How Grey Is Mid-Tone Grey?

I HAD NOT EXPECTED to create such a furore among readers from just mentioning camera calibration. It seems that I have inadvertently begun a topic that many of you want me to continue with.

Janet says: ‘Please talk us through the graph,’ and Martin asks: ‘What the hell is mid-tone grey anyway?’

Well, mid-tone grey is the term given to the colour of the industry standard 18% Grey Card, first produced by Kodak, to which all light meters are generally calibrated. Whatever you point your light meter at, the read-out you obtain will be expressed in terms of the amount of exposure required to reproduce the object measured as a mid-tone grey.

The 18% description of the grey card is misleading, because it describes reflectance – so photographers now tend to refer to 50% luminance instead.

In the RGB colour space, mid-tone grey has a value of 118, 119, 121 – a fact which enterprising readers can use to print their own grey card (if they are prepared to put-up with the small inaccuracies introduced by different types and grades of paper).

It is important for the photographic and print industry to have a standard by which everything can be measured and, as a photographer, mid-tone grey is the most important standard you are likely to run across because it defines what is the ‘correct’ exposure for any scene. Once mid-tone grey has been defined, all the other shades, from pure black to pure white, can be described in terms of it – and we can arrange to speak in terms of EVs (Exposure Values).

Few scenes, which you are likely to photograph, consist of a single tone. Individual objects absorb different amounts of light and reflect what they do not use back to the light meter – and it is those different levels of light that your camera’s film or sensor records to produce an accurate image of the scene.

Different levels of light in a scene can be described as different EV levels – with EV0 being given to the scene’s middle tone; positive EV numbers given to brighter levels; and negative values given to darker areas. EV1 is twice as bright as EV0; EV0 is twice as bright as EV-1 – and so on. The distance between one EV number and the next represents a doubling (or halving) of light strength (and is equivalent to the doubling or halving of exposure by changing a camera’s aperture by one full stop).

In a daylight scene, such as a landscape, there can be many hundreds of different light values that can be distinguished by the human eye; but no film or camera sensor can capture that full range (in a single shot). In digital photography, accurate scene representation is limited by the RGB colour space, which only has a range of 10EV (values: 0, 1, 2, 4, 8, 16, 32, 64, 128 and 255). That, however, is not the main problem facing photographers; because electronic sensors can only accurately record a much more limited range of around 8EV. (Cheap sensors may only be able to reproduce much less).

If you look at the inset graph, produced from calibrating a Nikon D40, you can see that the horizontal EV scale runs from –7EV to +7EV; but the only exposures that result in a sensor image lie between –5EV and +3EV.

If you examine the graph further, you should see that those elongated Ss (which each represent a grey scale image) become increasingly horizontal the lower they appear on the main curve. The more negative the EV (under-exposure) the fewer tones that can be captured by the sensor. You can measure the tones recorded by any S using the vertical Level scale which records the (single) RGB value.

(Although misrepresented on this graph – more later – the same ‘drop-off’ and compression is exhibited as exposure is increased).

Taking the green curve (representing the results at ISO 3200) we can see that the straight part of the slope at –1EV records grey values of between around 90 and 110; but at 0EV its range is extended by 50% to between about 135 and 165.

But the main point to note is that all the exposures, in this graph, are ‘wrong.’ The only mid-tone greys that have been captured lie on the 1600 ISO and 800 ISO curves at EV0 – and they are both located on their S’s shoulder, where any additional exposure will quickly degenerate the mid-tone into a much lighter shade.

It should be pointed-out that we are not calibrating the Nikon’s internal meter here. Rather, we are calibrating the Nikon’s exposure to a Sekonic incident light reading – so there is bound to be some discrepancy. (The difference here is around .5 EV in the Nikon’s ISO sensor ratings which are, relatively speaking, over-rated – if you accept that the Sekonic is correct).

The anomaly serves to illustrate the importance, for all serious photographers, to properly calibrate all new equipment before attempting that unrepeatable shot.

So what does this graph generally tell us? Well, first of all, that image tones are only accurately recorded by the film or sensor on the straight portion of the ISO slope. Furthermore, that getting the exposure ‘wrong’ moves the mid-tone to a new ‘mini slope’ in which fewer tones can be represented. It is the latter characteristic, emphasized when considering a single ISO slope, that is replicated between different ISO settings.

For professional photographers reading this: I now admit that the graph I am using is a bit of a con. (You don’t normally calibrate a camera with an integral light meter to a handheld without first calibrating both meters to ensure they produce the same mid-tone reading. That is why there is an apparent inability of the Nikon to reproduce an EV0 mid-tone here.) But I have chosen to use this graph for a purpose: not only does it illustrate that each ISO has a different tonal range (each ISO’s S would be adjacent to similar EVs and not offset from each other if that were not the case) – it also illustrates that different manufacturers appear to use slightly different benchmarks when calibrating their meters. The reason why TTL metering is nearly always off to a handheld meter is that the camera’s sensor is only reading what level of light is reaching it through the lens. (And all lenses absorb light to a greater or lesser degree).

If you use a professional camera (and a professional handheld meter) you will find that both provide a facility for calibrating one to the other. The standard adjustment range is 1EV in tenth steps, which is more than sufficient for any good equipment.

There is no need to take photographs when calibrating meters: just use both to take readings from an evenly lit standard grey card and adjust the one you know to be ‘incorrect’.

The other facility that most good camera manufacturers provide is an exposure compensation dial, which normally provides the ability to adjust metered exposure between –5EV and +5EV. It is not a coincidence that this range is the same as the range used to calibrate a camera’s actual exposure; because using it provides photographers with the ability to easily calibrate their camera’s automatic programs as well as its manual settings.

Camera calibration allows you to arrive at suitable adjustments, if necessary, to be applied to the TTL meter’s reading for each ISO sensitivity (and accurately place the mid-tone at 50% luminance). Keep a durable record of the adjustments required and dial those into your camera’s settings, as necessary, when using different ISOs.

In my reply to Mac’s last comment, I said that selecting the correct ISO was also all about mood – and, if you take a look at the different results of taking a sequence of calibration grey card shots, you can better see what I mean. Low ISOs tend to produce a light and airy mood; high ISOs dark and sombre depression.

Often, mood can be important – as beautifully expressed in Harry Borden’s portrait photography. Harry can make a baby appear positively angelic by exposing for the highlights – and make a politician appear positively menacing by concentrating on the shadows.

And that brings us nicely on to the subject of clipping-points…

Clipping points are those EVs in which the last vestige of any detail is still represented – before all is lost in pure black or white. They are located where the ISO slope begins to heel and to shoulder.

So how do you produce an ‘angelic’ baby portrait? The answer is to meter the brightest white in the scene and accurately place it on the ISO’s shoulder – allowing the shadows to take care of themselves. In other words, if the ISO’s shoulder clipping-point is at +2.7EV, use your TTL metering to set the brightest white at mid-tone grey – and then rack the exposure up by 2 2/3 stops.

To produce that menacing portrait, or gloomy scene, perform the same exercise for the darkest shadow and place its mid-tone reading on the clipping-point you have established on the ISO’s heel by under-exposing. (Let the highlights take care of themselves).

And no, you cannot reproduce those precise effects in Photoshop from a standard exposure; because you cannot reproduce the subtle logarithmic tone compression that occurs towards both clipping-points.

Would you ever place a mid-tone reading past the clipping-point? Yes. When producing knockouts (when a commercial product shot requires a pitch black or pure white background).

But then you need to know what ISO to use – to ensure all the tones (and colours) in the product are reproduced accurately. Manufacturers tend to invest a great deal of money in patenting the colours used in their logos; and, if you don’t reproduce them accurately, you will be unlikely to get paid or offered any more work.

And here endeth the lesson on photographic exposure and camera calibration. I am supposed to be on holiday 🙂.