Camera Image Sensors

Unraveling the Mystery Behind the Image Sensor

Before getting into the discussion of the what and wherefores of camera image sensors, take a look at the sensor size below and the visually comparison to get a feeling of the differences in sensor size.

Also, image sensors can be a complex and convoluted subject to discuss. We will attempt to keep the discussion in layman term so even I know what I'm talking about. If you are an electronics engineer type (I'm Not) there are many technical articles on the web for you to read.

Here, we are going to try to KISS it. (Keep, It, Simple, S*****)

Sensor sizes shown below are for relative size comparison purposes only. Actual size in mm is shown below each image.

For those in the U.S.A. who haven't got their heads functioning in mm yet, the smallest sensor shown, 6.35mm = 1/4" and 4.29mm = 3/16" (more or less) that's 1/4"x3/16". The 35mm full frame sensor is 36mm = 1-7/16" x 24mm = 15/16 or for discussions 1-1/2" x 1". We won't talk about the Medium format sensor since a camera (body only) with very basic functions like the recently introduced Leica S-2 using this sensor costs over $27,000! I paid less than that for my new truck! Now, on to the nitty gritty! Oh, the blue line around the sensor, is the comparison to the 35mm full frame sensor.

I'm going to start out by saying physical size matters. Its NOT the number of pixels that count, but the size of your the sensor.

Of course, there is the algorithms (mathematical formulas) and programming outside the image sensor that really helps with cleaning up the image quality, but that is another complicated story. We'll have enough to deal with explaining why bigger is better.

So, what's this thing called a "pixel?" Well, for our discussion a pixel is a teeny, teeny, TINY little bucket that holds light photons. If your camera has 12mp (megapixels) then you have 12 million pixels crammed on your sensor. 12 million on that tiny sensor? Yep! Of course it doesn't take a big bucket to hold a photon! You'll soon see why the more photons your bucket can hold the better your image quality. Oops, that's getting ahead of the story.

So far, so good. Sensors have pixels, and pixels are teeny, tiny buckets that hold photons. Don't touch your sensor; it's likely to make your camera go blind! And that's the truth.

An image sensor is a device that converts an optical image photon to an electric signal. An image sensor is typically a charge-coupled device (CCD) or a complementary metal–oxide–semiconductor (CMOS) active-pixel sensor.

There are many technical articles about the image sensor but "technical" is beyond the scope of this article. As a matter of fact many of the articles I read are far into physics and if you care to go there check out http://www.clarkvision.com/ Otherwise we'll just stick to the basics so we can decide what size image sensor we want in our camera and why.

Evaluating Image Sensors

There are many parameters that can be used to evaluate the performance of an image sensor, including its dynamic range, its signal-to-noise ratio, its low-light sensitivity, etc. For sensors of comparable types, the signal-to-noise ratio and dynamic range improve as the size increases. More on this in a minute.

Importance of megapixels

Generally speaking the more megapixels you have the greater the perceived sharpness of the image and the larger the print you can make. As always there are problems introduced when getting more pixels onto a fixed size sensor. As pixels increase other issues like quality may go down. So let's take a look at megapixels first.

Megapixels over rated?

Is the number of pixels over rated? When it comes to making a print of your image they may be, but it depends on your final objective. The human eye has great resolving ability but there is a limit to the detail it can distinguish when looking at a print at a normal viewing distance. When looking at a photographic print a resolution between 240-300 PPI (pixels per inch) is about it. As a side note, the terms ppi, pixels per inch and dpi, dots per inch are interchangeable. Generally in the 8 x 10 print size and larger you want to output your prints at 300 ppi even if your printer will put out 4000 ppi. You really won't see the difference and you'll just waste ink and take forever for the printer to print the image. So ok, you need to print at 300 ppi; then what is the relations ship to your cameras megapixels? Good question, lets see.

PPI to Camera Megapixels

As I said above, for sharp prints they should be printed out at 300 ppi. Remember "PPI" is pixels per inch? Well, let’s see how many pixels in our camera's sensor we need for an "uncropped" 8x10 print.

Simply multiply the pixels in the width of the print by the pixels in the length of the print.  It will look like this: PPI x width x PPI x length= sensor pixels. Well, let's see then; 300 x 8 x 300 x 10 = 7,200,000 pixels. Since "mega" means million we need a camera with a sensor of at least 7.2 megapixels.

How about an 11 x 14 print? Since we'll probably be viewing it from a greater distance, maybe we can get by with 240 ppi? So, 240 x 11 x 240 x 14 = 8,870,400 pixels or 8.9mp. What about at 300ppi? Well that will get you up to 13.9mp.

So, since your camera has a 10mp sensor it will more than do the job, or will it? Even though your image appears sharp, is it high quality? Not likely. Let's see what happens with the number of pixels and the actual sensor's physical size, after we look at a little caveat.

A  Caveat

Keep in mind, the engineering of image sensors and imaging software is changing almost on a daily basis. For example, for my Canon PowerShot SX-20 12.2 megapixel camera, Canon claims I can get quality images up to 16" x 20". Of course, they don't define what they consider quality. Well, let's see where the calculations come out using our lower print resolution of 240 ppi.   240x16x240x20=18.4mp.  At 300ppi 300x16x300x20=28.8mp.  Has Canon done some magic with their algorithms? Or are they saying 200ppi is an acceptable image? 90,000 x 320=12.8 mp.  Hummm, seems they may be pushing the envelope a little..

Table of Camera Sensor Sizes

Exact dimensions may vary, but those listed are typical and are for illustrative purposes only.

Grin a bear with me through these little calculations because they become significant farther on.

When you want to find pixel size you may have to look and dig into several locations. Since I have a Canon PowerShot SX 20 IS and a Canon EOS 5D. Let's check the SX20. The First place I looked is in the owner's manual. Under the specifications I found the following information:

Camera effective pixels:  approximately 12.1 million

Image Sensor:      1/2.3-inch type CCD and total number of pixels approximately 12.4 million.

Number of recording pixels (still images):    4000 x 3000 pixels (12 million)

Checking the sensor size table we find the 1/2.3-inch sensor is 6.16mm x 4.62mm (less than 1/4"x3/16")

Now, since we know the length and width of the sensor and we know how many pixels are in each direction we can find out how big each pixel is.

First divide the width 6.16mm by the number of pixels 4000. 6.16mm/4000pixels = 0.00154mm/1pixel and the other is 4.62mm/3000pixels = 0.00154mm/1pixel. 0.00154mm x 0.00154mm and converting it to microns our pixel size is 1.54µ x 1.54µ

Now, let’s compare these pixels to the pixel size of my Canon EOS 5D

Camera Effective pixels: Approximately 12.8 million

Image Sensor:      35.8mm x 23.9mm CMOS and the total number of pixels approximately 13.3 million.

Number of recording pixels at large/fine:     4368 x 2912 pixels. 12.72 million

Our Pixel size calculates out to approximately: 8.20µ x 8.20µ

Each pixel in the Canon EOS 5D is 5.33 times larger than those of the SX20

So you say, "Big deal" you still only have about 12million. Well a big deal it is, because of a thing called "photon counting" and a thing called "Noise."  This is where "quality" comes in or falls out depending on which size sensor  you have.  Now we are getting the the reason bigger is better.

 CLICK ME  FOR Adobe Products

Photon Counting

The sensor in today's digital cameras uses a charge-coupled device (CCD), CMOS sensor or other similar device that is a field of pixels. Each pixel is a material that absorbs light photons and generates electrons.

The photons are gathered and held in what may be called a "well" or "bucket" or "pixel" that prevents them from drifting away. This is kind of like a bucket holding rain drops, where the photons are the drops.

For a side note, as the process goes, the photons generate electrons and the electrons go through a converter that converts the analog wave signal to digital data. This digital data can be stored on a memory card that later can be read by our computer or printer and displayed on our computer screen as an image or output by our printer as a print.

Let's get back to our photon buckets. Here comes the bucket brigade. When we want to read the image, the photons from the first bucket is emptied into the converter. The photons from the second bucket are poured into the first bucket, which then gets poured into the converter, and so on. These buckets of course have very specific construction specifications so the photons don't leak out. This in turn limits their minimum bucket size and how big the opening must be for that photon to get in and the size of the area where photons are absorbed to generate electrons.

Another important limit is that our bucket can only be one photon deep. So, when collecting light photons, only area matters. With a wide bucket we can collect more photons in a given amount of time than with a narrow bucket.

The importance of this is that the accuracy or quality of the signal is directly proportional to the size of the signal, meaning we have less unwanted signal in the form of "noise" from our larger pixel.

Guess what? The amount of noise we get is significant when we count photons in the well. Let me explain.

The noise in the signal is equal to the square root of the number of photons in the well. Therefore, noise is NOT lineally proportional as in 1 for 1. How significant is that? Let's look.

If your photon well holds 9 photons the square root of 9 is 3 photons of noise. If your photon well holds 40,000 photons the square root of 40,000 is 200.

Is this significant? Well 3/9=1/3 and 200/40,000 = 1/200. That's 30% noise compared to 0.5% noise. I'd say that's significant. So, how many photons does our sensor hold?

 I haven't calculated it, but I do know the Canon 5D pixel or well is 5.4 times larger than the SX20 and therefore, I'm going to get a far higher quality image.

Also, I now know that in low light conditions where there are fewer photons flying around, the larger sensor pixels of the Canon EOS 5D are going to gather more photons then the smaller pixel a lot better because they are going to collect more photons over a given length of time and since we have more in the same bucket we have less noise.

In conclusion, I think we can say a greater number of pixels will likely give us an apparently sharper image because we collect a great number of photons, but we also have to say, that larger pixels and consequently a larger sensor size is more important and will give us an even sharper image as well as a higher quality image because we collect even more photons in one larger pixel and one pixel has significantly less noise because of the square root of the pixel photon count.

Now that you know why sensor size counts, there is more to the story. A couple more things that add to image quality are Dynamic range or contrast, and ISO, formally know as film speed, but now in the digital image world is known as signal amplification.  We'll look at each and again see why bigger pixels are better pixels.

Dynamic Range - Contrast

During the days of black and white film and when Ansel Adams was still alive, I spent days calibrating a particular black and white film through exposure and development so I could get fewer or more steps along a gray scale from pure black to pure white. Fewer steps between pure black and pure white meant more contrast between each step and lower dynamic range, and consequently some shadow detail went to total black and some highlights went to pure white. On the other had when getting more steps between pure black and pure white we had less contrast between each step and a higher dynamic range. With high dynamic range we can preserve the detail nuances in the shadows and maintain greater detail in our whites. The goal was to get 10 steps or stops between pure black and pure white.

As was then and is now, when you have a very high contrast image (low dynamic range - fewer stops) to start with it’s nearly impossible to increase the dynamic range acceptably via darkroom manipulation or software adjustments. However, today things are a little easier and you can work with Low Dynamic Range images through multiple exposures and software HDR image editing. What does this have to do with our pixels and image sensors?

Let's think back about our pixel size. Remember, with a larger pixel we can capture more photons right? Well, each of those photons are probably are at a different energy level or brightness. With more photons each having different energy levels we have a higher (wider, more steps) dynamic range to work with. With small pixels we capture fewer photons and consequently we will have a lower (shorter, fewer steps) dynamic range and more contrast in our resulting image.

So, with regard to dynamic range, the larger the pixel size and the more photons we capture in that bucket with more energy levels the higher (more steps) the dynamic range and the higher the quality of the image.

This is why Ansel Adams’ images are so smooth and detailed from the brightest bright through the deepest shadows. He was known to use an 8"x10" image sensor (film) and knew how to develop the photons (film grain) through exposure and chemical development to give him a high dynamic range of gray tones he could represent in his image prints. He not only developed the zone system method but he was also a master at manipulating it.

Signal to noise ratio

Parts of the following is From Wikipedia, the free encyclopedia

In common use, the word noise means any unwanted sound. In both analog and digital electronics, noise or signal noise is an unwanted perturbation to a wanted signal; it is called noise as a generalization of the audible noise heard when listening to a weak radio transmission. Signal noise is heard as acoustic noise if played through a loudspeaker; it manifests as 'snow' on a television or video image. In signal processing or computing it can be considered unwanted data without meaning; that is, data that is not being used to transmit a signal, but is simply produced as an unwanted by-product of other activities.

Noise can block, distort, change or interfere with the meaning of a message in both human and electronic communication.

In low light, correct exposure requires the use of long (slow) shutter speeds, higher gain (higher ISO sensitivity), or both. On most cameras, longer shutter speeds lead to increased salt-and-pepper noise in the image.

Also, the relative effect of noise increases as the exposure is reduced, when increasing the ISO sensitivity, since fewer photons are counted and since more amplification of the signal is necessary.

The size of the image sensor, or effective light collection area per pixel sensor, is the largest determinant of signal levels that determine signal-to-noise ratio and hence apparent noise levels.

The sensitivity of a given imager at the same noise level scales roughly with the sensor area. For instance, the noise level produced by a Four Thirds sensor at ISO 800 is roughly equivalent to that produced by a "full frame" sensor (with roughly four times the area) at ISO 3200, and that produced by a 1/2.5" compact camera sensor (with roughly 1/8 the area) at ISO 100.

This ability to produce acceptable images at higher sensitivities is a major factor driving the adoption of DSLR cameras, which tend to use larger sensors than compacts. The number of pixels on a sensor greatly affects the noise level per pixel, since a sensor with more pixels for the same size must use physically smaller pixels. However, when pixels are scaled to the same size on screen, or printed at the same size, the pixel count makes little difference on perceptible noise levels.

So, larger sensors with larger pixels hold more photons, and as such create less noise because we have less amplification noise at a given ISO amplification than small sensors with small pixels. (hint: shoot with an ISO setting of 100-200 for the best quality image)

Quality and ISO

Def: The International Organization for Standardization (Organization international de normalization), widely known as ISO (pronounced /ˈa ɪsoʊ/), is an international-standard-setting body composed of representatives from various national standards organizations.  – They attempt to standardize measurements.

Unity Gain Sensitivity (ISO)

Camera manufacturers set ISO based on some fraction of the maximum signal that can be recorded. The maximum signal is called the "full well capacity," which is the maximum number of electrons (converted photons) that a pixel can hold.

Larger pixels in general hold more electrons. For current technology of CCD and CMOS sensors, the full well capacities run about 800 to 1600 electrons per square micron. These values haven't changed much on over twenty years of sensor development.

The setting of ISO implies that cameras with different size pixel collect the same amount of light per unit time for a given f/ratio. That is incorrect because, the ISO definition relates to the fraction of light relative to the full well capacity, not the total light collected.

For a given f/ratio and exposure time, a camera with larger pixels collects more photons. The camera designers change the gain (amplification) of each camera based on the full well capacity.

A property called the Unity Gain shows the true sensitivity of a sensor. The much higher ISO's of large pixel cameras show that they have much better low light performance. (large sensors with large pixels work well in the dark)

Conventional film comes in different sensitivities (ASAs) for different purposes. The lower the sensitivity, the finer the grain, but more light is needed. This is excellent for outdoor photography, but for low-light conditions or action photography (where fast shutter speeds are needed), more sensitive or "fast" film is used which is more "grainy".

Likewise, digital cameras have an ISO rating indicating their level of sensitivity to light. ISO 100 is the "normal" setting for most cameras, although some go as low as ISO 50. The sensitivities can be increased to 200, 400, 800, or even 3,200 or higher on high-end digital SLRs.

When increasing the sensitivity, the output of the sensor is amplified, so less light is needed. Unfortunately that also amplifies the undesired noise. Incidentally, this creates more grainy pictures, just like in conventional photography, but because of different reasons.

Higher ISO is similar to turning up the volume of a radio with poor reception. Doing so will not only amplify the (desired) music but also the (undesired) hiss and crackle or "noise".

Improvements in sensor technology are steadily reducing the noise levels at higher ISOs, especially on higher-end cameras. And unlike conventional film cameras which require a change of film roll or the use of multiple bodies, digital cameras allow you to instantly and conveniently change the sensitivity depending on the circumstances with a push of a button or turn of a knob and in full auto it will make all the adjustment by itself.

Summary

In summary, small sensors have small pixels, collect fewer photons, create more noise, and have lower dynamic range meaning more contrast.  While Larger DSLR sensors have larger pixels, collect more photons in each pixel, create much less noise, have a higher dynamic range, and can control contrast better.

So, is it more pixels or a bigger senor that counts most?  Yes, bigger sensors with larger pixels is of the ultimate importance.

What image sensor do you want in your camera?