using full frame digital




Towards the end of 2005 I purchased a Canon 5D, my first full frame 35mm digital camera. I kept the 20D, which has a 1.6x smaller sensor, as a backup. Canon users have this choice of two formats, a choice denied to Nikon users. However since the launch of the 12mp Nikon D2X I don't sense much envy from the Nikon camp. The benefits of the the full frame format are being questioned and the small format ( 1.5 or 1.6x ) cameras are taking almost all the market.

Whatever the trends, I was excited about the arrival of my 5D. This camera is about the same size and weight as the 20D. The only major outward changes in the 5D are the viewfinder, which of course has wider coverage corresponding to the sensor size; and the absence of a popup flash. The camera takes all my lenses apart from the Canon EFs10-22, but the field of view and image quality of the lenses changes with the larger format.

First the bad news. Shooting for maximum reach, the 20D has higher pixel density than the 5D and can therefore render more detail. Below is a comparison made with the Canon 100-400IS at the long end. The test target ( the lens's own box ) was the same distance from the camera in both shots.

 Reach with the 100-400IS  -  100% unsharpened crops from 20D (top) and 5D (bottom)

The image widths of these two cameras are 4368 pixels (5D) and 3504 pixels (20D). However the 5D sensor is physically 1.6 times larger. From which you can work out that the pixel density of the 20D is 1.28 times higher than the 5D. Put another way, the 20D provides a 1.28 times enlargement compared to the 5D ( but only of a centre crop of the image ). The advantage of pixel density is evident in the samples. Actually I was quite surprised at how well the 20D did in this test. In the field I have often found that the 20D is soft at the long end of the 100-400IS, whereas the 5D is reliably sharp. The lesson is that the 20D needs to be held very steady and focused carefully to deliver the extra resolution.

For wildlife photography reach is all-important. But for most applications you can zoom the lens, or move the camera, to fill the viewfinder with your subject. That leads us to a second test. Here I use a Sigma 105mm macro and move the cameras to get the same field of view. The jump from the 20D's 8mp to the 5D's 13mp is quite a large one. In prints it means greater sharpness at A3 or above.


 Detail with a macro lens -  100% unsharpened crops from 20D (top) and 5D (bottom)


The best thing about full frame for me is not the resolution though. It's about seeing the whole image produced by the lens. I didn't like the thinking behind small-sensor cameras - that because of the cost of sensors, 60% of the image area of 35mm lenses is blocked out. With a FF camera the lenses I choose for my bag are different, but they are better aligned with the products on offer. So I can get away with a couple of primes for landscape (50mm and 35mm) if I want to travel light. These lenses are cheap, lightweight, fast and free from aberrations. The laws of optics make it easier to design lenses with a 'normal' field of view, and this is the view we are most likely to want  to capture.

I haven't found any issues with edge sharpness of the primes lenses ( I use Canon 35, 50 and 85mm ) on a 5D. Actually most of the zooms are pretty good as well, even the superwide 17-40L. I had been worried about this lens on FF after seeing some bad samples on the internet. I have seen problems with vignetting  though, when I am stitching panoramas. Changes in the brightness of the sky towards the corners can make stitching difficult. Photoshop camera raw has a tool to deal with vignetting but I haven't found it fully effective. However I still use stitching quite often. By combining two or more frames from a good prime lens I can get amazing wide-angle quality.

 Audley End, Essex. Two merged frames from a Canon 35mm lens and 5D


Depth of field (DOF) is cited as an advantage for both the small format and for FF. Larger formats do produce more background blur, other things being equal. The test shots below are taken with a Sigma 105mm at f5.6 and wide open at f2.8, on a 20D (top) and a 5D (bottom). I have moved closer with the 5D to get the same field of view.

The 5D gives noticeably more blur than the 20D. It should be equivalent to a factor of 1.6x aperture.  I have been using the 70-200IS for portraits and even with a 20D it's possible to isolate the subject nicely in most situations. So I didn't get the 5D for DOF control. But it's fun to open up a fast lens and really obliterate the background. for photography