is one of the highest resolution entry-level DSLRs.
D810 - 36.2 MP
Nikon D810 The Nikon D810 is one of the highest resolution pro DSLRs.

The Science Behind True Resolution

Airy Disks Determine Resolution

Because of the way light works (it's actually a wave) it can only be focused to so small a point. The size of that point dictates that there is a maximum resolution that can be projected on a given surface area.

Airy Disk Example
Image of four airy disks using red gren blue and white light. Red, green, blue and white Airy Disks are captured at f/20 using 4.65μ pixels. Credit: JMZawodny via Flickr.

Sensors on cameras range drastically in size, for example a 864mm² full frame sensor has 30x the surface area of a 28.5mm² digicam sensor. That means the digicam's pixels are going to take up a lot less space on the sensor. In fact with most digicams the pixels are so small that they are essentially wasted - the smallest point that the light can be focused being many times larger than the pixels.

The smallest points are called Airy Disks, it's difficult to see them in an image because you're focusing an entire scene so it just ends up looking like blur, here's a example of Airy Disks @ f/20 with 4.65 micron square pixels used to capture them.

You can see the Airy Disks because only 4 light sources are being focused onto the sensor. The dots show the maximum physical resolution and demonstrate how the pixels are basically wasted. The size of the Airy Disks depends on the focal length - the higher the focal length the bigger the Airy Disks and at f/20 those are some big disks - the 4.65 micron pixels are just a little smaller than you'd see in a full frame sensor SLR. There is no way to make those green, red and blue dots the size of a single pixel, in fact in this example 25 pixels are being used to capture the brightest spot in the Airy Disk ( the puddle effect is the result of lights wave nature. This example is using fairly large pixels (4.56 micron), many digicams have pixels around 1.5 microns or 1/3 smaller can, that means a digicam (if it could shoot at f/20) would be using over 200 pixels to capture a single Airy Disk.


How Many Pixels per Airy Disk?

Pixels per Airy Disk
Image of pixels being used to fully cover an airy diskMathematical relationship between an airy disk and the number of pixels required.

It can be argued that a ratio of 1 pixel to Airy Disk is insufficient to capture the bleeding and gradients, which do represent some level of extra information. We feel that 4 pixels per Airy Disk is more than sufficient to capture that detail and permit some extra sharpening either by the camera or post process. The illustration indicates that the hypotenuse of each pixel will be 1/2 the diameter of the airy disk. For calculating true resolution we use an f-stop of 5.6 which is happy medium and fairly typical. The true resolution represents the maximum you can expect to achieve with the camera you're using. That being said achievable resolution will be less given imperfect lenses, jpeg compression, noise reduction routines, imperfect operators, etc.

Cameras with the Biggest Pixels

The following cameras from the last two years have the biggest pixels, that means they are the least likely to be wasting pixels.

Sony Cyber-shot DSC-TX30 The Sony DSC-TX30 is one of the ultra compacts with the largest pixels.
Sony Cyber-shot DSC-RX100 III The Sony DSC-RX100 III is one of the pro digicams with the largest pixels.
Alpha 7S - 69.5 µm²
Sony Alpha 7S The Sony 7S is one of the mirrorless interchangeable-lenss with the largest pixels.
K-5 II - 23.1 µm²
Pentax K-5 II The Pentax K-5 II is one of the entry-level DSLRs with the largest pixels.
D4s - 53.5 µm²
Nikon D4s The Nikon D4s is one of the pro DSLRs with the largest pixels.

Real World Example

Image for Comparison
Sample image shot on a 5d Mark II Sample photo shot on an EOS 5d MkII

We've analyzed a 21 megapixel photo taken with a Canon 5d MkII - full frame SLR and used this to illustrate the difference between a full frame and typical digicam sensor. We calculated the true resolution of a 6.2mm x 4.6mm sensor @ f/4 and used that to illustrate how big the airy disks are relative to the true resolution. The true resolution of the full frame camera is the same as the rated megapixels, but with pixels bigger than the airy disks. Conversely, the digicam's tiny pixels are so small that it takes about 8.5 pixels to make up an airy disk - a complete waste of those little itsy bitsy pixels.

Image Crops

28.5mm² Digicam Sensor Crop
Image of digicam sensor crop 100% crop of the 14MP 6.2mm x 4.6mm digicam sensor, resized to the same size as the full frame crop.
864mm² Full Frame Sensor Crop
Image of digicam sensor crop 100% crop of the 21MP 36mm x 24mm SLR sensor.

You can immediately see the difference between the two sensors that belies the rated resolution. At an f-stop of 4 the 14MP digicam has a true resolution of 7.8MP vs 21MP for the full frame SLR (which is limited by the lens and operator at such high resolutions not the airy disks or sensor).

Digicam Airy Disks

We're showing a 2nd crop within the crop above represented by the black square. The first thing to notice is how much total sensor area is used to capture the detail - about 46 microns across, hardly any at all, the camera is trying to focus the scene on too small an area and the limitations of the airy discs are presenting themselves. You can see that a digicam sensor packed with 14 megapixels is simply capturing detail that just is not there, a suitable scaling algorithm on much larger pixels would accomplish the same.

Airy Discs with a 28.5mm² Digicam Sensor
digicam image crop showing pixel size vs airy disc size

Full Frame SLR Airy Disks

Relative to the digicam the SLR is attempting to focus this portion of the scene on 212 microns width of sensor (almost 5x the width), because of this the airy disks are much smaller relative to the sensor area. Put another way the full frame has such a large sensor that it has relatively ginormous pixels, actually larger than the airy disks - no waste. You can see that a full frame sensor packed with 21 megapixels is a very reasonable proposition. The benefit of focusing the scene on a larger area means more detail is being capture.

Airy Discs with a 864mm² Full Frame SLR Sensor
full frame image crop showing pixel size vs airy disc size

Conclusions

  • Don't blindly trust the megapixel rating on a camera - use Snapsort's true resolution rating to get an understanding of the real world capabilities
  • Read more about resolution to understand why just a few megapixels could be enough
  • Learn more about other features in a camera - there's a whole lot more to cameras than their mega pixels

Related Info

Discussion

Showing 25 comments

wow (2:36 AM, June 06, 2014)
wow, didn't even start to think about the math and instead insulted the other.
 
Avatar for Søren Mogensen Søren Mogensen (9:05 AM, June 02, 2014)
How is it possible to put 12 MP into a 1/2.5 sensor when it is not posible to put more than 36 MP into a full-frame sensor 30 times its size?
 
Pete (10:47 AM, May 24, 2014)
Check out techradar's resolution chart images for the Canon EOS M, which
show there is no discernible difference between ISO 100 and 200 -

http://www.techradar.com/reviews/cameras-and-camcorders/cameras/digital-slrs-hybrids/canon-eos-m-1089580/review/4

According to the Snapsort article the 18MP Canon should be be night-and-day sharper, by a factor of 2x to the point-and-shoot sensor (they estimate a maximum 'true resolution' of around 9MP). Refer to their mock up 'image comparison' of tree blossom pictures to see how totally obvious the difference should be. Real world testing shows no such difference to exist.
 
Avatar for GunnyNinja GunnyNinja (1:43 AM, May 23, 2014)
I wonder if it's still hard to sit down after that
 
Pete (9:54 AM, May 21, 2014)
Hi, Tony. For me, I'm satisfied that the Canon would be barely perceptibly different at 100 compared to 200 (check out comparisons between those two ISOs elsewhere online if you want), so I felt happy to make this post. My second head to head with a Sony NEX5N (above) were both at base ISO BTW. What do you reckon, as regards resolution and sensor size?
Best,
Pete
 
Avatar for tony tony (11:37 PM, May 08, 2014)
Yes for a real comparison you'd need the canon to be at ISO 100 too at the very least. The bottom one is the canon M?
 
Pete (7:49 PM, May 05, 2014)
Here's another 16MP resolution comparison to throw doubt on the airey fairey snapsort theory. WX60 P&S rough and ready comparison to a NEX 5N (APS size sensor) fitted with19mm Sigma lens (at f5.6) both at base ISO. Actual pixels centre crop with no interpolation or sharpening etc.
 
Ekmund Yong (9:16 PM, April 23, 2014)
Yes I do agree vocabulary is important. I suggest you read another article then. Perhaps he should have posted this article in his native language because you clearly have no interest in what he is trying to say other than judging his grammar.
 
Avatar for Pete Pete (5:08 PM, March 23, 2014)
I thought I'd contribute a head to head between a 16 Mpx Sony WX60 point & shoot with 1/2.3 sensor (and Zeiss lens) and a 18 Mpx Canon EOS M with APS sensor (fitted with the decent 18-55 kit zoom set to 20mm and f5.6), mirrorless but comparable to an entry-mid level DSLR in IQ. The Sony is near it's wide end, too - and at the only aperture it offers at that focal length! This is the centre portion of a wide field of view with no resizing of the shot at 100 per cent actual pixels, of OOC jpegs and with no post sharpening (apart from the PS default applied at the time of 'save for web'), just a tweak of levels and desaturation as the different white balance was distracting. (I think the Canon was set to around medium in-camera sharpening whereas the Sony can't be adjusted.) I'm aware it's not full-on scientific (I didn't own the cameras at the same time which doesn't help) so there are some small variations, for instance the Sony was at ISO 100 and the Canon at 200 which is not ideal, but I think it's a credible comparison and shows what centre resolution is actually possible - although the corners are another story, and the EOS M is way ahead. The snapsort so-called 'real world example' above is illustrative and not a genuine comparison between two cameras with different sized sensors, it simulates predicted poor resolution in a small sensor high megapixel camera.
 
Avatar for EnKiu EnKiu (3:28 AM, March 21, 2014)
This explanation makes no sense. In the second example for the full frame sensor the airy discs are smaller than the pixels meaning that more than one airy disc will be combined into a single pixel. This will lose information. The reason the larger sensor produces a better image is because the same area of the image is focused over MORE PIXELS. From your explanation at least, the full frame sensor should have about 4 more times pixels for optimal resolution of details, i.e. about 100 MP :) LOL
 
Avatar for Gordon L. Scott Gordon L. Scott (10:59 PM, August 24, 2013)
The size of the 'airy disk' is very much dependent upon the quality of the lens used to produce the image. Your examples appear to be based on some rather inferior lenses! While el cheapo cameras likely meet your presumptions, a high quality camera lens most certainly will outperform your examples. That said, signal noise ratio (SNR) will improve with larger photosensor sites and as a result, so will apparent resolution. The bottom line: you usually get higher quality when you pay more for your equipment!
 
Avatar for krvlucky krvlucky (5:23 PM, July 24, 2013)
i ask a simpler question if u need a small size and shoot at lower resolution and higher resolution ...... what will be the difference in the pic quality ?
 
Avatar for krvlucky krvlucky (5:19 PM, July 24, 2013)
is there any gain when we shoot with a camera of resolution 16 mpx at setting of 10 mpx when sensor resolution is say 9.7 or 10.6 ?
 
Avatar for krvlucky krvlucky (5:14 PM, July 24, 2013)
sir with new camera panasonic fz70 i have asked similar question ....... can shooting at lower mpx can we get better pics than say canon powershot sx50 hs ...... or we try to zoom full at 60 or 50 with canon at lower mpx to get great shots with these cameras
 
Avatar for Youlu Zheng Youlu Zheng (8:46 PM, June 11, 2013)
Dear Billy, You do have advantage with your native tongue (or just English name). My writing is not perfect, and my last name is even worse, but I am a proud Chinese American, and my book was published by Oxford University Press. (You can find my name with the publisher if you're smart enough, and I do not try to hide behind an alias.)

Still, I am unsure how my writing offended you.
 
Shamik (7:25 PM, June 11, 2013)
I cannot actually decide which one to buy, Nikon D3200 or D5100? Both are almost having same or equal features except that D3200 has more megapixel. Please suggest which one would provide me with better photography experience and image quality.
 
Mamey (9:29 PM, June 08, 2013)
Billy, by any chance your actual name is "Hill"-Billy? The guy wasn´t born in an English-speaking Country, but who cares? May be we could get a couple of good ideas from Zheng. Just get a life, man, and stop dating cousin Mary-Lou, otherwise the kids will have "lazy" eyes, hehe.
 
Avatar for Mike Svitek Mike Svitek (7:29 PM, January 18, 2013)
I think the "uses" was a typo. Other than that, I didn't catch any serious mistakes in his/her argument... And I read it all the way through.
Maybe you should just stop being a Grammar Nazi and begin paying attention to the valid points in the argument, not the three words that aren't in the proper formatting.


Cheers!
 
Avatar for Mike Svitek Mike Svitek (7:24 PM, January 18, 2013)
Its probably too late, but a higher number of pixels is better if you have enough light to expose them all properly. Once you need higher ISO performance and better Dynamic Range, you should start looking at dropping resolution for the benefit of an increased pixel size. I have a Pentax 645D with its huge pixels and a 40Mpix resolution but, in good light, it will be about on-par with my Nokia 808 PureView with its 38Mpix sensor...
That being said, the dynamic range I can achieve with the 645D is equivalent to the Nokia taking a perfect HDR image of about 5 photos...

You have to pick what's most important to you, then pick the camera's specifications after.
 
Avatar for Mike Svitek Mike Svitek (7:19 PM, January 18, 2013)
Lenses were rated in lines per mm (or inch) in the days of film... Those were typically rated for up to 5000 lines/mm which will out-resolve any modern DSLR sensor (You'd need to cram over 4 Gigapixels into a full-frame sensor to out-resolve that)... I don't believe modern lenses have decreased their quality since my grandpa's 70 year old Pentax glass was made, so you might as well note even mention the possibility of any sensor out-resolving a camera lens...

(Leica Summicron 35mm f/2 lens is the king of resolution, rated at 7500 lines/mm... That will resolve 6,480Mpix)
 
21 limited. (2:22 AM, November 10, 2012)
The pixel size of K-30 is about 4.8 mu, and in terms of resolution, it's hard to compare because you use different lenses on different systems. but the Pentax K-5 IIs is probably the highest resolving APS-c system and is on par with the 24 MP D600. If you blow up a K5 IIs image it will be pretty much identical with a D600, or 5DIII. But the old thing about pixel density and resolution, and even sensor size and resolution isn't holding up as far as I can tell. You can't tell anything looking at numbers, you have to test images.
 
billy (6:39 PM, September 15, 2012)
@ YL Zheng,
It is so funny that there is always, in every forum, a commenter that uses dizzying mathematics to show off their aptitude! Yet despite your amazing math ability, your writing composition stinks! I could send you a long list of questions regarding your sentence structure and vocabulary. I know for sure the first thing you will say is, "English is not my first language" but really no one cares! If you want to be taken seriously put the calculator down and learn basic grammar! Nobody is impressed with a guy who can calculate the world but can't articulate for crap! Math and science is real handy but vocabulary and communication are equally important!
 
Geos (5:35 AM, March 22, 2012)
I'm sure this is basic knowledge to most, but I'm a beginner. Does a high number or a low number represent the better true resolution? What true resolution number should I consider "good" true resolution?
 
Makjabari (1:29 PM, March 13, 2012)
what this mean, is the slr sony better than aps-c camera of nikon
for example
sony slt-a65 vs nikon dslr-d7000
 
YL Zheng (9:33 PM, March 11, 2012)


Thank you for the great article, which is a good attempt to
explain and simplify the theory about resolution.  A few questions here:


1.      
In the last figure under "Full Frame SLR
Airy Disks", you uses a 36mm x 24mm 21 MP full frame sensor for
illustration.   Assume a 1.5:1.0 aspect
ratio, the theoretical resolution is 5612x3742. 
That is close enough to my 21.3 MP Canon 5DII of 5616x3744.  So the diameter of a pixel should be 6.14
micron.  However, in this figure with 39
pixels counted per row with the width 212 micron, each pixel is only 5.44
micron as you displayed.  How does the
difference happen?


2.      
In the same figure,  you say this full frame sensor is
"almost 5 x the width" (of the 6.2mm x 4.6mm 14 MP sensor), which is
apparently far away from the right number of 36/6.2 = 5.81.  Should it be corrected as "almost 6 x
the width"?


3.      
We all know the size of Airy Disk varies and
depends on lens focal length and aperture size.    With the factor of lens imperfectness
taking into consideration, Canon and Nikon, among other brands, decide to use
pixels numbers ranging from 12.1 MP, 21.1 MP and 36.2 MP for their full frame sensor
cameras Nikon D700, Canon5DII, and Nikon D800 respectively with different
philosophy and marketing strategy.   But
the real question practical to photographers is how to use the setting to
maximize the resulting image quality.    We don't know how the sensor is used when a
photographer sets the camera resolution to less than the highest setting.   The question is, for example for the Nikon
D800, when the camera is set to, say, 4800x3200 instead of the highest 7360x4912,
how does the image quality in terms of "true resolution" compare to,
assuming a perfect lens,  


o  
The highest 7360x4912 image cropped to similar
size with longer focal length lens


o  
3000 lines/inch as the maximal available resolution
in real life measurement used by many standard tests.


o  
Nikon D700 at the maximal resolution of
4256x2832 with the 12.1 MP full frame sensor.


If the answer is "equivalent", or
at least "close enough", the Nikon D800 with 36.3 MP has the
advantage of providing more choices for the photographers, at least.


4.     For the same situation in the previous
question, what is the effect of lower resolution setting on low light
noise. 


5.     How are different ISO settings implemented
with a full frame sensor camera?


 


Thanks in Advance.