It’s impossible to read about cameras and not encounter the terms “pixels” and “megapixels.” Nowadays most cameras have 20 megapixels (MP) or more, and there are some monstrosities with 100MP. I still have memories of walking into camera shops as a teenager and seeing the higher end models having 2 megapixels, and probably in fifty years everyone will have 600MP. So what is a pixel, and how many do you need?
What are Pixels (and Megapixels)?
A pixel is simply the smallest visual unit making up a digital image. In other words, a digital image is made up of millions of tiny coloured squares, and each of those squares is one pixel.
A camera sensor is also said to have pixels. In this context, a pixel refers to the number of photosites on a sensor. Photosites are the individual sensing areas that capture light, which is then translated into pixels through software.
Pixels are also the units used to describe a camera’s resolution. For example, the Fuji X-T4 produces a 6240 x 4160 image, which means an image 6240 pixels long by 4160 pixels wide. This gives a total of 25,958,400 pixels. Since this is such an unwieldy number, it’s better to use the unit of megapixels. One megapixel is simply a million pixels. So, the X-T4 has a resolution of about 26 megapixels.
Total vs. Effective Pixels
Let’s say your friends just cancelled on coming to your birthday party. What’s the remedy? Reading camera specs on B&H, of course! Just knowing that the Canon R5’s DCI 8K records at up to 1300Mb/s is sure to make you feel like you don’t even need a birthday.
But if you look too closely, there’s something that may disturb your peaceful evening. Almost every camera has two different values its megapixel count: actual (or total) megapixels, and effective megapixels. What’s the difference between these two values?
The important value for photographers is the number of effective megapixels. This is the number of megapixels that will be in your full-size image when you open up your Raw developer or export a JPEG at maximum size.
For example, the Panasonic G9 is listed as having 20.3 effective megapixels. But what about the G9’s “actual megapixel” value of 21.8 megapixels? Can you unlock these hidden pixels for the low price of $329.95 to get even more precious resolution? Sadly not. Instead, these are pixels on the edge of the sensor outside the imaging area. Why are there extra pixels on the edge of the sensor? There are two main reasons.
1. First Reason: The Way Color Sensors Work
The first reason is the nature of color sensors. For example, consider the Bayer sensor present in almost all color digital cameras. It uses separate photosites for red, green, and blue light:
If you shoot Raw (and you should), these are combined by your Raw editor through a process called demosaicing to produce what you see when you open a Raw file. If you shoot JPEG, then the camera does the demosaicing.
However, if there were only as many photosites as the final number of desired pixels, then the edges would not have enough photosites for accurate color information. For example, this is what happens when you try and compute color values only from the edge pixels:
In the example on the right, there are 4×4 or 16 color photosites. Usually, the value of each pixel is computed using four photosites. But when you get to the end of the row, there are only two photosites for the fourth pixel of that row. Therefore, to get a 4×4 grid of pixels in the final image, you actually need a 5×5 grid of photosites so each of the 4×4 pixels has full color information. (I simplified this process a little bit. In reality, the demosaicing stage typically uses a better algorithm than I just outlined.)
2. Second Reason: Black Level and Unwanted Dark Signal
However, these additional edge pixels are not enough to account for all the extra pixels. In fact, most cameras have pixels that are completely obscured from light! You can think of them as pixels with black paint on them. These are the so-called optically black pixels. Why should there be pixels on the sensor that cannot even sense light?
Unfortunately, even in total darkness, a sensor will still generate a signal (the dark signal) that will be translated into something other than pure black. This is undesirable, because obviously you want black to register as black.
This can be partially compensated for by using these optically-black pixels. By reading in the signal generated by these pixels, the camera can apply a correction to the whole image.
This correction is typically derived from a model that depends on temperature, which in turn is estimated from the optically black pixels. In practical terms, the hotter your sensor, the more unwanted signal (noise) comes through, and the camera estimates this via these extra pixels to account for it.
A similar technique is used in long-exposure noise reduction, where a dark frame is taken either manually or by the camera to reduce noise. Unfortunately, not all noise can be predicted from the optically black pixels (nor can hot pixels), which is why dark frame subtraction is still useful for long exposures.
How Many Megapixels Do You Need?
Now that I have covered the nitty gritty details of pixels, it’s time for the fun question: how many megapixels do you need? The answer is at least 100MP, and less may cause the universe to explode.
Okay, joking aside, how many megapixels is really enough? Should you get a 45MP camera over a 24MP one? These questions can be answered by considering two things: what is your final output medium, and how much do you need to crop? Let’s go over these two in a little detail.
1. What is Your Final Output?
If you’re mostly displaying your photos on the internet, you do not need that many megapixels at all. For example, a 4K monitor can be covered by 8.3 megapixels. (An 8K monitor, on the other hand, needs 33.2 megapixels. Very few people have such high resolution monitors, but they’re becoming at least a bit more widespread, so keep the 33 MP mark in mind if you wish to create 8K desktop backgrounds.)
Printing is another great way to display your work, and at least at larger print sizes, it tends to demand a larger number of megapixels. For close viewing distances, most people recommend printing at 300 pixels per inch. What does that mean? I means for every inch on your print, you’ll want that inch to be spanned by 300 pixels. It’s not a hard rule, meaning that if you have 270 pixels per inch, your print will look pretty good. So how many megapixels do you need for printing? Just take a look at this chart for some common large print sizes:
|Print size (inches)||Resolution for 300ppi||Megapixels for 300ppi||Resolution for 250ppi||Megapixels for 250ppi|
|8 x 10||2400 x 3000||7.2 MP||2000 x 2500||5.0 MP|
|12 x 18||3600 x 5400||19.4 MP||3000 x 4500||13.5 MP|
|16 x 24||4800 x 7200||34.6 MP||4000 x 6000||24.0 MP|
|24 x 36||7200 x 10,800||77.8||6000 x 9000||54 MP|
|32 x 48||9600 x 14,400||138.2 MP||8000 x 12000||96 MP|
As you can see, the number of megapixels you need becomes insane as the print size goes up. I guess everybody who wants to make large prints should go out now and buy the Fuji GFX 100S, right? Well, not exactly. The number of pixels you need is also dependent on the typical viewing distance!
People’s vision is such that a low-resolution print can look perfectly sharp if viewed from further away. To be specific, if you need about 300 PPI at one distance, doubling that distance will reduce your requirement to 150 PPI. And moving twice the distance will divide the number of megapixels required by four.
Since larger prints won’t tend to be viewed as closely, I have come up with a more realistic chart for pixel requirements:
|Print size (inches)||PPI||Resolution||Megapixels|
|8 x 10||300||2400 x 3000||7.2 MP|
|12 x 18||260||3120 x 4680||14.6 MP|
|16 x 24||220||3520 x 5280||18.6 MP|
|24 x 36||200||4800 x 7200||34.6 MP|
|32 x 48||180||5760 x 8640||50 MP|
This is based on my own personal preferences and thinking about where I’d put different-sized prints in my house. In other words, it’s highly scientific and not up for dispute. (Well, at least it’s a good starting point.)
How much resolution you need also depends on the print medium and the subject. A photo with a lot of fine details like feathers will appear worse if those details are obliterated compared to a photo of a person’s face at the same resolution.
From these considerations, I recommend the following: If you are happy keeping your prints at most 16×24 inches, almost any modern sensor will be fine (since the entry point on cameras today is usually at least 20MP). So, this means any recent micro four thirds, APS-C camera, or low-resolution full-frame camera will suffice. Even if you lack a bit of resolution for a print at your desired PPI, you can use software methods that can do advanced upscaling. Some cameras like the Panasonic GH6 also have pixel-shift or high resolution modes that are suitable for some subjects and provide more resolution.
On the other hand, if you want to print 24×36 or higher, you will have more freedom with a high-resolution full-frame sensor like that in the Nikon Z7, Canon R5, or Sony A1. An even higher resolution full-frame camera, like the Sony a7R IVA, which has 61MP, is an excellent choice for those who need to make massive prints.
And if none of these is satisfactory, then the Fuji GFX 100S is an amazing camera.
2. How Much Do You Need to Crop?
The second consideration is cropping, which in some cases is unavoidable. As a wildlife photographer for instance, I am often cropping because not all species are easy to get close to. Cropping is also common in macro photography, because the size of the subject in the photo is often limited by the maximum magnification of the macro lens.
So for shooters who use need to crop substantially, I would recommend the higher megapixel bodies like the Canon R5 over lower megapixel ones like the Canon R6. Looking at the print chart above, the 45MP of the Canon R5 will give many more print options. Even after a 1.5x crop, the 45MP of the Canon R5 will still leave you 20MP, whereas the 20MP of the Canon R6 will become 8.9MP.
Alternatively, if you use a long enough lens, you can “crop” by using a crop-sensor camera, i.e., micro four thirds or aps-c. Most of these cameras are about 16 or 24 megapixels at the most, but thanks to their crop factor, you could end up putting more total pixels on a distant subject than a typical full frame camera could manage.
As a final reason to buy a camera with more megapixels, there’s no denying that pixel peeping is a very relaxing activity. More pixels can be an option for their therapeutic properties.
A pixel is the fundamental building block of an image, and generally, the more pixels, the better. However, photographers are very lucky with modern cameras because most of them have more than enough pixels for almost any situation. For very large printing and cropping, it is definitely worthwhile to have more pixels, and so cameras in the 40-60MP range can be very useful. However, even a 20MP camera can make a very nice large print, and fewer pixels should not hold you back. I look forward to hearing how 100MP is the ultimate level of photography in the comments!
Thanks for the detailed post.
But what is missing is the consideration of the other side of the shot.
Example: If I have a detail resolution of 1 mm at 20 MP and f/50 mm, I need 80 MP at f/25 mm for the same detail resolution! Or I have to halve the distance.
I was reading about current smartphones and recalled articles by Nasim Mansurov pointing out how it was cellphones that were driving the camera industry. Computational photography advances were the way of the future, he observed, and he sure was right. Cellphones destroyed the point and shoot market, and now do amazing things that flagship mirrorless cameras cannot.
Nasim seems to have vanished from the website he created and brought to excellence.
His influence and voice are sorely missed.
Is he associated with Photography Life anymore? What is he doing? Will we hear from him again?
Sorry to rain on your parade here in Comments, but this is where Contact Us sends me.
Stop chasing Megapixels! m.youtube.com/watch…u4tTGn633I
It’s a good point about viewing distances. If your viewing distance is proportional to the size of the picture, then yes the required megapixels doesn’t change much. However, I don’t really subscribe to that rule. This might have something to do with the wildlife genre, but personally one of the reasons why I like larger prints is because I can get a little closer to them and see a lot of fine detail. I think it has something to do with the idea that ordinarily, it’s not easy to get up close to wildlife so there’s a natural tendency in a lot of people to want to see more…see what they can’t see in real life in other words.
Basically what this means is I like to get a little closer to large prints of what I print (birds), and with this in mind, I disagree with the premise of that video.
Now, I do agree with the premise of the video when it comes to certain kinds of prints. Let’s say it’s a portrait of a person’s face. Well in that case I probably don’t really want to get that close and see just the endless details in their nose, haha! With a portrait of a person’s face, it’s probably better enjoyed as a whole. But with wildlife, I think it’s also nice to be able to enjoy small parts of a picture, and that’s why personally I prefer higher PPIs.
In short, I think it’s an over-generalization to assume that what works in one style can work in every style. I guess all I can say is, know what works for you! :)
Some years ago an “IQ test” (Image Quality) was carried out comparing scanned images from Fuji Velvia 50 and Provia 100F which found that the pixel array in MP of these images correlated to Canon 6.0 or Nikon 6.3 MP cameras available at that time. I personally have printed A3 (420 x 297mm) images from a Canon 300D 6.0MP (Rebel) with very good results and have been successful with it in print competitions. So I believe it is not the MP count that is important but the quality of the original image. What about the algorithm’s used by the in camera processor?
The algorithms in camera do nothing worth mentioning, if you shoot RAW. After focusing and selecting exposure (shutter/apperture/iso) and pressing the release. RAW data is captured, no demosaicing, no denoising, nothing is done with the data (maybe settings and a thumbnail jpg are added).
Nikon cameras output “cooked raw data” not “raw data”. I think you will find that some, if not most, other manufacturers do the same.
Other terms used include:
• half-cooked raw
• half-cooked RAW
You are right that the quality of the image is also important. If you don’t have enough light for example then a lot of the pixels in a 50MP image won’t contain much or any additional information. Perceptible edges also matter too. I mean, of course a lot more to the image will contribute to its apparent quality than pure resolution. For example, if you shoot a 6MP camera in the most ideal conditions with a great lens of a bird, it’s obviously going to look a lot better than a mediocre lens on a 50MP camera from far away and then cropped. So definitely, a high quality 6MP file will go a long way!
I really miss here the information about needing more stable tripod or faster shutter speed with higher resolutions. 1/focal_distance sec may leave pixels sharp, if they’re quite big, but once those pixels get smaller, every small movement will make them blurry, hence either faster shutter or tripod is needed. So having APS-C 600Mpx would prevent handheld photos unless shooting the Sun itself.
After reading your article (Very clearly written), I came away with some questions based on my experience over the last 10 years. I did not get into digital until 2012 and bought a 16 megapixel sensor Olympus E-M5 as my first digital camera. I had prints made from these files usually about 13-14″ by 20-22″ depending of the orientation, etc. The prints were excellent and I was able to sell a number of them. The RAW files were converted to Tiff files for printing. The Tiff files used for printing were in the range of 30+ to 80+ Megabites and printed by a pro photographer/printer friend of mine. So my question is how is it possible I got excellent prints (based on my own eye, other photographer’s comments and sales) with 16 MP sensor given your criteria above. I’m somewhat (as you can probably tell) unsophisticated in this whole area, so please tell me what I’m missing/don’t understand. Thanks!
Thanks for the comment, Rene. If you take a look at my second table (the “realistic table”) for print sizes, your numbers are actually pretty close to what I got (I wrote about 18MP for 16×24), so that is quite similar to your 16MP for 14×22-ish figure. And you can get away with even less if the print will be viewed from slightly farther as my figures are also on the safe side. So basically, you’re not missing anything :)
Haha I guess I wasn’t paying enough attention and thought I was reading this on PetaPixel. After seeing the headline I went in with a very cynical attitude, ready to count how many errors you made. I was incredibly surprised by how good your article was compared to most PetaPixel intro-level articles! Finally as I went to give my kudus I realized I’m on PL, not PP. Then it all made sense! Thank you for the great article and maintaining PL’s high standards.
Thank you, Brian. Your comment is very much appreciated.
You actually need a 39.3 MP 3:2 image to cover an 8K 16:9 monitor. (39,321,600 pixels to be exact.)
If your image is 4:3, it will need to be 44.2 MP (44,236,800 pixels).
Yes, that is very true! Since the aspect ratio of most (all?) cameras is not 16:9, an image will have to be cropped to fit on a 16:9 monitor and so your greater megapixel counts will be needed from the source to account for cropping.
I am happy with the megapixel count on my DSLRs (D800 &D500) but I find nicer results on my Z6ii and Z9. I am guessing that the Z series lenses are responsible for that. They are cleaner in several ways and that gives you a different perspective on how many megapixels you need.
The extra mp on the Z9 are very nice. If I drop into DX mode on the Z9 it is actually collecting less information than the D500 (I believe), well just a bit. That said the Z9 has other things it does.
I also have to say that the Z6ii produces stunning images. It has more than enough clarity (dynamics, detail & dimensionality, to use a term from the audio world) for my printer, paper and ink. I only print A3+ (13×19) on pearl paper. Surprisingly different papers can “look” more detailed, which is another factor.
Thanks for your comments. You are absolutely right about different papers (not to mention different print shops). And you are also right that the Z9 has very slightly fewer pixels in crop mode, though probably not enough to affect anything. I also agree about the Z6II – I have the original and while I was a bit worried about the anti-aliasing filter, in practice it doesn’t feel constraining. Happy printing!
“The answer is at least 100MP, and less may cause the universe to explode.”
No, Jason: the last time having fewer than 100MP affected the universe, it caused the universe to implode, quickly followed by the rapid expansion named the Big Bang by scientists who were frightfully unfamiliar with the history of photography, obviously 😀
An astronomical blunder caused this cosmological cataclysm!
I stand corrected, Pete!