Doing some reading on allegedly advanced photo websites you might come across a popular claim that the higher the megapixel density in a sensor the more image noise you will end up with in your images. And at first this seems to make sense:
- First, the sensor has precise dimensions, so to get more pixels in it will naturally mean that the pixel’s size itself must be shrunk down in order to fit. This way it’ll occupy less area. Unfortunately, these smaller pixels also allow less light to be captured and hence create more artifacts (or image noise).
- Second, if you look at the digital camera market you will come across models such as the Sony A7S (12 MP) and the Nikon D4S (16.2 MP) which have specifically reduced resolution sensors for low-light shooting (compared to their counterparts Sony A7R (36.4 MP) and Nikon D810 (36.2 MP) respectively) which only seems to prove the above statement.
- Thirdly, even when looking through dpreview.com reviews, you might come across comparisons like this:
Do you see more image noise in the higher megapixel camera? Of course you do.
“So what, now I shouldn’t trust my own eyes!?” you ask. No, not at all. There’s nothing wrong with your eyes, but there are some questions you should ask about methodology used to get the sample comparison shots above.
So what’s wrong with the methodology? Look at the pictures more carefully. You will notice that the ones with more image noise also depict a bigger subject. The reason for this is because they have more megapixels and on a screen of equal resolution you need to zoom in to correspond one pixel on the screen to the same one on the photograph - this, is called 1:1 (one-to-one) or screen comparison. And yes, the image with more pixels will also have more noise.
However, you don’t look at pictures in 1:1 in the real world. When you get your picture printed or upload it to Facebook, or look at it on your computer screen, you fit them in the same frame or screen area, so they all get scaled to the same size. That means that all those extra megapixels get combined into one. So, one pixel on the screen or one dot on the photograph eventually combine the light of several camera pixels. What does this means in practice? Let's take a look at the images scaled to the same size of 16 and 12 MP:
As you can see, the results are much different now.
Even though you can see a little bit less image noise with the reduced megapixel cameras, and especially color noise in the darker parts of the image, the overall difference is pretty small considering all the drawbacks that less megapixels will lead to (especially to justify 2+ times the increase in price). If we take a look at the comparison between two non-high-end pro low-light cameras with the same APS-C sensor but different amount of pixels (such as Sony NEX 7 and NEX 5n) the difference in noise in the scaled images becomes very subtle if not non-existent. Furthermore, images from the higher-megapixel cameras have not only the same amount of image noise but very often more details to them. Again, this happens because even though in higher resolution cameras each pixel gets less light, on the screen (or in print) those pixels get combined into one, so the pixel on the screen will have the same amount of information (light) as the one from the lower megapixel camera.
So, given all the information above, you basically gain nothing from less megapixels? Well, that’s not entirely true. High end cameras developed specifically for low-light photography have an advantage over their ordinary counterparts in dynamic range and color depth at high ISOs as well as offer, on average, about 0.33 stops more in exposure given the same amount of noise.
Is 0.33 of a stop worth it, given the compromised quality at good light? For most people – absolutely not. But if you’re that guy who does a lot of astro photography or takes pictures inside coal mines and those photos put bread on your table – then yes, by all means, those cameras are the best you can get.
The only example where those pro low-light cameras do kick ass is when shooting videos in low light. As it gets processed in camera, unlike photos, the difference is really HUGE:
So, here's another reason for you to splurge on a low-light camera: amazingly clear video footage with nothing but candle light!
As for the consumer level cameras, there’s really no correlation between the amount of megapixels in sensors and the dynamic range, the color depth or the amount of image noise at a given ISO.
So, all in all, there’s no objective ground to pay a premium for fewer megapixels, nor is it worth paying for more megapixels unless you’re going to print your images billboard size.
Megapixel count is becoming a rather an unimportant characteristic in today’s cameras as very little depends on it. All modern cameras have enough of them (8 MP) to print images up to A4 format with excellent quality and those who want more, they can easily find much higher resolution ones.
As for the low-light photography, reduced megapixels don't make a mind-blowing difference like the advertisers would have you believe. On average (on 18% gray) they will give an advantage of only about 0.35 stops over their higher-megapixel version. The advantage will be a little more on the darker areas of the image, but will hardly ever exceed 0.5 of a stop.
It might not be much for practical use, but technologically it's a giant leap forward:
Still, given everything, I wouldn't recommend spending extra money on those cameras unless you're absolutely sure why you need them.
This article uses some illustration materials from following sites (you should check them out, too as they are very useful!):