On Image Editing

On Image Editing

Digital photography didn't just make it easier to go from scene to sending the finished picture to all your friends, it also made it a whole lot easier to edit the image. But what is acceptable?

A photograph is always only a partial view of reality. Even if you have film or a sensor with infinite dynamic range and infinite sensitivity, you'll only ever capture what is in the viewfinder, and then only what is in focus.[1] In reality, of course, we don't have infinite dynamic range, nor infinite sensitivity, nor infinite depth of field. A picture may be worth a thousand words, but we have ample possibilities to control just what those words are.

Given that absolute truth is unattainable, then, what is attainable? Here I think we must primarily look at what claims we make about the photo. If we claim that the photo contains everything relevant to the situation shown in it - for example, a newspaper photo - then we better not remove or add anything that a reasonable person would consider relevant. If our only claim is that the resulting image sure looks good, then we can do a lot more editing.

When it comes to photos of people there is a conflict between these two claims. On one hand, nobody looks as good on an unedited photo as in real life:

  1. Due to a photo being a still image, there is more time for the viewer to look at details, such as (but not limited to) pimples.

  2. A photo can be enlarged. For example, a 6 megapixel photo can be printed at A3 size. At that point, the aforementioned pimple is ten times larger, if we let the face fill the frame.

  3. The camera's in-camera processing tend to increase contrast and sharpen edges, making the ten-times magnified pimple stand out even more.

  4. My own editing process tends to increase contrast even more, just to make the image stand out more. By now, any normal human visual cortex is only seeing said pimple and nothing of the rest of the face.

It is therefore quite unfair to the person being portrayed. But on the other hand, people tend to study photos of models very carefully and walk away with the conclusion that the single red spot on their face makes them horribly ugly, when, in fact, nobody but them even notices the red spot - because everyone has them, our brains simply don't bring them to our attention - unless they happen to be ten times the normal size and contrast-enhanced.

Here we can see two approaches to portraits: Do we show what actually is, or do we show what is actually seen? I don't do what I would call "documentary" photography. That is, I make no claims beyond that I think the photos I put up are kinda cool to look at. My policy is therefore that if you wouldn't notice the defect in real life, and I'm not claiming that the photo is exactly what the person looked like, then just get rid of said defect.

This policy was put to the test when I took a photo of a relative. A teenage girl relative. When I brought up the photo on my laptop I noticed three pimples on her chin. Too small to be noticed in real life, the processing had brought them out. But even if they had been noticeable, I'd have the same problem. On one hand, I don't want to propagate the notion that perfect skin exists[2] - on the other hand, as I stated above, the camera isn't fair, and using my young cousin as the battering ram seemed very unfair to her[3]. But if I start to remove pimples, where do I stop? Should I turn everyone into a porcelain-faced doll? Finally, I decided that since I hadn't noticed the pimples until I saw them ten-times magnified and detail-boosted, at which point they did an excellent job of attracting the eye, they'd just have to go.