Hints and Tips

A Case for Compensation?

It is a well known fact that many of the most dramatic and successful images are the result of unusual and often extreme lighting conditions that would have most camera exposure systems unable to choose the required exposure themselves, simply because, brilliant as they may be, they are (as yet) unable to read the photographer's mind. Is this a case for suing the camera manufacturer for compensation, or applying a little yourself?

Whether you select spot, centre weighted or integrated metering, all a camera can do is to evaluate the average light intensity of the subject areas sampled. Maybe with an integrated or centre weighted system, greater weight is placed on certain areas of the image so that the average intensity calculated by the cameras computer is pushed a little in one direction or another.

A centre weighted system does what it says and records average intensity in the central area of the field of view. Hence bright or dark fringes are ignored, the idea being that what you want to expose correctly lies near the centre of your picture. Such a system is often fooled by a large area of bright sky so that foreground tends to be under exposed and the sky slightly overexposed - not the best scenario. Some centre weighted systems are offset towards the bottom of the screen to reduce such problems.

Integrated systems have the field of view split into sections, often five to ten or so but sometimes as many as forty sampling areas. The cameras onboard computer can choose priorities for any combination of these sampling areas, often based on pre-programmed typical situations, such as the classic bright sky, dull foreground conditions. Such highly complex systems can reduce the percentage error rate when used on 'auto' by inexperienced photographers. But they may still be fooled at other times. And what happens if you decide to turn your camera 90 degrees and shoot in portrait format?? This can play havoc with off-centre weighted systems and integrated systems (very few of the latter can be switched to portrait mode).

Spot metering systems usually have a central field diverging from the camera at an angle of 5 - 10 degrees. The spot coverage is sometimes indicated by a central circle in the viewfinder, making it much easier to be sure that sampling is taken from one area of the subject or another. As we shall see, uncompensated spot metering only gives accurate results if readings are taken from a mid-toned (18% grey) part of the subject.

Thus however achieved, using the computed average light intensity, the camera selects an exposure for you. If fully auto it will also select an appropriate shutter speed and aperture that hopefully meet your needs. If using Aperture or Shutter priority the camera selects a suitable shutter speed or aperture. If full Manual is selected the camera indicates by needle or graphic display when your manually set combination of shutter speed and aperture is correct.

These settings are intended to make the brightest tones near white and the darkest tones near black and the average tones around 18% grey (a grey tone that reflects 18% of the white light falling on it), or their coloured equivalents. It is most important to remember whether you are using centre weighted or integrated metering, as in these cases most or all of the subject is sampled. In the case of spot metering, only a small area is sampled and this can greatly alter the way your camera, or you yourself, interpret the meter readings. It is also important to bear in mind that all recommendations in this article presume that you are METERING THROUGH THE LENS (TTL). Hand metering has similarities but also important differences. This subject will be the topic of a later article.

We must also be aware that contrast range, between the lightest and darkest tones in many sunny scenes with dark shade in places, is more than films and digital sensors can record (see fig.1). Consequently some light areas may be over exposed and/or some dark areas may be under exposed. We say that the highlights are burnt out and the shadows blocked up - a thing most judges penalise. The average b/w or colour negative film may cope with a brightness (contrast) range of about 6-7 stops, colour positive slide film covers only a 4-5 stop range. Remember each increased aperture stop represents a doubling of light intensity so that 6 stops covers brightness ratio of 2x2x2x2x2x2 = 1:64. Hence the lightest tone the film can record can only be about 100 times brighter than the darkest tone. Our eyes are able to distinguish contrast ratios of up to 1:500! Little wonder that the scene we look at is often very different from the one recorded.

What can be done to offset these deficiencies in the sensitivity of our films and sensors? For close-up work we may be able to shade or diffuse the light falling on the subject, thus reducing contrast. For more distant landscapes we might try using neutral density filters, graduated or not, to reduce contrast between sky and land, or even try some fill-in flash for middle distance or closer subjects. If contrast is still a problem then all we can do is to make a choice to deliberately allow the highlights to burn out or the shadows to block up depending on which is more important to the image. Usually blocked up shadows are preferable to glaring, white highlights.

We may well decide to come back when the light is less contrasty, as in the early morning or late evening. But we may be moving on, and our only chance of any record is during a period of highest contrast (sunny midday) and we are forced to make the choice between highlight or shadow detail. To retain highlights we compensate by giving less exposure and in order to retain shadows we give more exposure. Maybe we need to bracket between half and two stops in half stop increments if using b/w or negative colour stock or digital sensors; for colour slide film, bracket in one third stops. Then again we may want to produce a dramatic shot with silhouettes against the light (less 1-2 stops or meter off the brightest area) or a high key image bathed in sun (plus 1-2 stops or meter off the darkest area).

Only the photographer can make these choices and must step in and reduce exposure to favour the highlight details or increase exposure to bring out shadow details. This is done by means of the compensation settings that most cameras possess or by manually changing aperture or shutter speed.

Compensation allow us to manually increase or decrease the camera's computed exposure, usually by up to 2-3 stops in one third or one half stop increments. Another compensation method is to manually reset the speed rating of the film, lower for more exposure, higher for less exposure.

As with all manual adjustments it is important not to forget such settings have been made when moving on to new subjects.

Pub at Arlington, Sussex. Deryck Ford

In many ways an 'average' scene, but in order to stop the white walls 'burning out' some blocking of the shadows was inevitable, given the bright, sunny conditions.

The graph above represents scenes where there is an even spread of light and dark tones. In practice most scenes will have some degree of bias towards dark or light tones that will skew the graph to one side or the other, as shown in figs 2 and 3

In such circumstances it may be necessary to make exposure compensations so as to bring the peak of the graph back into the central 18% grey area on the film or sensor. This can be done by increasing exposure in darker scenes or decreasing exposure in lighter scenes. The 'integrated' exposure systems of some cameras may endeavour to do this automatically, some may not. Only trial and error can inform you of the degree of manual compensation required for a particular camera in any particular set of lighting conditions. Grey card readings cannot respond to such tonal imbalances and may be misleading as a result.

Bias towards a light scene or image - possibility of losing some highlight detail. Reducing exposure improves shadows and highlights.

Bias towards a dark scene or image - possibility of losing some shadow detail. Increasing exposure improves shadows and highlights.

When working in colour it should be noted that histograms are normally a combination of the tonalities of red, green and blue in the image (RGB). Often more information can be obtained by switching to the histograms for individual colours, although this is not needed for basic compensation calculations.

So far we have only considered subjects where there is a fairly even spread of light, dark and middle tones. Looking at Fig.1 the now familiar histogram of image tonal content (as seen in software such as Adobe Photoshop and in many digital cameras). This would be a symmetrical or 'normal' curve, with a few near white highlights (reflecting nearly 100% light), and a few darkest shadows (reflecting almost no light), but mostly mid-tones centred around the 18% reflectance level.

The kind of thing most photographers would easily recognise as an "average" scene. For 18% grey card fans, this is the kind of scene where the card can produce accurate exposure readings. It places your exposure in the middle of the curve (fig1) so that both light and dark areas are given equal weight and a good result can be expected so long as the light falling on the card is the same as that falling on your scene and the contrast range in your scene does not extend beyond the sensitivity of the film or sensor. On a bright sunny day with dark shadows the latter is most likely and you may have to make compensation choices (see above) about which your grey card can tell you nothing!

Now suppose we do not have an even spread of tones. Perhaps a predominantly dark scene with just a small area of near white, for example, a sunlit white swan on a mostly dark, shadowed lake. We now have anything but an "average scene" (see histogram fig.4). We might equally well be trying to snap a black swan on a white frozen, snow covered lake fig.5. In the first case your camera, swayed by the massive dark background, would be screaming "open up, open up"!! …. And of course, if you let your camera have its way, the white swan would be hopelessly over exposed. Or with the black swan with its white background the camera would be chanting "close down, close down"!! …. And if again you capitulate to your camera, your swan would be seriously under exposed. Only you know that the small area of white swan or small area of the black swan is what you really need perfectly exposed in each case. So once again you need to step in and give your camera the lead. Compensate by perhaps as much as 1-2 stops less for the white swan and 1-2 stops more for the black swan.

Now if you are using the spot metering setting on your camera only a small area of the subject may be sampled (typically a cone of 1-5 degrees subtended at the camera). In the case of the two swans above, you may be able to get near enough to be sure that you are only metering off the black or white swan feathers. Some cameras give a central ring in the viewfinder to indicate spot metering coverage. In these cases the camera metering system ignores the background and simply indicates the exposure required to produce a mid-tone (18% grey) from each swan. If you follow the camera's advice your white swan will be an under exposed greyish white and the black swan will be an over exposed greyish black. Both should have detail but neither will show their true tone in the scene.

Hence with spot metering the compensation action is reversed; plus 1 stop for the white swan, minus 1 stop for the black swan; the opposite compensation to when using centre weighted or integrated metering modes. This apparent anomaly lies at the heart of much confusion and bad advice concerning exposure compensation. Remember, how you compensate depends on the metering mode used, and whether or not you are metering only from the key area of the subject and are ignoring the background. Also note that in dull conditions less compensation (either way) will be required, than in sunny or bright conditions.

One should also be certain of the angle of sampling of spot metering systems. Built in SLR systems tend to have much wider sampling angles than purpose made, hand-held spot meters.

An alternative technique can be tried using an 18% grey card. If you take a spot reading off the card it should give a 'mid range' reading that will be close to the compensated spot reading of each swan - provided you can be sure that the card is in the same intensity of sunlight as each swan!

It is an easy matter to test this out by trying both metering techniques and comparing results.

Spot metering is also useful for measuring the tonal range in a scene. Taking readings from brightest subject areas, mid tone areas and darkest areas, will tell you if your are beyond the range of your film or digital sensor.

This type of metering is difficult or impossible using camera TTL techniques. The meter or camera is pointed directly at the same light source as is reaching the subject. Before reaching the meter sensors the incident light must be diffused, usually with some form of translucent white plastic cone (Weston meters used an inverted cone). Accurate readings would be difficult if pointing the camera lens towards the light and as far as is known camera diffusers are not made. Unless light is well diffused, the lens could act like a burning glass and severely damage film or digital sensor. If a separate hand-held meter plus diffuser is used, the advantage of incident readings is due to the fact that the reflectivity of the subject is irrelevant and unable to distort accuracy in the ways described above. Many photographers retain a hand meter with its diffuser cone accessory for this reason as a check on what camera exposure systems are indicating. Exposure compensations are not appropriate or required if incident light metering is used.

There are a number of small compensations that apply in certain circumstances and relate to the nature of the film source. In general negative film should be exposed for the shadows rather than the highlights, i.e. slightly over-exposed. This is because the least dense areas of a b/w or colour negative will be the shadows, which if too dark create almost clear negative film with no detail. The opposite is true for positive stock such as slide film which should be slightly under exposed. In this case overexposed light areas will quickly loose detail becoming almost clear film and producing burnt out highlights. Slight underexposure also improves colour saturation in slide film. In the case of negative film stock there is generally more exposure latitude (up to three stops) and general compensation may be up to half a stop under, whilst with slide film which has an exposure latitude of only half to one and a half stops, a general decrease of one third to half a stop is all that is required. Many photographers 'build in' these general compensations by rating the films above or below the 'official' figures. For example, Velvia users up rate from 50 ISO to 60 ISO, overriding any DX auto coding, where this is possible.

General compensations for digital cameras, at the present time, can only be established by trial and error with each individual camera. If the manufacturers have done a good job there is no obvious reason why any general compensation should be needed initially, unless experiment shows that colour or noise levels can be improved by small exposure changes.

The compensations suggested in the table should hold good where centre weighted or spot metering is used and assume that the camera's integrated metering system is not programmed to make some allowance for such special situations. The most common built-in adjustments of integrated systems are for a typical light sky, dark foreground scenario. Even so most systems are defeated by large differences from a normal range of tones as portrayed in Fig. 1. Just how much compensation is needed over and above built in compensation can only be discovered by trial and error.

If you are still in doubt concerning the application of compensation in the different metering modes, a digital camera can be used to try things out without even taking a single shot. Simply apply the compensations suggested for the situations listed and inspect the digital display of the camera. The effect of applying compensation can be seen straight away; something that can only be tested after your films have been processed if you use a film camera. I have used my own digital camera to test in this way and the results are very reassuring!

Another quite different approach to solving problems related to lens and film/sensor limitations is to take two or more tripod shots at different settings and merge the 'best' parts of each image in an image manipulation programme such as Photoshop.

If you have a higher contrast ratio that your system can record try taking two shots one compensated for the highlights the other for the shadows. The two images can then be merged in Photoshop without any trace of a join (it needs practice - but so did dark room work!).

Take one shot focused on the nearest parts of your subject and another focused on the more distant parts then merge the two images together to get an overall sharpness that stopping down cannot achieve. You can also afford, in such circumstances, to open up the aperture a bit to allow higher shutter speeds, if these are useful.

Carefully set up your tripod with a spirit level. Take two or more shots with about 20-30% overlap by swinging the tripod head round. Then you can 'stitch' the separate shots together using a 'stitching program' obtained free off the internet. Complete 360 degree panoramas are possible but perhaps more modest increases in horizontal (or vertical) coverage are most effective.

Church at Rodmel. Deryck Ford

In this shot the relatively large amount of bright sky could easily fool an integrated exposure system into stopping down and under exposing the foreground. If using centre weighted point the camera down slightly when taking a reading then holding the shutter button half down recompose and fire. Alternatively, take a spot reading off the mid toned grass, or church wall, or 18% grey card.

Deryck Ford

View or download this article in Adobe Acrobat format (1.4Mb)