Lessons

Lesson Info

Pixels

The topic is the sensor and it is time to dive in and talk about the individual pixels that are recording a light. Now the word pixel is a conglomeration of the words picture and element. It is the light sensitive cell that is gathering light in our cameras. A very common sensor these days will have four thousand rows and six thousands columns and four thousand times six thousand is twenty four million and that is what we call a twenty four megapixel sensor. Now each of these different pixels receives a little bit of light that usually comes through some micro lenses to help keep the light directed coming straight in whether it's in the middle of the sensor or off to the corner. And something you've surely noticed is that newer cameras have more pixels than older cameras. What's going on? Well if the image area is the same, they need to make smaller size pixels. In some cases they kinda push them closer together, but for the most part the pixel size is reduced on newer cameras so that ...

they can fit more pixels into that same imaging area. And that is where things can get a little complicated because smaller sensors, smaller pixels, are not necessarily as good as larger pixels. And so there is gonna be definitely a balance between the number of pixels and how big they are. And so I wanted to do a little test between a ten megapixel camera which is actually from a camera about ten years ago, about ten years old. And comparing it up against the top of the line fifty megapixel full frame sensor and I wanted to see how much difference would we see at various levels? And I'll just kind of remind you here in the middle, these are both twenty four megapixel sensors, one's on a crop frame, one's on a full frame. The twenty four megapixel in the crop frame, that is where many of the most popular interchangeable lens cameras are these days, both mirrorless and crop frame. So let's go ahead and take a look and enlarge the middle section of this and take a look at the results between all of these images. And if you look closely, you'll notice there's not a huge difference there between ten and fifty megapixels and we're getting pretty good light on this. We are getting a little bit sharper definition over here on the edges, and so you really have to blow things up very large in order to see much difference. And so for most people, I think fifty megapixels is way overkill. In fact, twenty four megapixels is probably a much sweeter spot and for some people, ten megapixels is more than enough. And so it is not necessarily just more is better. In strict resolution terms, yes. Everything else being equal, more is better. Having said that, you still have to deal with the storage issues which is another problem. And so, wanted to do a quick comparison between a bunch of different megapixels there to see what they look like. Alright, next up. More or bigger, is the question. Imagine you were to go to the donut shop, and you work in a small office and you've got eleven coworkers and you're gonna buy donuts for everybody so you wanna buy a box of donuts for everybody, alright. So you get a box of a dozen donuts. And as you're checking out, they say, hey, did you know we're running a special today? And you can get twenty four donuts for the same price as twelve. What would you say? Twenty four donuts for the same price as twelve? Sign me up! Well they take out the box and they start loading it up with donuts and you suddenly realize, you're like, excuse me, those donuts are not as big as the other donuts. Well, no we can't fit twenty four donuts into that size of box if they are that size. We gotta reduce the size of them. And you're like, but I only have eleven coworkers and we have a rule at work that only allows people to eat one donut a day. And so this doesn't do us any good at all. And so the question is, is what's better, more or bigger? Okay, size versus quantity, it's a very important question whether you're choosing donuts or whether you're thinking about megapixels. And what it really comes down to is that more is better if you need it, but it's not better if you don't need it. And so whether we're talking about twelve donuts, twelve pixels, or twelve megapixels versus twenty four megapixels. More is better if you have a need for it. If you don't need all of those pixels, they're not gonna help you out. And you really have to make a compromise and it's a tough compromise in many cases. But let's continue because you'll see more about this compromise as we go through this. So what is the difference between twelve and twenty four? Everyone thinks they know the difference between twelve and twenty four, right? Twenty four is twice the number of twelve, are we right? Am I playing any tricks on you yet? Alright, so it's a hundred percent more information, right? Well what if we were to count resolution, where we would count rows of information. Over here we have four rows, over here we have six rows of information. So when it comes to resolution, horizontal resolution, it's increased by thirty three percent, not a hundred percent. And so it's a little math game going on here and it takes a lot of pixels to really double the resolution. Even though you've doubled the number of pixels, you have not doubled the resolution of your sensor. So keep that in mind when choosing a sensor that has a few more megapixels. It doesn't make much of a difference. I wanted to compare two different twenty four megapixel sensors to see how they compare between a crop frame and a full frame. What's the difference on these? Well if we were to look at a small section of each of these sensors, and you could look in real closely, so closely that you could see the actual pixels on them yourself. We're gonna look at this really tiny area on the sensor. And we're looking at one millimeter across. And there's gonna be a whole bunch of pixels, and there's still really too many to really get a grasp of so what we're gonna do is we're gonna go all the way down to one tenth of one millimeter across. So this is a square one tenth of a millimeter on each side. And the difference between the twenty four crop and the twenty four full frame is that you've got twenty five pixels and sixteen pixels. Now what's better, more pixels or bigger pixels? Well if you need the resolution, you're gonna get a little bit more resolution here because there's more pixels to define that fine tuned area. But these pixels are bigger, and remember pixels are a little like solar arrays collecting light. The bigger that array is, the bigger that pixel, the better it's gonna do at absorbing light. And so this is gonna be better under the light, that one is gonna be better for higher resolution. And so when we look at the difference, at ISO one hundred, I'm hard-pressed to tell the difference between these two cameras. When we get good light in on a twenty four crop or full frame, I can't see the difference. Where I can see the difference, is when I pump up the ISO, and I jack it up to twenty five thousand. And now you'll notice the image on the left is more noisy or grainy, and we're gonna talk more about this as we get into the ISO settings. And so if you plan to crank up your ISO to twenty five thousand, you're gonna get much better results with a full frame sensor than with a crop frame, because they have larger pixels that can pick up low light. Now I am jumping ahead of myself a little bit here, but a lot of you know exactly what I'm talking about at this point. If this is the first time on this merry go round, you'll catch it the next time around. Alright, let's talk about those pixels. So a good analogy on the way the pixels absorb light is very much like a bucket out in a rainstorm collecting rain. And so as we have light coming into this first bucket here, if the bucket is completely filled to the very top, we know that that will be a white pixel. There we go, white pixel. So on our next bucket of light, we're gonna fill this up half way. When the bucket gets half way filled, it's gonna be a middle toned gray pixel, half way between white and black, and just as a side note, we're forgetting about color. This is just between white and black. And so the camera looks at how much light this pixel is receiving and that is a representation of how bright that pixel will be in the display or when you actually produce the actual image. So if just a little bit light comes as in this bucket here, it's gonna be a very, very dark image, almost charcoal type gray. Alright, so this next one, what we're gonna do is we're gonna put in a whole bunch of light. Our cup runneth over, okay. We've pointed our camera at the sun. What happens there is that we've had too much information on the sensor, and this something that we call clipping. What is brighter than white? The answer is nothing, you just can't get any brighter than white and so that is information that we've lost, and that's something we need to be aware of, and we'll talk about when we get more into the exposure section. Alright, so let's take this next one, let's fill it up half way, and we have a middle tone gray. Okay, we're good. How do we know it's middle tone gray? It's because the bucket is half way filled up, that's what it's supposed to look like. And we have our cameras that are knowing where the top of the bucket is, but what we can do, is because it's an electronic device that we can digitally manipulate, we can lie and we can say, that's not as bright as it should be. Pretend that's the brightness and we're gonna crank up the brightness. Some of you, I bet, remember the old TV sets that had the brightness knob on them. You remember the brightness knob on that? Did that change how bright it was in the studio where they were filming? No, that just kind of amplified the signal and this is what you're doing in the camera when you raise the ISO setting, is you're saying well, I know it's gray, but let's just make it white by making it brighter. So what we're gonna do is we're gonna take these final four pixels and we're gonna put in just a little bit of light and we're gonna crank up the ISO and we're gonna look at what the problem is with just cranking up the brightness on your camera by raising the ISO setting. And so we're telling all of these, even though you didn't receive very much light, just make these bright. And in theory, all four of these little areas here should be exactly the same tonality. But the fact of the matter is in the sensor, there's all sorts of electronics, there's heat that's built up. And that causes just a little bit of interference in the signal. If we go back to the analogy of the bucket of waters, it's like a layer of scum right at the bottom of that bucket of water. Now if you had to drink water from a bucket that had a layer of scum down at the bottom, would you choose a bucket that hardly had any water in it at all, or would you choose a bucket that was completely filled with water? Well we wanna have as good a signal as possible. And this is where we get into the whole signal to noise ratio. How strong is the signal? How much noise is in there? And so in this case, it's equal noise in all of them. But when we don't have much light in there, the signal is not very strong and so it's got a very poor signal to noise ratio, which means that it's gonna interfere and we're gonna get mismatching results and we're gonna end up with something called noise that looks like this gritty sand paper. And that's what's gonna happen, especially in the dark areas of our photographs if we crank up our ISO and just try to make everything brighter in the camera itself. And so this is where pixel size comes into play. The Sony a7S II has only twelve megapixels, but it is a highly sought after, fairly expensive camera, because it is very good under low light conditions. Now on the other end of the spectrum, we have an Apple iPhone 6s, which has a very nice general purpose camera that is also twelve megapixels. How big are the actual pixels on these two products? Because they both take good pictures for what they do. One is on a full frame sensor, one is on a one third inch type sensor, which means it's smaller than one third inch. And so forty three millimeters versus six millimeters. If you want to look at them relative in size measured in micrometers, it's eight point four micrometers across versus one point five. Once again, these are like solar arrays collecting light. The bigger the pixel, the much better it's gonna do, especially under low light conditions. And this is why the Sony a7S II is so much better under low light conditions than your typical phone, is that the pixel size on it is humongously bigger. Alright and so this is the extreme example, the largest and the smallest of the common ones that I could find. The other differences are smaller, but that clearly shows you the difference in the pixels. Now I like to test cameras, and I wanted to do a good test, comprehensive test of a variety of cameras out on the market. So pretty much hard to find any new camera less than sixteen megapixels these days, but we have a lot of cameras around twenty two, and I did test this against the twelve megapixel full frame all the way up to the fifty megapixel Canon 5DS, which is the highest resolution camera of anything full frame or smaller. And I wanted to look at the results of these, so let's go ahead and take a look at these results. And at ISO two hundred, you're not gonna notice a huge difference. Because if you're getting good light in, it's more about the number of pixels, and you can probably see a little bit more sharpness with the fifty megapixels but it's not incredible. We'd have to blow it up even larger to see that difference much more. But let's crank up the ISO pretty high, so we're gonna go up to sixty four hundred. This is a pretty high number here. And this is where you're gonna notice it's gonna be cleaner with the full frame cameras. And over there on the far left, we're getting the worst results because we have the smallest size sensor and the smallest size pixels on that as well. And so bigger sensors will allow you to shoot under lower light with better quality. Now the pixels themselves can be laid out in different types of arrays. The common system is known as a Bayer system. And it has to do with a quirk in the way the humans see with their own eyes. And it's alternating green and red, green and red, and then green and blue. And so there's twice as many greens as there are reds or blues, and this is what our sensors look like if you could get up really, really close to them as far as which color they are recording. Now this has a little bit of a problem, and that is if you had a red line, somebody had a red stick that you were photographing, that red stick could be very easily photographed horizontally or vertically, and depending on how many pixels you had, it could go pretty easily forty five degrees, but it's a little bit harder because we don't have as many pixels going in these other directions. And this can be a problem with something called Moire. And so by this illustration, you can see how disturbing this video looks when we have these two sets of lines that are slightly mismatched as we move them back and forth. And this is a problem that we had, oh, I think it was mostly back in the seventies with television announcers on TV that did not have TV-friendly suits. And their patterns of their suits and their ties did not match well with what was being used in the cameras these days. So to avoid this problem, most cameras on the marked have what's called an anti-aliasing filter in this. And so the way it works is we got our image sensor which usually has an infrared filter, which we're not gonna get into, but it's there. And then they will have a couple of other AA filters within the camera. And this is most cameras on the market. And so what happens is, as light comes through the first AA filter, it splits one beam of light into two, and then it goes to the second one and splits those two beams into four beams so it's taking one pinpoint of light and it's spreading it out into four. And it's blurring the image ever so slightly. And you think this would just be totally harmful on the image, but it's blurred ever so slightly that you would barely notice it just looking at the photo. And so this is what they've had to do because they didn't have enough pixels on the sensor. Well as more and more sensors have been raising up in resolution, they found out that they may not need this AA filter, because now there are so many pixels, we don't have a problem with these lines not lining up with the pixels on the sensor. And so there is an increasing number of cameras out that have dropped out the AA filter, and it records much more of the pure light coming in the camera. And so it would record a little bit of a cleaner signal, and I don't have an example to show you between with an AA filter and without, because generally the cameras are either made with or without it and the difference is so incredibly subtle, you can find it, there are some test sites like dp review that will show you the test results between models that can turn it on and off. But it's a very, very slight difference but it's one of the things that's going on with the sensors in your camera. And so, wanted to give you a list. This is not a comprehensive list, but this is a list of some of the cameras out on the market. There's a number of Nikons that have this and we've started to see this through some of the other parts of the industry as well as cameras that don't have this. And they will talk about this in the literature of the camera that it has no AA filter. Sometimes they call it an optical low pass filter. And these are cameras that are capable of a little bit higher resolution than their pixel count might indicate. But you might have a Moire problem if you were to, say, focus on fabric or textures that had a really fine weave to it. And so if you were doing textile photography, you might not want one of these type of cameras. But it's one of the choices that you'll find out on the market. There are some other different types of sensors we're not gonna spend much time on. Sigma has one called a Foveon sensor, which in theory looks totally awesome, but in practice, it's been really hard for them to bring this to the market and we haven't seen it on that many products. But it actually has the colored layers stacked, so that each pixel could be either red, green, or blue. We don't have that red, green, blue, green pattern system on it. And so it seems like they're still working on it and it just hasn't gone very far and I don't know that it is. Now Fuji has taken a completely different approach, and what they've done with their Bayer type system, is they have kinda reordered the pixels, and they've kinda scattered the green and red pixels. And what they were trying to do was to emulate film grain in its randomness. And they have come across something that works out quite well and those who use Fuji will tell you that this system does quite well. And so you will find that when you are comparing a Fuji camera against another camera where they have the same number of megapixels, Fujis usually do a little bit better. And it's partly because they have some really nice lenses, and most of their lenses are pretty nice lenses, but their sensors, they pack a really powerful punch for what they do. And so I would expect to continue to see further development and innovation in this department. Because this is something that's gonna get better. Kind of one of the theoretical ideas is the organic sensor that has some sort of organic material where the pixels could be any size, as small a size as you can possibly imagine. The theoretical best sensor would be one where you could dial in how many pixels you have. Because you could say, I only need twelve megapixels for this image, and they're really big pixels that do really nice under low light conditions. Or let's crank it up to a hundred megapixels because I got a lot of good light and I need really fine detail. That would be really nice to have. Doesn't exist though, not yet.