Digital cameras have brought a great flexibility and power to the world of night photography. The principal advantages relative to film are three: rapid feedback of exposure accuracy through the histogram function; no reciprocity failure; and the no image quality loss from digitally scanning the film or print image. But nothing comes for free, and with long digital exposures comes pixel noise in the sensor, and the limited charge life of batteries required to keep the shutter open. Another thing to watch for is the inaccuracy of current light meters when the light falls below LV 2 or so.
Finding Correct Exposure Time and Tonal Scale Management
Finding the correct exposure for a night image is much easier with a digital camera than it is with film. In fact, one can use the digital camera's histogram function to iterate on exposure until an optimum image capture is obtained even without knowing ahead of time how much light you have to work with. Though that knowledge does speed up the process, sometimes substantially. The iterative method involves guessing what the exposure settings need to be, making a trial exposure, and then applying the in-camera histogram function to the results. If the histogram is off to the left (or more likely all piled up at the left edge of the screen), add more exposure and try again. If the histogram is off to the right, reduce exposure and try again. Repeat until the histogram is centered on the display screen. If you have a lot of light to work with, the correct exposure can be found within minutes. With moonlight work this can take far longer, sometimes hours if long exposures are being used. You can speed up the process by making trial exposures with a very fast lens, such as a 50mm f/1.4, and then adjusting to the desired aperture once the optimum exposure has been obtained. The lack of reciprocity failure means that a scene has an EV value, unlike film where the long exposure time varies so much with changing aperture that the notion of EV for similar camera settings is meaningless.
Knowing how much light you have to work with is the same with digital or film. For light levels down the range of LV 2 to LV 0 or so, a light meter can be used. In-camera meters are convenient, and spot meters are good for reading extreme areas within the planned image frame, however all become less accurate as the light levels fall off. Tests with the internal meter of my Canon 20D show me that at moonlight levels around LV -4, the meter estimates the light levels are a couple of stops too high, with the error growing as the light levels continue to fall. At starlight levels, the meter continues to give readings, but they are entirely useless. So for moonlight work, I still rely upon brightness predictions from my NightLandscape program for the first iteration. Scenes lit by urban light or other manmade sources are typically bright enough the meter remains a reliable guide, and so the camera's internal meter can serve as an initial reading. The nice thing about the iterative method is that, no matter how far off you are to start with, you will get to optimum results eventually, with the time to get there depending on how close your initial estimates are.
Applying the method on the last few iterations takes skill, since at this point the camera can record the scene adequately, however slight changes in exposure effect image quality. Experience, gained by practice and learning, is needed to know how the image quality relates to histogram shape and placement on the screen. In general, two key image management issues arise, and there may be more that I haven't thought of yet. These issues also relate to how the image has been pre-visualized. First of all, there is getting the exposure scale of the whole image within the histogram. Digital camera image characteristic curves have a very abrupt shoulder when the pixel luminosity reaches 255 counts, which is the ultimate top of the tonal scale Any scene brightness above that recorded at 255 counts will be rendered in the image as featureless white. So if you want some part of the image to have highlight detail, you need to be careful about where those highlights fall in the histogram. If they are that part of the histogram that is just a spike at the right margin of the screen, you have lost your highlight detail. Digital camera image characteristic curves have a very shallow toe, which is wonderful for preserving shadow detail, so if you have to bias the image one way or the other, bias a little toward underexposure is the way to go. But bear in mind that the brightness difference between pixel "count" values decreases as you approach the toe, and sensor noise becomes apparent much faster in shadow areas.
However, too much bias toward the dark end of the tonal range can bring another problem. If the image is generally too dark, and you rely upon digital contrast expansion in Photoshop, too great an expansion can cause banding in the sky where a slow transition of light levels occurs in the scene. The digital tonal scale has only 256 steps, and tonal scale expansion in Photoshop makes the step size larger. If you expand it too much, the step size becomes noticeable as banding in areas where the tonal gradation in the original image is very smooth and gradual. I hope to provide some sample images to illustrate this issue soon.
I find that placement of the histogram on the scale so that the highlights are about one stop from the top of the scale, while the majority of the image is clustered slightly below the midpoint, gives an image that I can best work with in Photoshop. Once in Photoshop, the scale, and shape of the curve can be adjusted to achieve desired results, and things like color shift and saturation can be controlled. These are all tools to be used in producing the final image, just like taking film into the darkroom. With film, choice of developer and development methods allowed adjustment of the tonal scale, which could be further adjusted along with color shift and saturation while making the print, providing you know your materials. The digital darkroom offers these same controls, only with greater flexibility and ease of control. The digital darkroom also offers local contrast control in both tonal and color dimensions, much like polycontrast photo paper, but again with greater flexibility. As photographers, we are truly blessed with the range of options digital darkroom techniques have brought to us.
Tonal range can be extended on a high contrast image by making one exposure to record detail in the highlights and another to record shadow detail, and then combining them in the digital darkroom. I know that some artists feel this is not allowed within their guidelines of "what is photography", however I would submit that if this were done internally in the camera by means of the image processing chip instead of using Photoshop on a PC in the digital darkroom their complaints would vanish, so what is the point?
Digital Noise Management
Digital noise can be a frustrating problem. It comes in two forms, a uneven light fog that slowly builds up in the image, and hot pixels.
The light fog version of digital noise can be reduced by making a "dark exposure" (identical exposure to the image exposure only with the shutter closed) immediately after the image exposure and then subtracting the dark exposure from the image exposure. This is something that is done internally by most camera digital image processor chips when the camera is in "noise reduction" mode. That is why "Noise Reduction" mode doubles the amount of time required to expose an image, since the exposure must be made twice, once with shutter open and again with shutter closed, before this internal image processing step can be performed. You can also make the "dark exposure" separately and then subtract it from the "image exposure" using layers in Photoshop (use the "Difference" blending mode). The noise on any sensor changes with time and temperature, so the dark exposure must be made as close in time as possible to when the image exposure is made. If the temperature where you are working is stable, you can save some time by taking a dark exposure only once every several image frames and then performing the subtraction in Photoshop rather than using the internal camera noise reduction. This is useful if you are trying to make a series of half-hour exposures and don't want to extend them into hour-long waits between frames. However doing this does run some risk of not being able to get rid of all the noise if the temperature changes. Remember that use of the chip causes the chip to self-heat a little, which changes its temperature some.
However even if you use this method to remove digital noise fog, it doesn't seem to completely remove hot pixels, which are not fogged but are at full saturation (255 pixel count value) and can happen no matter how long the image exposure has been. Hot pixels are not always just single image pixels located randomly within the image frame; often the pixels immediately adjacent to the hot one are affected as well from some sort of electrical bleed-over effect, leaving a bright cross shape within a circle centered on the hot pixel. You can sometimes use the dark exposure method to remove them, but that can leave you with the black opposite image of the affected area. Hot pixels that are not fully or properly removed using the dark exposure method must be corrected in Photoshop using the clone stame tool. Typically, the first step in the digital darkroom with the image loaded into Photoshop is to carefully inspect the entire image for hot pixels and correct them using the clone stamp tool. I have (as of this writing) a 2 1/2 year old Canon 20D, and I have noticed the hot pixel problem has gotten worse with age. I seldom had the problem when the camera is new, but most images made with it these days require careful correction of up to about 50 of the little buggers!
A very good read on digital noise management is this page on photo.net.
That Damn Sensor Dust!!
Sensor dust is as much a plague on digital night photography as it is on our daylight brethren. I have purchased a very nice sensor cleaning kit (the "Mega Kit" from Copper Hill Images ) for about $60 that includes a battery powered blower with a small nozzle and a soft hyper-clean nylon brush that can be given a static charge by brushing it on a supplied piece of velum paper before brushing the sensor. It also has a special shaped wand very much like a soft plastic cooking spatula that you wrap with optical cleaning tissue that is then moistened with reagent-grade methanol (both supplied, with replacements available) and then wipe across the sensor surface. The kit also comes with some nice lens cleaning tools. It all works very well, and I highly recommend its use.
The blower removes anything that isn't clinging very tightly to the sensor surface - and most anything that clings to the surface is static-charged and so clings pretty tightly (thus the origin and the crux of the problem). The soft brush does a great job of removing loose static-charged particles that haven't been on the sensor very long. It only takes a few seconds to charge up the brush by wiping the bristle tips across the supplied velum paper a dozen times or so. You have to blow the brush off after use and store it in its little container to ensure it remains clean, and don't touch the bristles with a finger (though if you do, it comes with cleaning instructions). Particles that have been in on the sensor for a while cling more tightly to it, and can be removed using the optical-quality wipe taped to the soft wand and moistened with the methanol.
However you still have to check the sensor for dust every damn time before heading out with the camera if you want to come home without a dust blob ruining the sky in an image. I do that by putting a long lens on the camera, setting the focus at infinity, and then photographing a white wall from a distance of a few feet (keep your shadow out of the image!). Then transfer the image to the computer, open in Photoshop and apply "Auto Levels" or "Equalize" to it. Any sensor dust just jumps right out at you. Remember the image made in the camera is reverse to what you see on the screen, so a dust blob in the lower right of the image is actually in the upper left of the camera sensor.
I start cleaning by opening the shutter (use the camera "sensor clean" mode) and then applying the blower to the sensor and the space around it. Hold the camera with the lens mount side down when you do this so that gravity will help pull the dust out of the shutter space instead of helping to settle the dust back onto the sensor. After a good blowing-out, make another test. If there are no blobs, you are done, however you may need to go the next step and use the charged brush and then blowing it out again. If that doesn't work, use the optical wipe and methanol on the wand - I have never had that technique fail. The recommend practicing use of the wand on a CD Jewel Case to get the feel of it first, something I have found really helps.
I hear that Canon and Nikon are taking the sensor dust issue seriously and their engineers are devising better self-cleaning tools. However until I see that in front of me, I'll continue to test the sensor before every shoot and clean as necessary. If on the road for a week, and staying in a relatively dust free place like a friend's house or a motel, you can test and clean every day. You can reduce the risk of dust by being careful not to change lenses in a dusty place (like the desert with a strong breeze blowing), or better yet not changing lenses at all in the field (but who wants to be constrained by that!).
Digital Star Trails
One of the difficulties in capturing star trail images with film is that with very long exposures eventually the small amount of urban light reflecting from airborne dust in the sky will fog the film. This makes it very hard to capture bright star trails against a dark background if there is any urban light in the sky at all. The star trail images I have made up to now have relied upon my being far away from urban light sources so that a three- to six-hour exposure will not result in objectionable light fog ruining the shot. It turns out that digital techniques can greatly relieve that problem. The trick is to break the exposure time into a sequence of many short exposures using a camera remote timer. This must be done with the noise reduction feature turned off so that the exposures follow each other with as small a time gap between them as possible (most remote timers have a minimum 1 second gap, but that is so short that gaps in the star trails will not be noticeable with most sensors and all but very long lenses). Once you have the sequence of images, you can combine them in Photoshop as layers using the "lighten" blend mode. There is also some specialized software such as a program called "Startrails" that can be used to combine the images (I have not used any of them yet and so don't have an opinion on them to share). This technique could also be used with film, using all the frames of a 36-exposure roll of film, however with film you then have to scan and register each image before combining can begin. So this is another matter in which the digital camera is superior to film for night photography.
Topics to be added (when time allows) -
- Extreme low light photography (the crescent moon and starlight as a light source!)
- Stacking (compositing) Digital Images
** This is a work in progress. Come back soon to see more! **
C. D. "Kit" Courter
Torrance, California, USA.
This web page and all contents (c)2007 LunarLight Photography.
All Rights Reserved.
Contact LunarLight Photography