# How many pixels in a megapixel?

I am trying to find out the definition of a megapixel? Some references on the web place it at 1 million pixels, and other places say it is equal to 2^20 = 1,048,576 pixels.

• Just to satisfy normal human curiousity? Or is there something where it matters in which way megapixels are counted? Jun 27 '13 at 15:47
• "Enough and more than enough for poor Catullus" Nov 17 '14 at 13:23

I think that in general due to rounding — and more importantly, other real world factors which mean that megapixels only relate loosely to actual resolving power — it doesn't really matter if "megapixels" is binary or decimal. It is a useful term because it happens to be in the range where we get human-useful small numbers with the digital cameras (so far). It's rarely used to mean a precise value — one 16-megapixel camera will likely generate photos with a slightly different size than one from another brand.

For the same basic reason, "kilopixel" isn't a real word, because there's no particular case where it would be useful.

Overall, a lot of us coming to photography from a tech background, be it programmer, engineer, or otherwise, have a tendency to look for precision. When it comes to exposure, anything under a third of a stop is unlikely to be a big deal, and when it comes to pixels, a similar basic rule makes sense: until we're talking about doubling or halving the number, don't sweat it.

I originally posted this as a comment to another question, but I think it answers this one.

• My thermoimager is 20kPixels, and Basler racer series are 2-16 kPixel . That is as useful as MPixels. Or in Baslers case it is more useful than the usual MP measure as they dont mix up 2 dimensions in an unrecoverable manner. Jun 26 '13 at 21:24
• I stand corrected. However, it's not a term in general photography. Jun 26 '13 at 22:35
• and the threshold where it matters as much as 1 Mp whether you use x1000 or x1024 as a base is at 20.5MP. So a 21MP camera would be a 20MP camera in base 1024. No way the manufacturers would advertise it as a 20MP camera, though, so I bet they agree, haha. Not that it matters if it really has a few more or less pixels though. a factor 2-4 is what really matters Jun 26 '13 at 23:31
• Right, exactly. At 20.5MP, even though the difference in rounding is One Million, that amount isn't so important that we care about the details. I completely agree about it only really mattering when you're getting to significant factors. Above 20 we should probably round to the nearest 5, and once cameras are routinely in the 40+ mpix range, rounding to the nearest 10 would be reasonable. Jun 27 '13 at 2:54
• Which of the (currently) 8 other answers is "the" other answer? Mar 29 '18 at 11:41

A megapixel is defined as 1 million pixels, not 2^20.

• what he said / points at MikeW Jun 26 '13 at 19:55
• True but not the whole story. A sensor with 1,048,576 pixels would also be a "1 megapixel" sensor. The precise difference isn't important. Jun 26 '13 at 20:46
• So if you round a from 9.7 to 10 it doesn't really matter if you are actually rounding to 10 or 10.2 but just choosing to write 10? Jun 26 '13 at 21:19
• @mattdm If you really want to see the most ridiculous extension of the idea that cameras are precise to the 9th decimal place (in terms of shutter times expressed in decimal values of one second), check this out: scantips.com/lights/fstop2.html Mar 29 '18 at 11:52

It depends how you count but almost every company multiplies the number of photosites and divide by one million. They rarely make the distinction if those photosites are next to each other or layered. For this reason, a 45 MP Sigma SD1 makes an image which has the same resolution as a 15 MP Canon 50D.

They sometimes quote two numbers, effective megapixels and actual. Effective are the ones that make into final maximum resolution images and which may be a little less than the actual ones which are how many are on the sensor. Some of these may be masked out to read the back levels and others lost because of the imaging area of the lens.

In computing, when talking about kilobytes, and megabytes, the terms kilo and mega have traditionally been modified, letting kilo=2^10, and mega=2^20.

This has led to confusion, because hard disk manufacturers would use megabyte to indicate 1 million bytes instead of 2^20 (resulting in more impressive numbers).

This has led to the definition of two new terms, Kibibyte and Mebibyte, meaning 2^10 and 2^20.

But when talking about something different than bytes, kilo and mega should still refer to their original meanings, one thousand, and one million.

Thus a megapixel should be 1 million pixels. But this can often be an approximation. E.g. my 18 megapixel Canon EOS 7D 'only' has 17.9 million pixels.

• The corruption of the terms "kilo" and "mega" made sense for RAM and other electronics that were forced by their physical structure to be a power of 2 in size. As sizes increase and the difference between the powers of 2 and powers of 10 increase, this becomes more confusing. I consider it an unfortunate mistake of history that powers of 2 were ever used. Dec 2 '19 at 17:36

To answer you have to understand what a `pixel` is.

In digital imaging, a pixel, or pel (picture element) is a physical point in a raster image, or the smallest addressable element in a display device.

So `Mega` being a unit prefix, it simply means `1'000'000`. Knowing that, 12 Megapixels means simply 12'000'000.

This said, when your camera constructor displays 'Around 14.3 Megapixels' in the data sheet, it's a simplification to avoid writting stuff like : 14,204,928 `pixels`.

This value being calculated from the resolution of the pictures you're taking : 4352 x 3264 `pixels` = 14,204,928 `pixels`.

It depends on whether you're selling or buying. When you're writing marketing literature, you want the Mpix number to be as high as possible. That means you use 106 for "mega". When it's to your advantage to make the number look small, you use 220, which is 1,048,576.

In reality, a 5% difference in the total number of pixels is pretty much irrelevant. Note that the linear resolution goes with the square root of the total number of pixels, so 5% more pixels is only 2.5% more linearl resolution. You won't be able to notice that difference even in two prints at the proper size you get to compare side by side.

I would like to also say that each "pixel" in a DSLR is actually only a portion of a pixel. So, the sensor itself has say, sensitive light elements for an R and a G and a B, and maybe another G. Now, these three, or four, together should form a single pixel but it don't. They interpolate it and make the four, count as four pixels.

Or something similar (http://en.wikipedia.org/wiki/Bayer_filter)

This means that your 20 megapixel camera might actually be a real 5 megapixel camera but it's interpolated up using algorithmic magic.

Same applies to the LCD screen. A "1 million dot" screen only has 300kish pixels. Sadly.

• This underestimates the quality of those interpolation algorithms. If it were simply a matter of 4:1, the practice would probably have stopped as sensor density increased and file size correspondingly went up. But in fact, the interpolation really does contribute to extra resolution. It's not as good as 1:1, but it's also not 4:1. Jun 27 '13 at 23:05
• As far as I understand, it is not that THE SAME 4 pixels are interpolated but EACH PHOTOSITE is used 4 times to calculate the color value of the 4 neighbouring pixels in the final image. (exept the photosites in the border of the sensor). Nov 17 '14 at 15:03
• This is not correct. The pixels (photosites) on the sensor all correspond to a pixel in the image, however on a normal sensor 1/2 are green, 1/4 are red, and 1/4 are blue. The bayer algorithm simply uses each pixels neighbor to determine what the real color was at that photosite. This is entirely different from the interpolation used to make a larger image from a smaller sensor. This is also completely different from the 3 RGB sub-pixels of an LCD screen that make up one pixel. Dec 1 '19 at 13:43
• @mattdm is correct, modern de-Bayering algorithms are much more sophisticated than a simple interpolation. You need only look at a picture of a resolution test target to see. You can't simply overlay the different colors over each other because they aren't aligned. Dec 2 '19 at 17:44
• @MarkRansom you can overlay them if you make 4 single-channel images each from the subset of the photosites with the same filter (or same position in the Bayer tile) and shift them by half a pixel to align (using an interpolation, of course). But this is just one of the approaches to reconstructing RGB colors, a simplistic one. Dec 2 '19 at 21:47

Converting megapixels to pixels is very similar to converting megabytes to bytes.

1000 Bytes are 1 KiloByte (not to be confused with Kibibytes which are 1024-based)
1000 KiloBytes are 1 MegaByte (not to be confused with Mebibytes, which are 1024*1024-based) which is 1 Million Bytes.

So we can simply convert the units like this:

1000 Pixels are 1 KiloPixel (rarely used in practice) 1000 Kilopixels are 1 Megapixel (1 million pixels)

### Here are some examples:

A Camera With 96x128 Pixels is a 0.012MP Camera (or 12kP Camera)
A Camera With 120x160 Pixels is a 0.019MP Camera (19kP)
A Camera With 240x320 Pixels is a 0.07MP Camera (70kP)
A Camera With 320x480 Pixels is a 0.15MP Camera (150kkP)
A Camera With 360x640 Pixels is a 0.23MP Camera
A Camera with 480x640 Pixels is a 0.30MP Camera
A Camera With 480x854 Pixels is a 0.40MP Camera
A Camera With 540x960 Pixels is a 0.51MP Camera
A Camera with 600x1024 Pixels is a 0.61MP Camera
A Camera With 768x1024 Pixels is a 0.78MP Camera
A Camera With 720x1280 Pixels is a 0.92MP Camera (920kP)
A Camera With 960x1280 Pixels is a 1.22MP Camera
A Camera With 900x1600 Pixels is a 1.44MP Camera
A Camera With 1200x1600 Pixels is a 1.92MP Camera
A Camera With 1080x1920 Pixels is a 2.07MP Camera
A Camera With 1440x1920 Pixels is a 2.76MP Camera
A Camera With 1536x2048 Pixels is a 3.14MP Camera
A Camera With 1440x2560 Pixels is a 3.68MP Camera
A Camera With 1800x2400 Pixels is a 4.32MP Camera
A Camera With 1920x2560 Pixels is a 4.91MP Camera
A Camera With 1944x2592 Pixels is a 5.03MP Camera
A Camera With 2048x3072 Pixels is a 6.29MP Camera
A Camera With 2448x3264 Pixels is a 7.99MP Camera
A Camera With 2160x3840 Pixels is an 8.29MP Camera
A Camera With 3072x4096 Pixels is a 12.58MP Camera
A Camera With 2880x5120 Pixels is a 14.74MP Camera
A Camera With 3264x4896 Pixels is a 15.98MP Camera
A Camera With 3600x6400 Pixels is a 23.04MP Camera
A Camera With 4096x6144 Pixels is a 25.16MP Camera
A Camera With 4320x7680 Pixels is a 33.17MP Camera
A Camera With 5720x10240 Pixels is a 58.57MP Camera

It’s easy to calculate. Just multipliy `width[px] * height[px]`, and you’ll get the amount of megapixels.

• -1: 1000 bytes are not one kilobyte in general use. 1 kB is 2 ^ 10 = 1024 bytes. The whole point of this question is the distinction between the two, and this answer ignores that. Nov 17 '14 at 12:28
• I'm voting this down because a) it doesn't add new information beyond other answers and b) the wall of numbers and repeated "Camera A Camera" lines is just unreadable noise. Nov 17 '14 at 21:12
• Sorry i forgot that in the flow of answering about pixels...bytes calculation is the same as the pixels...1024bytes is 1Kilobyte, 1000pixels = 1Kilopixel.....1024Kilobytes is 1Megabyte,1000Kilopixel = 1Mega Pixel...1024Megabytes = 1Gigabyte and etc....... ok? :) ..Sorry for the mistake.... Nov 30 '14 at 11:12