The importance of negative size has been discussed for more than 100 years. When Oskar Barnack designed and constructed his first Leica for 35 mm perforated film it was concidered a joke. It took many years before the 24 x 36 mm format was accepted.
Today we are comparing the so called full frame sensor (24 x 36 mm) with the smaller DX-size sensor (16 x 24 mm) and discussing pros and cons with the two. Looking into the professional market we find even bigger sensors like the "double-full frame" 36.8 x 49.1 mm CCD from Kodak and the slightly smaller CCD from Dalsa, both sensors beeing used in camera backs for mid and large format cameras.
In this article we are going to take a closer look at positive and negative aspects of different sensor sizes.
In order to make a fair comparison and keep as many factors as possible constant, we are going to look at three "fantasy" sensors not excisting on the market. They all have the same amount of pixels, 13,5 MP, the same fill factor and the same aspect ratio. We are going to use three cameras with normal focal length lenses.
1) 16x24 mm. Normal focal length = 29 mm. Pixelpitch = 5,3µm.
2) 24x36 mm. Normal = 43 mm. Pixelpitch = 8,0 µm.
3) 34x51 mm. Normal = 61 mm. Pixelpitch = 11,3 µm.
Optimized lenses
A camera lens of a certain quality has a resolution of a certain number of line pairs (lp). Please note that we are talking about a certain number of line pairs over the image circle, no matter the size of the circle. We are not talking about lp/mm here. A top quality normal lens of 29 mm for the 16 x 24 mm format can resolve the same number of line pairs over the sensor diagonal, as a top quality lens of 61 mm for the bigger sensor. The number of lp/mm is less than half for the 61 mm lens compared to the 29 mm lens.
Understanding this fact, it is easy to realize that it is better to design lenses for the smaller DX-size sensors of many of today's SLR cameras, than to use lenses designed for full frame film or sensors. There is a common misconception about using only the "sweet spot" of a lens for 24 x 36 mm, leaving the (sometimes) softer edges out. In fact there is a lot to win in designing lenses with an image circle exactly covering the sensor surface.
Same image size
13.5 MP gives an image file of 40 MB. The images from all three cameras will come out with the same file size. There will be nothing in the file size revealing what camera was used. The perspective and DoF (explained later) will be equal. If we make big prints from the three images, they will all be enlarged equally.
Depth of field (DoF)
In our ambition to make a fair comparison between the three formats we decided to stop down the lenses to the same DoF. We use f5.6 for the 29 mm, f8 for the 43 mm and f11 for the 61 mm lens. These settings will give approximately the same DoF. Stopping down the larger sensor with two f-stops means we need to compensate with two stops longer exposure time. This is a clear advantage for the smaller sensor.
On the other hand, it is impossible to take pictures with a very shallow DoF with a small sensor. Even with wide open aperture DoF will be considerable with a small sensor.
Diffraction
Diffraction is an optical phenomenon limiting resolution when stopping down the aperture. Click here for very good explanation of diffraction.
The smallest dot can not be reproduced smaller than the diffraction "airy disk" and the size of the airy disk is approximately the f-number in µm. Stopping down to f8, with any lens, gives a blurry airy disk of 8 µm. Stopping down a lens to a f-number beyond the pixel pitch of the sensor results in blurred images since the airy disk will cover several pixels.
For our 16 x 24 mm sensor the pixelpitch is 5,3 µm and it is stopped down to f5.6. The 24 x 36 mm sensor with a pixelpitch of 8,0 µm is stopped down to f8. And finally the big 34 x 51 mm sensor with a pixelpitch of 11,3 µm is stopped down to f11.
At the same DoF with appropriate f-settings, the diffraction problem is proportionally equal for different sensor formats.
Pixel size
Like we discussed in the article Dynamic Range and Pixel Size a large pixel can collect more photons and emit more electrons than a small pixel. The positive effect is better noise control, cleaner shadow details and more dynamic range. Advantage for the larger sensor.
Pixel density and camera movement blur
There is a discussion going on about pixel density and blur caused by hand held camera movement. The basic idea is that smaller pixels packed on a limited area will cause more blur when a hand held camera is shaken slightly.
Let's compare the DX-size with the double-full-size sensor. Shooting similar pictures with the two, involves a 29 mm lens with the small and a 61 mm with the big sensor. In order to get the same DoF we need to stop down to for example f8 with 29 mm and to f16 with the 61 mm lens. At the same ISO setting, the camera with the small sensor, with high pixel density, 29 mm lens stopped down to f8 and (for example) 1/30 sec, compares with the other camera with lower pixel density, a longer lens of 61 mm set to f16 and 1/8 sec. To me it is obvious that pictures from the lower pixel density camera will be much more blurred than the other one.