As I recently added a used Canon 1Ds mk II full-frame digital camera to my existing arsenal of 1Dmk III (1.3x crop) and 40D (1.6x crop) it occurred to me that I now had a camera of each sensor size in Canon's digital SLR lineup. I was therefore perfectly placed to conduct some interesting comparative tests.

Through the two parts of this tutorial, I wanted to challenge - through practical means, whether some widely stated opinions on the characteristics of each sensor type can actually stand up to close scrutiny.

In part 1- I examine the reputed telephoto benefits of crop cameras.
In part 2 - I examine image quality between the cameras and the claim that full frame cameras have less's depth of field than crop cameras.

Background

(You may wish to skip to claim 2 if you are already familiar with the basics of sensor sizes and wide angle benefits)

A full-frame sensor in a digital camera is approximately 35mm in width and will produce a image that resembles a 35mm film camera's in that it shares a similar angle of view and magnification on the recording medium (whether it be film or a digital sensor.)

Full frame sensors are currently found in the Canon 5D, 1Ds range and the Nikon D3 and D700.
Benefits of full frame sensors come from the fact that their pixels are usually large and more spread out over the sensor - which should give superior image quality and lower digital noise at each ISO setting compared to a camera with a smaller sensor (crop camera).

The sensor in a 1.3x sensor camera (such as the Canon 1D range) crops the image by 30 percent and a 1.6x sensor (such as the Canon 20D-40D crops the sensor by 60 percent.). Most Nikon cameras such as the D80, D300 etc have 1.5x crop sensors.

Here is an illustration of how each of these sensors would capture an image with the same lens from the same distance:

Claim 1 Full - frame cameras are better for wide-angle work

Well, this is a no-brainer really. Look at the image above - the full frame image is totally uncropped in camera, unlike the other two cameras, so a wide-angle lens will appear much wider on full frame than on a crop sensor camera - end of story.

The focal length of the lens does not change irrespective of what size sensor the camera has. A 28mm wide angle lens for example is still 28mm on a camera of any sensor size, but the field of view does change, so a much smaller part of the image fills the frame on the crop sensor.

To all intents and purposes the 28mm lens acts as though it has become 1.3x or 1.5x or 1.6x longer than it actually is on the crop cameras. To differentiate this pseudo-multiplication effect, we need to talk about effective focal length as opposed to actual focal length. So a 28mm wide-angle on a full frame camera is still 28mm x 1 = 28mm. On a 1.6x crop camera it becomes effectively 28mm x 1.6 = 45mm. Such a 45mm lens is not considered very wide angle at all. This is why crop cameras are frequently supplied with an 18-55 kit lens - as this gives a very useful equivalent field of view to 28-88mm on a film or full frame camera and makes a great walk-around lens.

One disadvantage of full frame sensors for wide angle work is the fall off in light at the corners - vignetting. The crop sensor has effectively had it's edges and corners "removed" and is therefore working in the lenses "sweet spot" for image quality.

Claim 2. Crop-sensor cameras are better for telephoto work, where you can't get as close to your subject as you would like.

Imagine that each of the sensors represented by the white boxes in the above image held the same number of pixels - let's say 10 million (or 10 megapixels) for example. The 1.6x crop sensor - being much smaller, will require it's pixels to be a little smaller and to be packed tighter together in order to fit them all into the small space. The density of pixels is therefore high. In theory, this means that when photographing a bird or animal that is hard to get close to, there is a telephoto advantage to a crop camera.

Let's clarify this a little. In the image above, the blue butterfly will have more pixels covering it if the sensor has a high pixel density - as is the case in the 1.6x crop camera. The 1.3x camera with the same number of pixels (10 Megapixels) should be at a disadvantage due to lower pixel density (same number of pixels but bigger sensor area). A full frame 10 megapixel sensor should be even more disadvantaged.

However, to make things a little more even, manufacturers have been able to put additional pixels on the big, full frame sensors in order to fight back the deficit due to lower pixel density.The 1DsmkII that I use has 16.7 megapixels. Even so, it is still theoretically disadvantaged as you can see in the table below when cropped to 40D size (a high pixel-density 1.6x crop camera). It is important to understand that it is the high pixel-density of the crop camera that is providing a theoretical telephoto advantage - not the crop factor itself. If a large full frame sensor had enough pixels packed on it to match the pixel density of the small sensor crop camera it would not be disadvantaged in telephoto comparisons.

In the chart below, the 40D is a 1.6x crop camera, the 1DmkIII a 1.3x crop and the others are full frame.

* Notice how many more pixels are available on the 40D crop camera.
Since this article was written, Canon have release the 7D camera with a massive 18Mp on a 1.6x crop sensor it has the highest pixel density of any digital SLR (as of September 2010). In theory this should make it very powerful telephoto tool.

So in theory the crop camera has a big advantage in telephoto work. Let's see if this actually born out in practice.....

Expt 1. Images taken from the same camera to subject distance

In the first experiment, I attached a 180mm macro lens onto a tripod and (without changing the tripod to subject distance throughout) I then attached the full frame camera and took a shot at f8 and 1/1000 sec to eliminate any chance of camera shake.The lens will work near its best at around f8 and the reasonable depth of field that f8 provides should eliminate any minor focusing inaccuracies.


Full frame sensor camera image (Canon 1Ds mk II)


1.3x sensor camera image (Canon 1Dmk III) taken from same distance with same lens as the first image.


1.6x sensor camera image (Canon 40D). Notice how this uncropped image is much bigger than the others ? this is because of the multiplication effect of the crop camera. If I wanted to crop the image down to a single butterfly, this image will require a much lighter crop - and will retain far more megapixels on the target than the equivalent full frame image.

Next, we need to open up the images on the computer and view at 100% on the screen.The images below are totally unsharpened crops and have undergone no post processing in Photoshop (apart from the usual RAW conversion process).


1Ds mkII full-frame sensor camera - *100% crop, unsharpened

1Dmk III 1.3x sensor camera - *100% crop, unsharpened


Canon 40D 1.6x sensor camera - *100% crop, unsharpened

*NOTE:

As the images from the 40D contain more megapixels than the other two cameras due to the high pixel-density, there is a problem in comparing apples with apples. I could have presented you with one large image from the 40D and two smaller images of identical (smaller) size from the other two cameras - as you will remember they share the same pixel density. Unfortunately this makes them rather difficult to compare, so I upsampled the smaller images and downsampled the larger image to the same number of pixels using Photoshop's interpolation tools. This only required a modest 22% increase and decrease respectively, so does not distort the results here. I have stared for ages at the non-interpolated files at maximum size on my computer screen but can still see no difference.

All other tests conducted throughout this tutorial will adopt this approach to better enable comparisons to be made.

TEST 1 CONCLUSIONS

There is absolutely no discernable difference between any of the three images to my eye. So despite the enormous handicap in terms of number of pixels, images from the larger format sensors appear to withstand much harder cropping than those from the 1.6x crop camera. Clearly the larger, better spaced pixels on the full frame camera can punch above their weight ! If the full frame camera had less pixels than the 16.7 Mp of the 1Ds mk II then it would struggle harder to keep up with the other cameras here unless the individual pixels were of even higher quality.

This all came as big surprise to me. I firmly believed that a crop camera would be more beneficial for telephoto work - such as photographing small birds than a full frame camera. I would mistakenly remove my 1Dmk III from my 500mm lens and put on the 40D if I wanted to get a bit more "pull". It appears from these experiments, that I would not get any benefit from doing this. To fill the frame more, it is necessary to either change to a longer lens, get a bit closer (not always possible) or add a teleconverter. Doubtless this will change as technology improves and I expect high-density crop sensors will soon be able to match the quality of today's full frame sensors. In turn the megapixel race will also continue on the large sensors too and these will probably hold massive amounts of small pixels packed onto the larger surface area.

TEST 1 CONTINUED

The results from test 1 were very surprising. The crop camera should have a big advantage in this test as in theory it puts far more pixels on the target area. I therefore set up another test to challenge my findings. This time I enlisted the services of my co-operative chimp. Here he is taking it easy on the garden furniture..

Chimp starting image taken with the 1DsII

This and all the test images was taken using the Canon 400mm f5.6 L lens, tripod mounted, using mirror lock-up and cable release. ISO 200, Aperture fixed at f8, shutter speed set by camera to around 1/250 sec. Images shot in RAW and individually matched for white balance etc in Photoshop CS3.

40D 100% crop at original size

1Ds mk II 100% crop interpolated to match 40D size

40D 100% crop interpolated to match 1Ds mkII size

CONCLUSIONS FROM THE REPEATED TEST

If you compare the 100% crops at the original size, the 40D has produced a larger image because of the greater number of the pixels on the area covering the chimp (as it has a 30% higher pixel density than the 1DsII ). On paper this should mean that the 40D has a big 30% pixel advantage and should beat the 1Ds mk II easily if sheer number of pixels is all that matters.

I find it difficult to compare the images from the two cameras because of this size difference. In some ways the 40D image looks a little better than the 1DsII image - but is it more impressive simply due to it's size alone - or is there actually more detail present ? If you look at the fur in the area between the eyes and the brow of the chimp, I think I can see more detail in the fur on the smaller 1Dsmk II image. If you compare the nose, I think there is a tiny bit more detail in the 40D image.

In the last two size-matched images above, I matched them again (as before in the first experiment) by interpolating up or down as appropriate to the mid point so both cameras will be equally disadvantaged. To differentiate them is really splitting hairs and pixel-peeping at the extreme. I would be happy to call this a draw again, but if you put a gun to my head and forced me to say which image is best, I would have to say that the 40D just has it. If I had to put a value on it I would say it is about 2% better - whereas the 40D should have a 30% advantage with all those extra pixels.

So, if you were in the field and you saw a bird that was a bit small in the frame, should you put down your 1DsII full frame camera and swap to a 40D for extra reach ? I would say not -
although this goes against the theory and also some other tests that I have seen on other websites. The full frame camera's individual pixels must make up in quality for what they lack in quantity (on target not in camera base megapixels). What would the story be with the 1Dmk III ? From the first experiment with the butterflies, I would suggest that it can still hold it's own OK.

Finally what about the 5D full frame camera ? It is a better match to the 40D in base megapixels (see the table again), but it has a very low pixel density (50% lower), however it also has a very high per-pixel quality. I can't answer that question as I don't own a 5D, but I've got a feeling that it might be a very close call once again.

Things should swing into the full frame 1Ds mkII's favour when you can get closer to the subject. and put all of it's 16.7 Megapixels on the subject. To find out if it does, you will need to read part two of this tutorial.