There is a widespread recognition that using a digital medium for photography makes it much more prone than film to "image manipulation", whatever this means. Recent reports of press agencies distributing "manipulated" photographs makes the subject timely. What is "image manipulation", and where lies the border between "straight" and "manipulated" images? Is the problem more widespread in digital photography, indeed?

Let us quickly enumerate some obvious factors, if only to get them off our way. A photographer can distort the truth just by choice of time, place, and subject of a picture. This works because photographs are usually perceived as more "objective" than most other media (except motion pictures and TV). While you cannot show a picture of a dinosaur grazing in Kansas (this would be an outright fabrication), you can easily imply that only a few nuns and elderly people were greeting the Pope in Poland; we've seen that.

Now, some of the more technical aspects. Framing, angle, and perspective. Shoot the same street demonstration up close with a wide-angle lens, framing your image off-balance, and it will look more dramatic than if taken with a longer lens and framed straight. (On the other hand, a long lens may make the crowd look thicker.) Shoot a policeman with your wide-angle lens at waist level, and you get a bad, threatening cop. The same guy from a longer distance, with the camera at eye level will be just someone doing his nine-to-five job. You can think of similar effects related to light, subject juxtaposition, and more. None of this changed in the digital technology.

And here we come to less obvious aspects. In addition to a potential for blatant image manipulation (adding or removing contents), the digital darkroom allows for easier image adjustment of color, tonality, and some other image attributes in postprocessing. Here is where the purists start their laments, not about a falsification of reality, but about the postprocessing being less "natural" or "photographic" than anything going inside the camera, digital or (preferably) film-based. Most of that is, however, a misunderstanding.

A "straight" image is one where the color and contrast are not altered, right?

Let's start from color. There is no such thing as "true" colors in photography. A scene lit by incandescent bulbs is really hopelessly red-shifted, as their light is much more red than sunlight. Our brain adjusts the perceived color accordingly, and, in a similar fashion, image processing (inside or outside the camera) also does that, to create a natural-looking effect. In case of film, this is done either at the printing stage (negative film), or by choosing the proper type of emulsion and/or corrective filter (transparencies). Does that mean that it is OK to let your digital camera do it, but not OK to do it by hand on your computer? And what about people who use Kodak Portra for portraits and Fuji Velvia (more saturation) for other applications? Is that really "straight"?

Tonality and contrast adjustment is an even stronger case. A sunlit scene has a brightness range of about 1:10000. A good print would carry 1:100, and a good computer display 1:500 or so. Some of the original range has to be either lost or flattened, and both film and digital technologies try to do that. Still, lots of highlight and shadow detail is lost in the final image. This is a detail which was perfectly visible to a person who was there. Some digital cameras use special technologies boosting the shadows and darkening highlights. Is it OK to use such a camera, but not OK to mask these areas and adjust them selectively in Photoshop? Besides, this is nothing new; for decades we've been dodging and burning our prints, at least until we've given up our own darkroom work, content with "straight", drugstore prints. Moving a cotton ball on a piece of wire under the enlarger lens is more awkward than using a mask tool in an image editor, but does it make the process more "natural"?

Dodging and burning aside, each of us may also have different preferences regarding how an image should look. Ansel Adams liked his Yosemite pictures printed strong, with the midtones moved way down; not exactly a drugstore print. Your taste may be different than that of your camera maker; the required effect may also change from one image to another. The fact that the adjustments are easier to do in the digital domain is actually a plus; it allows you to bring the final image closer to your subjective perception of the scene, or to the effect you want to impose on it.

To avoid beating a dead horse I will skip the subject of color filters in B&W photography, or of a polarizer used to deepen the sky in color.

One may even consider using a flash to be one of the most intrusive ways of image manipulation, much more so than anything I mentioned above. We are drastically changing the scene by bringing our own source of light, and the effect may be very different from what anyone could see on the spot. Compared to that, dodging and burning (or selective tonality adjustment, as we call it now) is small change. Should we ban flash use in news photography?

Sharpening of digital images is also sometimes considered "bad". Well, the digital image is built of three color components, each laterally shifted with respect to the others; it is also blurred on purpose by the antialiasing filter in front of the sensor. Sharpening will restore the detail crispness lost in that process, so why not? It will not add detail which was not actually there. Besides, for decades some photographers preferred this or that lens because it was softer, sharper, or had a nicer "bokeh"?

Another complication is that viewing the original scene our eyes re-focus when we move our attention from one point to another, an option lost when looking at a two-dimensional image of that scene. There is already so much distortion of reality here that any discussion of sharpening seems just out of place.

In short: all ways of image postprocessing discussed above are (a) nothing new, and (b) perfectly kosher, as long as their aim is to restore our subjective perception of the scene (or, in non-documentary applications, to convey a given atmosphere). Therefore they do not deserve to be called "image manipulation", at least not in the sense that the image is less "real" than it would be unprocessed. There is no such thing as unprocessed image, as imaging itself is a process.

Now we come to the more obvious cases of manipulation: adding or removing actual contents of the picture. There are fewer doubts here: obviously, an image loses or gains something which really was not there. While situation is quite obvious at the extremes (straight image versus Aunt Winnie replaced with a flower pot), the borderline can be quite fuzzy. Is it OK to remove a cigarette butt in the foreground before taking the picture, but not OK to do the same with a healing brush in postprocessing? While technically you are altering the image detail, does this make your picture less "true"?

One of the famous American landscape photographers travels on location with a sweeping broom, to prep the foreground before taking a picture; he even refers to it as his most important accessory. Remove a footprint here, or a couple of broken twigs there: how much better is this than doing the same with the clone brush of your graphics program?

Or, shooting with a flash, we often capture some dust motes and bright reflections off surface imperfections, not visible under the ambient light. Removing them will actually make picture closer to what we saw there. One may argue the result will be more "straight" than the original image.

In general, no set of rules or guidelines will replace a common sense. Even in documentary photography, some flexibility must be applied to avoid getting into a paranoia. This amount of flexibility is even larger for us, who take pictures for pleasure, without a mission of "bringing truth to the masses".

As an illustration how an attempt to codify the "truth" may lead to unintended (and sometimes comical) results, let me quote verbatim the summary of the Reuters' guidelines for their staff photographers and freelancers (the summary was extracted from the article as of February, 2007):

The title of the guidelines, "The use of Photoshop", already hints that something is wrong here. A closer look at the summary amplifies the impression. Take the "lasso tool" and "quick mask" items: these are tools for selecting a fragment of an image, but they do not introduce any changes to it. "Adjustment to histogram limits" is practically the same as "Auto levels", but performed manually; one os OK, and the other a no-no. "In-camera sharpening" is something you cannot turn off; you can only adjust the magnitude. "Sharpening at 300%" indicates that the writer never heard of methods other than the Photoshop Unsharp Mask (a quite intrusive sharpening tool), in addition to the fact that the final effect at given settings will look very different for a 2 MP image than for a 10 MP one. Lesson: do not go into technicalities if you do not understand them.

More, every camera has at least one, default, "saturation style", so I'm allowed to bring three different cameras to an assignment, but not one camera switchable between all three color renditions, how clever. ("You may use three different films, but not in the same camera".) A clear impression is that the more specific a rule is, the less sense it makes. While the effort is well-intended, Reuters deserves a credit for helping me to make the point about the common sense.

Better and more technically literate guidelines have been published at the digitalcustom.com site, if you would like to have a look.

To add things up, again: use the common sense. Sometimes the borderline is crossed in an obvious way: the well-known cases of Newsweek combining Martha Stewart's head with somebody else's body on a cover picture, or National Geographic moving a pyramid closer to another to fit both on a vertical cover. A British writer, David King, spent half of his life collecting blatantly manipulated photographs from Soviet sources, with people disappearing and replaced in various edits of the same picture (his coffee-table book, "The Commissar Vanishes", printed in 1997, is a highly recommended reading). And this was before digital; most of the work had to be done with an airbrush and a hobby knife. But is removing someone's elbow getting into the frame a falsification? Only the common sense (and your real intent) may answer this.

If you really want to cross the borderline, there is nothing to stop you. After all, photomontage is recognized as an art by itself. Just don't try to disguise the result as a documentary photograph. Quite often the nature of the image will be recognizable at the first glance; if not, provide a brief explanation with the caption, just to keep the viewer from being misled.

A Postscript

After the original publication of this article in the Quest, I received two nice examples from my Internet respondents; I think they fit well into this article.

The first shot is by Bob Alexander, who uses an E-500 for most of his work. Bob is retired and teaches photography classes in his free time; as you can see, he has a strong background in flash lightning techniques.

While you can tell this is a digital effect (for example, there is no reason for the left part of the cleavage cut to be more blurred than the right one; both are at the same distance), the overall result is very pleasing.

For me, however, the strongest effect is due to the tilted cropping, entirely changing the viewer's impression of the model's attitude: from reserved to outgoing. This much about "kosher" image adjustments.

The other example, a delightful and whimsical beach shot I've got from Taffy Ledesma from the Philippines, looks like a superposition of two images, perhaps the heaviest form of image manipulation.

Web Links

For your convenience, here are the two links mentioned in the text of the article:

• The Use of Photoshop by David Schlesinger at the Reuters' site;
• DigitalCustom Guidelines —more realistic and less entertaining that the Reuters' ones;

Last but not least, look what I recently found:

• An article by Ron Bigelow on the same subject. A recommended reading.