Apart from the need to make the equipment waterproof, the major differences between photography in water and photography in air are due to the fact that that water absorbs and scatters light. Human vision adapts automatically to this change of environment, but camera systems require help and persuasion if they are to produce good results. The specific problems which have to be addressed in this respect are as follows:

Water absorbs light across the whole of the visible spectrum. Water may look clear in a drinking glass, but light absorption over distances of several metres is significant. Hence the underwater environment is dark, artificial lighting is needed in most circumstances, and flash and video lights are not as effective as they would be in air.

Water absorbs light at the red end of the spectrum more strongly than it absorbs light at the blue end. Water may look colourless in a glass, but a glance at a swimming pool shows that it is a pale cyan-coloured liquid.

Water usually has particles suspended in it; from small solid particles, organisms, and gas bubbles, which give the impression of fog; to large fluffy particles which give the impression of a blizzard. The brain processes visual information to enable us to see through partial obstructions, but the processing relies on stereo vision and movement and is much less effective when we look at two-dimensional still images.

These additional factors change the rules considerably in relation to flash intensity requirements, working range, white balance, and lighting position. It might also be said that 'lighting technique' not only encompasses the use and positioning of lights, but the image adjustments which can be carried out after the event in order to restore colour balance and contrast. With the latter in mind, those interested in obtaining professional quality results should use cameras (or film scanners) which can provide output in a 48 bits-per-pixel 48bpp) format (e.g., RAW), the point being that an image file with 4096 brightness levels for each primary colour (rather than the usual 256) can be subjected to large adjustments without significant quality loss.

Reduced Guide Number:
Flash illumination underwater no longer follows the inverse-square law on which 'in air' exposure guide tables are based. At a flash to subject distance of about 1 metre, the guide number is reduced to about a third of its 'in air' value. At greater distances, the guide number is reduced even further, and only when the light source is very close to the subject does the guide number begin to approach its 'in air' value. Nowadays, we don't worry about guide numbers too much, taking care of exposure by looking at the image histogram or by using TTL flash, but it means that a large flash unit is needed if an appreciable range is to be obtained.

Colour Cast:
What is meant by 'appreciable range' may also come as a surprise, the issue being that clear water acts as a strong cyan-blue filter. We need to remember here that white light is a mixture of all the colours of the spectrum. The problem is that the different colours are not absorbed equally by water. Red is absorbed particularly strongly, leaving pictures taken at intermediate range with a cast of the complementary colour, i.e.cyan. For very long range (10m or more), both red and green are absorbed and the only colour substantially recorded is blue. In the latter case, little or no colour information is present in the picture, and the best presentation may well be obtained when the colour saturation is turned down to zero and the picture is reproduced in black and white.

In clear water (free from green algae or brown mud) the red component of white light is reduced relative to the blue component by about 1 EV (i.e. halved) for every 2.5m of light path. Since the light has to travel from the flash unit to the subject, and then from the subject to the camera; if the flash is aimed from the vicinity of the camera, a photograph taken at a distance of 1.25m has its red component reduced by the equivalent of 1 f-stop. In the old days of direct-projection transparency film, this meant that the maximum flash illumination range was in the order of 1 to 2m, and the strong cyan cast was taken to be a fact of life. Compensation was possible at the printing stage, and nowadays, we can adjust the image data to pull out the cast; but electronic noise becomes a problem if the gain of the red channel is adjusted too far, and so the trick in producing clear colourful pictures still lies in working at short range.

Backscatter:
A particular problem arises when using flash in cloudy water, and is exactly analogous to that of trying to use full-beam headlights when driving in fog. Light is scattered back by the suspended particles, and the subject can sometimes hardly be seen at all. Natural water is always a little cloudy, and swimming pools are only clear when properly maintained*; so if high clarity is required it is important to check the state of the water and keep extraneous swimmers and divers away from the location for a few hours so that particles can settle. For poor visibility which cannot be avoided, the problem is reduced by aiming the camera and light source from different directions (the point being to light only the subject and not the suspended particles between the subject in the camera); and by working at very short range (i.e., by using wide-angle and macro lenses). Remedial action at the post-processing stage involves adjusting the image black-point (Photoshop: levels) to crush-out the general background fog and increase contrast; and using the cloning stamp to remove any distracting white spots.

* Sand filtration alone is not enough. Perform a good backwash a few days before a shoot and add some flocculating agent (e.g. Aluminium Sulphate) to the input side of the filter.