Today's advertising would have us believe that megapixels are exclusively responsible for the quality of digital cameras, but there is more to technically good-looking images than just raw pixel count. Optics and electronics play a crucial role, and shortfalls in either are responsible for two of the most recognised and disliked digital artefacts: noise and purple fringing.
Electronic noise covers images with masses of tiny coloured specks, which are particularly apparent on what should be flat or smoothly shaded areas such as tones or blue skies. The situation gets worse as sensitivities are increased under dim conditions, or with the smaller sensors used in compact cameras.
The dreaded purple fringe effect produces garishly coloured edges to subjects placed against high-contrast backgrounds. This is a combination of chromatic aberration, where a lens fails to converge all wavelengths of light at a single point, and blooming, where the light-sensitive elements on the sensor are overloaded. The result is coloured fringes around high-contrast areas, which may be green or red, but most commonly a pinky-purple. While some cameras suffer less than others, all are susceptible to a degree. Fortunately, there are software tools and tricks that can reduce noise and purple fringing.
Noise reduction is a case of smoothing the image without losing too much detail or sharpness. While you could experiment with filters, we recommend shelling out the £20 or so to download a copy of the amazing Noise Ninja from www.picturecode.com. Removing purple fringing, however, may look an unsurmountable task, but it is simply a case of isolating the offending areas and removing their colour using Photoshop. Here is how it is done.
Big chips vs small chipsSmaller sensors suffer from higher levels of electronic noise. This is down to the reduced surface area of their individual light-gathering pixels. Bigger pixels gather more light than smaller ones, so require less electrical amplification, enabling most noise to remain hidden.
The problem today is that many sensors sport increased resolutions without being physically any larger. This means their pixels are smaller so suffer from higher noise.
Most modern five-megapixel cameras employ a so-called 1/1.8in sensor, which actually measures just 7.18 x 5.32mm. This is the same physical size as the sensors used in the first four- and even three-megapixel cameras. Clearly, the pixels have become much smaller, so noise levels are higher.
This is the big advantage of digital SLR cameras with larger sensors. Take Canon's 300D, which has a six-megapixel sensor measuring 22.7 x 15.1mm. Many owners complain that because its sensor is smaller than a frame of 35mm film, their lens focal lengths are effectively multiplied by 1.6 times. But compare the surface area to a consumer camera and you will find that it is almost nine times larger with roughly the same number of pixels, meaning much lower noise. A digital SLR will produce less noise at 400 ISO than a consumer camera does at 100 ISO.
As a final example, take the latest eight-megapixel prosumer cameras. These have a 2/3in sensor measuring 8.8 x 6.6mm. This is the same size as the first five-megapixel models, which means smaller pixels and higher noise levels. To help counteract this, most have a super-slow 64 or 50 ISO setting.
Aperture and fringingAs explained in the introduction, purple fringing is often the result of imperfect optics. A camera lens must bend incoming light to converge at a focal point. Light gathered towards the edges of a lens must be bent the greatest and is, inevitably, more prone to error. One way to maximise optical performance, though, is simply to close the lens aperture by a stop or two, thereby allowing it to disregard the extreme edges. This can eliminate or at least reduce purple fringing on many cameras.
Author: Simon Aughton
Reducing the effects of purple fringing