In the first of an occasional series, photographic journalist and technical author, Joel Lacey, suggests that the new all-singing, all-dancing supermarket digital cameras may not be all that they are made out to be...
Churchill may have been referring to statistics when uttering his famous quote about lies and damned lies, but there is now an even more pernicious numerical assault taking place… at a supermarket near you.
We are, so the supermarket buyers think, all obsessed with numbers; ever-bigger, ever more impressive numbers. Especially when those numbers are bigger than other numbers on products you may be familiar with, even if the names of the products are not.
Let’s imagine a digital camera that has a 8MP CCD, 16bit image processing, and a 28x total zoom. And all for under £100. Sounds great doesn’t it?
The only trouble is that the camera mentioned would produce results in most lighting conditions that were not quite as good as a 3x zooming, 3MP, 8 bit per image processing Brand-you’ve-heard-of model from 2002. In low light conditions the supermarket special may well not produce results as good as sketching the scene with crayons.
But how can this be? Well let’s look at the numbers that really matter.
The actual effective number of pixels in a camera are important… all other things being equal. But all other things rarely are equal. Let’s take size a an example: specifically sensor size. Like most things, a good big’un will beat a good littl’un.
Sensor size is normally given for some unknown reason in inches.
For example, a relatively big sensor for a compact camera may be 1/1.8inch in size. What does that mean? Well it actually means a sensor is 7.18x5.32mm in size – that’s about the size of a man’s well-pared little fingernail.
On this sensor, there are, say, 8 million pixels arranged 3264 pixels wide by 2448 down. Each pixel site is (on the incorrect assumption that there is no gap between adjacent pixels) a maximum of 2.2micrometres wide by 2.2micrometres down.
Just to put this into mind-blowing perspective, find the finest human hair in existence, cut across it and the area of the cross-section of that finest human hair in existence is 1000x bigger than the area of each pixel site of our 8MP 1/1.8inch digital sensor.
If we compare that size to an SLR with full frame 35mm sized sensor (36x24mm) of the same 8MP resolution, then each pixel site of the latter would be 22x bigger than the compact’s CCD.
Shot taken with Canon Ixus V3 red area enlarged.
Shot taken with Supermarket special red area enlarged. Notice the artefacts on the enlarged example.
If we assume the base sensitivity of the two chips is the same, the smaller chip would need 4 1/2 stops more exposure than the larger chip in order for it to receive the same amount of light. That’s the difference between a shutter speed of 1/125sec for our DSLR and 1/6sec for our compact at the same aperture and in the same lighting to get the same exposure. Which of those two shutter speeds would you feel happier trying to hand-hold?
Now what if our compact’s chip is not even as big as 1/1.8inch? A 1/3.2inch 8MP CCD would require a further 1.5 stops of exposure (down to 1/2 second now) than our full-frame sensor DSLR.
Let’s take that 1/2 second shutter speed and take into account the 4x optical zoom (a focal length range that is equivalent to a 38-152mm). It has a maximum aperture range of f/2.8-5.6. So at the long end of the zoom we are getting in only 1/4 of the light (two stops less) that at the wide end, which means we’re now down to a two second exposure time.
And this at a magnification where it only requires a third of the movement to see the effects of camera shake as at the wide end. For good measure, the manufacturer has added the lunatic idea of a 7x digital zoom. We’re now getting 21x as much camera shake as we were at the wideangle setting. By this stage, it wouldn’t frankly matter whether we were using 16 bit per channel image processing, 8 bit per channel or frankly finger-painting from memory.
MEMORIES ARE MADE OF THIS
The great drawback of 8MP or 10MP image files (no matter how awful they may be when you actually get to see them) is that they slow the whole operation down. A good 3MP CCD model (say the 2002 Canon Ixus V3) will deliver A4 prints of impeccable quality with a 1.5cm white border around them. It will do so quickly, and you get loads of images on a card that can be processed quickly even on older computers.
A badly designed 8MP or 10MP supercompact will deliver slow-to-process results that are noise-ridden, blurred by camera shake, or badly affected by the effects of image stabilisation systems, take immense amounts of room on your hard disk.
LESS REALLY IS MORE
So next time somebody tries to impress you by quoting massive numbers at you, they are either trying to sell you a pup, or they’ve bought one themselves.
If someone is waxing lyrical about their specification, ask to see their pictures, preferably their hand-held low-light shots.
And if you’re after a bargain compact to carry around, why not try to pick up a decent secondhand compact from a manufacturer you’ve actually heard of?
About the author
Joël Lacey is the former editor of What Camera and Total imaging and has written on many different aspects of technology and imaging for national and regional newspapers, as well as a host of magazines including MacFormat, What Digital Camera, T3, Homes & Gardens, Amateur Photographer, New Scientist and Motor Boats Monthly. He was shortlisted for the PPA (Periodical Publishers Assocation) Specialist Writer of the Year and is the award-winning author of four books on photography and imaging. Visit his web site here Joël Lacey
Do you have an opinion about Joel's view? Go to the forum thread here Lies, damned lies and Megapixels and air your views