Color reversal film exposure is rather critical in comparison to other film materials. Aside from the limited possibilities for corrections after exposure there are several reason for this. Slide film has high contrast and hence small changes in the exposure will result in noticeable changes in the image tones. The amount of dye in the slide determines both, optical density and color saturation. Due to this interdependency the exposure of slides affects the image beyond a mere tonal shift or tonal compression. To avoid color distortion, only the part of the characteristic curves in which all color layers have a similar response can be utilized.
The preferred exposure for color slides is therefore very much dependent on the desired colors in the final image. Below I will describe the procedure I use to determine exposure values in cases where I want to be careful about color rendering. It goes without saying, that this procedure will yield good results only, if the favored exposure index (EI) for the film/apparatus combination has been established. The procedure is geared towards digital hand-held light meters, but could be adapted to in-camera or analog meters. Particularly when using a hand-held meter, it is important to establish the exposure index (EI) under similar conditions to those under which the actual images are exposed. Focus distance, transmission and flare of a lens (and hence, e.g., aperture setting, and the use of a lens hood) and filters (incl. UV and skylight filters that are supposed to have a filterfactor of one), all will affect the luminance and its range on the film plane.
Given the small latitude in color slide exposure it is rather astonishing, that the issue is often treated vaguely. Two noticeable exceptions are due to Langford [1] and Egglestone [2]. The latter being the best text on the subject I have seen so far. The method described here has been adapted from Egglestone's white card method [2, pp. 176--181]. There are two modifications to the original method: I use an incident measurement instead of the white card measurement, and the exposure shift is expressed in EV values rather than exposure factors. This is convenient for digital light meters (e.g.,the Gossen Luna-Pro digital) that do not permit filter factors below one, but accept positive EV corrections. Please note, that a positive EV correction (e.g., +0.5) will raise the measured EV value and hence result in less exposure. I.e., a EV correction of +0.5 is equivalent to a -0.5 stop exposure change.
The schema in Fig. 1 shows the method I use for determining color slide film exposure... if I do have the time. The first step is, to decide whether the scene presents one of the following two cases:
If 1. is the case, set the EV correction at the light meter to +0.3, i.e, underexpose by 0.3 stops. If 2. is the case, set the EV correction at the light meter to -0.4. Take an incident measurement and use an exposure setting equivalent to the meter reading.
If neither 1. or 2. is the case then the contrast of the scene has to be taken into account. Take a reflex measurement of the brightest significant tone in the scene. (Glare in which color and detail is not important should not be considered as a significant tone.) Remember the EV value of this measurement. Next, take a reflex measurement of the darkest relevant tone. Subtract the EV value of the second measurement from the EV value of the first measurement, the result is the subject luminance range or contrast. Color films and light meters differ in spectral sensitivity. Hence reflex measurements should not be taken from pure colors and ideally neutral tones (white, gray, black) should be measured.
Consult the table in Fig. 1 for the appropriate EV correction, and set the correction on the light meter. Then take an incident measurement and use an exposure corresponding to the meter reading. See the next paragraph, however, for further considerations.
A subject luminance range exceeding about 4.25 stops can not be recorded satisfactorily with current color reversal films. In this case the usual approach is to trade off shadow detail against color saturation in the highlights. Applying the EV correction (-0.4) from the table in Fig. 1 will do the same and consequently result in loss of shadow detail. If the color or detail in the shadow area of a contrasty scene is more important than the highlights, the latter could be burned out to preserve the former. To do so, the EV correction should be set to a value below -0.4 (even more exposure).
In the following examples, the favored EI is assumed to be 100. The EI has to be set on the light meter, but it is not directly involved in the procedure. However, it will affect the absolute exposure values (EV) which are measured. Let's say we have a scene in direct evening sunlight with a full range of tones. A reflex measurement of the lightest relevant tone in the scene gives an EV of 12.4. A reflex measurement of the darkest shadow in which we want to preserve detail gives an EV of 8.5. From the difference, 3.9, we find in the table in Fig. 1 the EV correction of -0.2. After setting -0.2 as correction on the light meter we take an incident measurement. The light meter shows the measured EV with correction, let's say EV = 10.3. We use an exposure that corresponds to the EV of 10.3, for example an aperture setting half way between 5.6 and 8 (N = 6.7) and a shutter speed (t) of 1/30 s.
Now assume we want to photograph an indoor scene in diffused window light. Let's say the highlight reflex measurement gives an EV of 15.5, and a reflex measurement of the darkest important tone shows an EV of 13.2. The difference is 2.3. From the table in Fig. 1 we get the EV correction of +0.3 and set it on the light meter. Then we take an incident measurement; assume it yields an EV of 14.1. We use an exposure equivalent to EV = 14.1, e.g., N = 11 and t = 1/125.
There is sometimes confusion over the direction in which the photocell should point for an incident light measurement. Some authors claim the it should point from the subject towards the lens, others state that the photocell should point from the subject half way between the lens and the main light source. Both are right. If the diffuser that covers the photocell is hemispherical the former is correct. In case the diffuser is flat, the latter is a workable shortcut for taking the logarithmic average (i.e, the midpoint on the logarithmic meter scale) of two measurements: one, with the flat diffuser pointing towards the brightest light source, and one with the flat diffuser pointing towards the lens [3,pp.542--543].
Fixing the favored exposure index (EI) requires about 6--10 frames to be used for a test. Choose a scene that is representative in color, saturation, tone distribution and lighting for the scene(s) you want to photograph. Make sure that the lighting is constant for several exposures. Use the film/equipment (lens, filter, light meter, camera, lens hood...) combination you intend to use for the actual scene. Set the ISO (ASA/DIN) film speed stated on the film package on your light meter, this is your current EI. Use the procedure described above to determine the exposure value (EV) and take an exposure at a corresponding shutter speed/aperture setting. Than take a several more pictures systematically varying the exposure, keeping track (notes) which exposure setting is used for which frame. It is best to work out a scheme for the variations beforehand. The steps should be 1/3 or 1/2 of a stop, depending on what the equipment will allow.
The strategy for varying the exposure depends on the scene and on the equipment. For example, if the scene has high contrast with large bright areas, flare will affect the luminance range. Since flare will be influenced by the aperture setting, it is better to vary the aperture as little as possible. Often the aperture can be set to 1/2 stops, but the shutter speed only to full stop values. If this is the case, the aperture could be alternated between two positions for 1/2 stop steps and the shutter speed adjusted for full stop steps. If the shutter is known to be inconsistent over its range (as it is sometimes the case with older mechanical shutters) then it is better to vary the aperture. Exposure variations covering the range of -1.5 to +1 stops around the EV determined with the procedure described above should be adequate for the first test, if the light meter is reasonably accurate. (For subsequent tests, utilizing the same light meter, a more limited range is likely to be sufficient.)
After processing of the film, mark the frames so, that the exposure setting corresponding to each frame can be identified. Then compare the pictures under the conditions under which they will be used. Decide which of the frames is the best compromise in color rendering and tone reproduction. If it appears as if a more extreme change in exposure might lead to the desired result, a new test is necessary; probably the light meter is not accurate or a major factor (e.g., bellows extension) had been neglected. If none of the slides shows a satisfactory color rendering, another film type could be tried.
If the frame with the preferred rendering of the scene was taken with an exposure different from the EV determined with the above described procedure, the EI of the film has to be changed from the ISO film speed to an EI that will result in the exposure favored among the test results. This will be the favored EI and should be used for scenes similar to the test scene when photographed with the same film/apparatus combination. I.e., the light meter is set to the favored EI instead of the ISO film speed.
To get from the ISO as EI to the favored EI which includes the exposure adjustments found to be necessary in the test, some formulas are helpful. Film speed (EI) is related to aperture setting (N) and exposure time (t) by
where C is a constant built into the light meter [3, p.538]. It is chosen according to assumptions and compromises made by the manufacturer regarding the likely scenes to be measured. The favored EI will also compensate for the difference between the light meter manufacturers choice of C and your preference. E is the light falling onto the light meter's photocell. During our tests we assume that the illumination of the scene was constant. (Even though we used the procedure only once to get the EV for the first exposure of the test, a reiteration of the procedure for each test step would have resulted (for the constant scene under constant illumination) each time in the same amount of light falling onto the light meter's photocell.) Therefore E is constant in the test and hence E/C is constant for the different exposures during the test.
Without knowing the light meter constant (C) and the amount of light falling on the photocell (E), we can still determine the constant E/C:
Where EI' is the ISO value as EI, and N' and t' are the aperture setting and the exposure time for the frame that was taken with the exposure according to the EV value determined by the procedure described above. (It might be of interest that 2^EV = N^2 / t [3, p.538].)
With the constant E/C determined, the favored EI can be computed from the aperture setting (N) and exposure time (t) used for the best frame of the exposure series:
That's it.
Well, let's consider an example. Assume we use a film, for which the film speed is given as ISO 100/21°. So 100 will be our EI for the test; we set this value on the light meter. We choose a test scene that is under constant illumination. Now we apply the procedure described above to get the EV for the scene, let's assume it is EV = 10.8. We make an exposure with t = 1/15 s at N = 11, which corresponds to EV = 10.8. We take exposures with t = 1/15 s at N = 9.5 (this is +1/2 stop, the setting between N = 8 and N = 11), t = 1/8 s at N = 11 (this is +1 stop), t = 1/15 s at N = 13.5 (this is -1/2 stop), t = 1/30 s at N = 11 (this is -1 stop), and with t = 1/30 s at N = 13.5 (-1.5 stops).
Assume that after processing we find frame number 4 has the nicest colors. The first step now is to compute the constant E/C from the EI used in the test (EI') and and the exposure setting which corresponds to the EV in the test. The exposure setting corresponding to EV = 10.8 was t = 1/15 s at N = 11:
We can now compute the favored EI from the constant E/C = 18.2 and the exposure setting of the favored frame. For frame number 4, this was t = 1/15 s at N = 13.5:
The favored EI for situations as in the experiment is EI = 150.
You may notice, that the resulting EI is not precise. E.g, if the best frame had a -1 stop change in exposure, the resulting EI might not be exactly twice the ISO film speed, as it should be. The discrepancy arises from the fact that the aperture numbers and shutter speeds are rounded values. The series of aperture values can be calculated from:
with, p iterating over the range of stops. Here is a series in 1/2 stop steps:
2.000, 2.378, 2.828, 3.364,
4.000, 4.757, 5.657, 6.727,
8.000, 9.514, 11.314, 13.454,
16.000, 19.027, 22.627, 26.909,
32.000
If in doubt how to round the favored EI, recalculate it using for N' and N the corresponding values from the list above.
No matter how sophisticated an automatic metering system is claimed to be---if it is unaware whether your camera is loaded with black and white negative film, or with color slide film, it has no chance to give good results in both cases.
© K.-P. Zauner, 1999. Located at: http://www.cs.wayne.edu/~kjz/