Moonlight is a magical medium in which to live and work. There is something mystical about the soft light that fills the landscape with a calm so unlike the harsh rays of the sun, and the presence of stars in the sky and man-made lights on the ground (or even in the air) adds a dimension similar to that found in Chinese landscapes. For my entire life, I have been fascinated by moonlight. Through a desire to share the visions I found in moonlit landscapes, I began to photograph them 1972. During the more than thirty years since, I have used various methods for determining proper exposures, experimented with different types of equipment, and did a lot of research and thinking about the problems involved in rendering moonlit scenes onto film.
The two key problems are to determine how much light the moon gives to work with, and how film responds to such low levels of light. Though careful experimentation, testing and study, I have mastered both of these problems, and thus can predict, in advance, exactly what camera settings and length of exposure is required to obtain a pre-visualized result.
There are several approximate methods of exposure that are useful as points of departure for developing a practical theory for finding optimum camera settings for making photos by moonlight. These can be found in many basic photography texts in chapters dealing with ambient light, which typically means urban light. The better texts and published guides also give some basic ideas about moonlight, however they never seem to get beyond using maximum aperture under full moon. This method suffers in that depth of field and image sharpness are sacrificed so that minimum length exposure times can be used. The biggest reason for this that I can see is that it also minimizes the problem of reciprocity failure in film exposure response.
One rule-of-thumb that I have heard is the Looney f/16 Rule , where the light level is assumed to be 1/250,000 that of sunlight, or 18 stops less. (Actually, the full moon is 402,000 times dimmer than the sun, or 18.6 stops). Another is the similar Looney f/4 rule. These can be applied whenever the "Sunny f/16 rule" would be applied in daylight, and take into account the difference in brightness between the full moon and the sun, but becomes inaccurate in a hurry whenever the moon is not precisely at full phase or at an elevation at least 35 degrees above the horizon in clear air. A third such method is the Rule of Three 4's.
There are also the guesswork methods, such as widely bracketing the exposure in hope that one of the exposures so obtained will be close enough to be acceptable. This method is very hit-or-miss, and does not allow pre-visualization of the final image. Pre-visualization requires an excellent understanding of how the materials used will respond, and often requires balancing several technical aspects of the photograph in order to obtain an optimum result.
The method of exposure calculation I use today has evolved through several generations. Through experimentation and test, the approximate luminance of various phases of the moon was determined, and the means of turning this into exposure predictions was worked out. In those days (1972-1986) exposures were generally made by trial-and-error, though by taking good notes I was able to achieve a reasonable level of success. However, I knew much better results were possible, and was determined to have them. In 1987, an algorithm suitable for a computer program was developed, which resulted in the program NightLandscape 1, which printed a table of predicted exposures based on the "age of the moon" (in days from previous new moon), with tables made for various film speeds at lunar apogee, perigee, and nominal altitude. Use of the tables required knowledge of the "lunar age" and distance from the earth to the moon, which were obtained each month from the lunar data column in Sky and Telescope, a leading amateur astronomy magazine. After I discovered the work of Peter Duffett-Smith, a lunar age and position calculation algorithm was devised, which when combined with the previous computer program resulted in NightLandscape 2, which was hosted in Microsoft Basic on an Apple Macintosh SE computer. Further refinement of the moon brightness at various phases, and incorporating a graphical display resulted in NightLandscape 3, which was hosted on an Apple PowerMac 7100/80 computer. A further revision of the program, NightLandscape 4, includes many more astronomical calculations to do such things as predict the location of the moon in the sky, and the end times for twilight after sunset and before sunrise. This program produced the desired exposure calculations in three steps. These are:
* For a given location and time, calculate the phase (fraction of the lunar disk that is directly illuminated by sunlight) and position in the sky of the moon using the astronomical and celestial mechanics routines originally written in the BASIC computer language and published by Dr. Peter Duffett-Smith of Cambridge University. These routines are based on algorithms developed by the US Naval Observatory, and described in the book Astronomical Algorithms by Jean Meeus. Orbital phenomenon are modeled using Fourier Sine and Cosine expansions, a technique that is frequently used in engineering and scientific applications for widely varying periodic phenomena. This is necessary since the moon's orbit around the earth varies widely due to the perturbing influence of the sun. This causes the moonlight brightness to vary a lot from one lunation and the next, over a period of 18.6 years. I have modified and adapted Dr. Duffett-Smith's routines extensively, and now have them hosted in Visual Basic 2001 on an Apple G4 laptop computer.
* Using the lunar phase (fraction illumination), calculate the amount of light provided by the moon, using lookup tables with values obtained by test. I performed the tests that lead to the data in these tables over a period of approximately 16 years. The result from the lookup table is then corrected for various parameters that can affect the amount of available incident moonlight, including the current distance from the earth to the sun, the distance from the earth to the moon, and the position of the moon above the horizon. However beginning in mid-2003, the test data lookups were replaced with a set of calculations based on information in Meeus' Astronomical Algorithms and Allen's Astrophysical Quantities that produce a much more accurate theoretical brightness of the moon. This is combined with a much better atmospheric scattering model to give very accurate predictions of the available light from the moon.
* The available incident light value is then used to calculate a series of possible exposures, depending on film speed and lens aperture. The original method used was the fx exposure method devised by astrophotographer Barry Gordon, however recently I have been using a Light Value / Exposure Value system . These calculated exposure times are then corrected to account for film reciprocity failure using either another lookup table or a curve fit to test data. The reciprocity failure correction data table began as the standard Kodak reciprocity failure correction data, but has been updated over the years to be accurate for specific films I like to use.
NightLandscape 4 is entirely coded in Visual Basic, and is hosted in a Microsoft Excel 2001 workbook. The results are presented as formatted Excel spreadsheet tables within the workbook and can be easily printed. However the entire package also resides on a Macintosh Powerbook G4 computer that I can easily take with me into the field. NightLandscape 5, which is currently under development, replaces most of the data table lookups based on my own flawed test data with much better calculations based on information culled from the scientific literature.
The exposure times calculated by NightLandscape 5 provide a daylight-like rendering of the scene, meaning that the film density for values in the Zone V (middle-gray) range are equivalent for objects photographed both under moonlight and average sunlight conditions, where the "standard exposure formula" (or Sunny 16 Rule: shutter speed = 1/EI, f=16) is valid for the sunlit case. Standardizing in this way allows easy pre-visualization and interpretation of the scene. A moody, brooding feel can be obtained by under-exposing the image by a half- or full-stop (or even more), while a balance between ambient moonlight and low-levels of urban light can be obtained with a slight over-exposure. On the whole, there is not much room to maneuver, though, since the shadows in moonlit photographs are hard to manage, and easily become unexposed areas in the film due to the increased reciprocity failure in such areas of the film. Likewise, urban light quickly overwhelms moonlight in intensity, and a balance between the two in a single image is very difficult, if not impossible, to achieve. (Urban light can work in an urban image if is kept to a very small area in the frame, with care taken to control lens flare and film halation.)
The exposure times calculated by NightLandscape 4 are only strictly correct when the moon is at least 35 degrees of arc above the horizon. The reason for this is that the light from celestial objects (stars and planets) must pass through the earth's atmosphere in order to reach us. When a celestial object is positioned close to the horizon in the sky, the light must pass through more of the atmosphere along the way. Since the atmosphere absorbs a portion of the light passing through it (through interaction of the light with molecules in the air), the greater the path through the atmosphere, the greater the amount of light absorbed, and the dimmer the object appears. Astronomers call this effect atmospheric attenuation (or atmospheric extinction) , and go to great lengths to avoid it. I have made tests to determine the amount of darkening this effect has on incident moonlight, which you can read about here . I incorporated this information into a version of NightLandscape (4.1b), but lately have begun to rely on corrections based on more accurate methods from the scientific literature. NightLandscape 5 will incorporate a full three-factor atmospheric scattering model.
Besides the conventional reasons for using less-than maximum lens aperture (optimizing depth of field and lens sharpness), selecting the aperture for use in a moonlight photograph can also be used to balance the exposure obtained by moving lights (automobile lights, airplanes) relative to the moonlight exposure. With a moving object, use of a wider lens aperture allows more light to be recorded on the film. Thus if a dim moving object is an element in the photograph, a relatively wide aperture can be used, while a small aperture can give an equivalent result for a brighter light in motion. In either case, the exposure time is adjusted to obtain the same ambient moonlight exposure. Thus the relative brightness between moving and still objects can be easily controlled.
Color photography by moonlight brings an additional problem. Color films are typically made of three separate emulsion layers, each of which records the image component in one of a triplet of complementary colors. Each emulsion layer has its own exposure characteristics, including emulsion light sensitivity (speed), contrast, and reciprocity failure response. Thus in low light conditions with very long exposures, the differing reciprocity failure response in each layer can lead to one layer responding far differently from the others, which produces a color shift in the final image. The amount of color shift varies from film to film. I have not found a single color film that has absolutely no color shift with long exposures, however some are a lot better than the others. The worst one I ever had the misfortune to use was Kodak Ektapress, which produced essentially a yellow and black image, with nearly no blue or red response. I have found that Kodak Kodachrome films typically shift dramatically toward the magenta, while Kodak Ektachrome tends to shift toward blue by an unacceptable amount. Kodak Max 800 film is about as lacking in color shift as I have ever found, though it has more grain than I like. I have also used Kodak 3200 speed Ektachrome, which is really nothing more than a high latitude 400 speed chrome film that can be pushed three stops, resulting in a huge gain in contrast and excessive grain, though the color was pretty good. I used to use Fuji RHP 400 D slide film before it was discontinued, and found the slight shift toward green acceptable and easily removed in printing. Until recently, I used mostly Fuji Provia 400F for slides, however recent testing of Fuji Provia 100F RDPIII emulsion shows it is far superior in color and grain, with remarkable low-light speed due to a near lack of reciprocity failure. I have found Fuji NHG 800 negative film has really good color and tight grain for a fast negative film, and the speed helps keep exposure times to a reasonably low value. It also has a long latitude that helps when urban light is a part of the scene. However, it seems that Fuji has discontinued this emulsion, and I haven't had a chance yet to test its replacement.
As for useful exposure times, I find that exposure times from three to fifteen minutes work the best. Longer exposure times are needed if there is a man made light traveling through the image, since it takes a finite amount of time for the light to move from the desired starting point to the desired ending point in the frame, thus producing a streak of light. Reasonably good star trails can be recorded with exposures of about five minutes, though fifteen minutes is better and a half-hour is better still. (You need trails at least 30 degrees of arc in length to give the circum-polar ring effect, requiring a minimum exposure length of two hours.) Exposures longer than fifteen minutes try my patience to too great an extent, and I sometimes find my attention wandering with such long exposures to the point that I loose track of time and ruin the exposure. I have found that taking a digital kitchen timer into the field helps with timing long exposures. Many times I have been roused out of reverie by the beeping of the kitchen timer telling me that I need to get up and close the shutter. I recently started using a Cannon EOS 3 camera with a remote timer that allows the exposure to be programmed. This tool has improved both the accuracy of the exposure and the pleasure of making them, since I am now free to enjoy the moonlit ambience and let my mind wander once the exposure has begun.
While I do not use filters at the lens to enhance landscape images, such could be tried. Moonlight is of similar color composition to sunlight (the lunar surface is a dark, neutral gray color, producing broad-spectrum reflected sunlight), so in theory any filter used to improve image contrast should work for moonlight as well as sunlight. In particular, I have thought that use of a polarizing filter might help darken the sky to produce brighter stars, though I have not yet tried this. It could be that the reduced light reaching the film due to the filter factor might negate this effect. I would be interested in hearing from anyone who has tried it.
Most of the moonlight work gets done in 35mm format, using cameras with all-mechanical shutters, though I do some work in 4x5 format as well. The big limitation in using the larger format for moonlight photography is the larger f/stop numbers required by the larger format. An exposure that can be obtained in seven minutes with a 35mm camera and EI 400 film can take two hours or longer with 4x5 format Tri-X. Despite numerous queries made of photographic suppliers, I have yet to find a 4x5 color film that is fast enough to be useful in moonlight photography. I did some work in the late 1980s and early 1990s with 160 speed Kodak LPS, however the exposure times required (in excess of 2.5 hours) meant that only two images could be obtained on a given night. Though the sharpness and image quality of the 4x5 moonlight image can be astounding, setting up and focusing the camera in the dark can be a maddening experience.
And as a final note, I find that landscapes with moon high in the sky tend to look just as washed out as landscapes made under noon sunlight conditions. Consequently, I tend to do more moonlight photography in summer months, when the moon grazes the southern horizon, and reserve the high moons of winter for performing film tests and other craft-enhancing chores.
This whole enterprise is a work in progress, as I am constantly seeking to improve the technical basis of my chosen exposures. I can try to answer questions sent to me at this email address. General comments are also most welcome.
C. D. "Kit" Courter
Torrance, California, USA.
Photography by Moonlight This site has a lot of useful information. I wish it had more sample photos to show what this artist can do.
Adams, Ansel, "The Camera", New York Graphic Society, Boston, 1980.
Adams, Ansel, "The Negative", New York Graphic Society, Boston, 1981
Allen, C. W., 1976, Astrophysical Quantities (London, Athlone)
Duffett-Smith, Peter, "Astronomy With Your Personal Computer", second edition, Cambridge University Press, UK, 1990
Gordon, Barry, "Astrophotography (featuring the fx system of exposure determination)", second edition, Willmann-Bell, Richmond (VA), 1985
Hedgecoe, John, "The Photographer's Handbook", Alfred A. Knoph, New York, 1980
Meeus, Jean, 1998, Astronomical Algorithms" 2nd Ed., Willmann-Bell, Richmond, Virginia.
Upton and Upton, "Photography", Little, Brown and Company, Boston, 1981
This web page and all contents (c)2003-2005 LunarLight Photography.
All Rights Reserved.
Contact LunarLight Photography