Photo IQ - Week 1 -- Sensor Size and You

One of the comments that I hate more than anything when people see my images is, "You must have a really nice camera!" The implication is that the only thing standing between an amateur and a professional is the cost of their equipment. In picking a teammate for basketball, would you choose LeBron James or my uncle Earl? What if Earl had better shoes on and a nicer basketball? Yep, still LeBron. While there is a huge body of knowledge, skill, and experience that separate a true professional from an amateur, I admit that yes, I do have a really nice camera...and yes, it does make a big difference.
Here's why:
First humor me as I define the two types of cameras we're talking about here.
A "point and shoot" is the standard compact camera that makes up 80% of the cameras people carry around with them. They're designed to be compact, inexpensive, and easy to use. Most digital point and shoot cameras are designed to be used by framing the photo on the LCD on the back of the camera, many don't even have physical viewfinders anymore. Digital point and shoot cameras also have very few moving mechanical parts.
Here are some examples of point and shoot models:

A DSLR or Digital Single-Lens Reflex camera has interchangeable lenses, a mechanical shutter, and are designed for image quality, versatility, and speed. Despite recent changes, most photos with a dslr are taken while looking through a physical viewfinder.
DSLRs look something like this:

>

From the advent of digital cameras we have been conditioned to look at pixel count as the sole statement of quality on the cameras we buy. For years all camera marketing was geared towards more and more megapixels...at least until October of 2009. In October of last year Canon released their G11.

The latest in their flagship point-and-shoot line had one very interesting difference from the G10, its predecessor--5 million FEWER pixels!
Whaaaaa?!? As your entire photographic world comes crashing down, realize that there is a method to Canon's madness and it leads us into the heart of today's issue.
The megapixel race is over.
Pixel number determines the physical size of the image, not the quality. 10 million pixels is around 3888×2592. To translate that to a full-resolution print it equates to capturing an image that is 13x8.5 inches. This means that anything you print smaller than an 8x10 you actually have pixels you don't even need. I've printed picture-perfect 30x40s from 10mp files. All cameras with 10 mp will print essentially the same sized image at full resolution.
The holy grail of megapixels was always 8. Eight million pixels is the basic equivalent of 35mm film so anything you could do with a 35mm negative you can do with 8 megapixels. That is the reason most camera makers have held at around 10.
The size of that standard 35mm frame of film is still the standard when it comes to sensors. A "full-frame" sensor refers to a sensor that is the same size as the classic 35mm negative. Typically, digital equipment is still measured in comparison to that standard.

The device on a sensor that captures the data is called a photosite. There's one photosite per pixel therefore if the camera has 10 million pixels, then it effectively has 10 million photosites.
The problem is that as you pack more and more photosites onto a sensor they need to get smaller and smaller and must be closer and closer together. This doesn't make a difference in size on your computer or in a print. (Pixels on your monitor are the same size and spacing regardless of the size of the original.)
Larger photosites absorb more light in the same period of time so they have more color data to work with and more dynamic range so colors are truer and stronger. Shadows become deeper and highlights keep detail in more extreme conditions. More contrast means clearer details and more depth to the image. Also, when the sensor is hit with electricity, the closer the photosites are to each other, the more heat is generated. Heat affects the way the photosite records data and translates into noise in the final image. So to go back to the G11, by only putting 10 million pixels on the same area that previously had 15 million you can have larger and more spaced--meaning cooler--photosites. That all comes down to better low-light and high ISO performance. Canon had decided to sacrifice quantity for quality.


What about my camera? I'm glad you asked. Below is a diagram of the typical sensor sizes for most of the cameras on the market:
I have removed a couple of the less-common sizes for ease in understanding the drawing. This diagram is not actual size but you can see the huge differences in the size of various sensors. (Actual size is further down.)

For the purpose of visualization, we will imagine that each sensor has 10 million pixels on it. It is easy to conceive the difference in the size of the individual photosites between the different sensors.
The smallest sensor, the 1/2.5 (That means the diagonal is 1/2.5th or 0.4 of an inch) is the standard sensor for point and shoot cameras--about 80% of the cameras you see. Our example above, the G11 has the 1/1.7 sensor (or .6 of an inch diagonal.) That's the same 10 million pixels but they have a full 50% more space than the average point and shoot.
You get out of point and shoot models into the four-thirds system in cameras such as the Panasonic and Olympus 4/3 DSLRs.
By the time you get to the standard APS-C sized sensors in the majority of the Canon and Nikon DSLRs you have more than 13 times the sensor space for those same 10 million pixels. At Full Frame you are looking at 2.5 times the size of the APS-C and 35 TIMES the size of the standard point and shoot!
Another aspect of sensor size is the ability to shorten the depth of field to blur the background and get the beautiful soft background that is characteristic of professional portraits. To get the same depth of field at 100mm and f2.8 as a Canon 7D with an APS-C sensor a point and shoot must be at f0.8 which simply isn't possible and would be extremely expensive to make if it was.

If you're curious about the ACTUAL size of the sensors, they would be about this:

 (Note: Cell phone camera sensors are much smaller than even the smallest sensor on this chart so hopefully that helps you understand why mobile phone cameras are so...uh...basic.)

To illustrate the concept, I've attached three photos from three different cameras with three different sensors.
These are each a pixel-to-pixel crop at 100%. All were shot at 10 mp and 800 iso. (800 ISO was chosen because it is a common choice for the auto setting in all three cameras and the differences are clearer at that sensitivity. All three sensors would look better at lower ISOs)

Canon Elph SD1100 - 1/2.5 Sensor (Standard Point-and-Shoot Sensor)

Canon G11 - 1/1.7 Sensor

Canon 7D - APS-C Sensor

The noise difference is apparent as is the color saturation and the contrast. The last image is clearer, cleaner and has more depth to it.  
What do we learn from all this jargon and mumbo-jumbo? We learn that as a start in determining Image Quality, equipment matters. A valid way to take technically better photos is to spend more money. With that out of the way, in the rest of this IQ series over the next few weeks we'll discuss ways to improve your Photo IQ that you can apply on any camera.

So to sum up, this is your image...and this is your image on APS-C. Any questions?

For more information about exactly how all of this works, here is some homework:
http://www.cambridgeincolour.com/tutorials/digital-camera-sensor-size.htm