I was watching an interview with a photographer specializing in headshots for actors. This photog (I forget his name—someone in LA) described how he subtly blurs out the ears and certain other features of faces so that the eyes are more prominent.
He explained that casting agents look first at the eyes, then at the mouth. This is where human beings tend to focus, he said, so to get a feel for what a person really looks like, he modifies the images so that those features are salient. “It’s more like meeting the person that way,” he told the interviewer.
Interestingly, “scanpaths” of eye movement show the same thing. This image was presented to a subject and his eye fixations were recorded. The blue dots represent the movement of the subjects eyes over the photo. This is a fairly typical pattern, from what I’ve read.
This may help to answer the question, “Why draw someone instead of just taking a picture?” The answer might be something like this: “Because in a drawing, the artist can present the features the way a human being will best process them.” The study below is unlike a photograph, if you look closely (especially the nose). It seems absolutely lifelike (to me at least), and yet, it is not photographic.
Cameras are, at this point in time, dumb. They don’t know how to look at a human face. They just record whatever light is reflected off the subject. They don’t do any cognitive processing.
Human beings do. So maybe this is the value-add of the artist, particularly the portraitist: making pictures for human beings, picking out those features that will tickle the brain of other people.
This is not to say that a computer could not in principle do this—I see no reason why it couldn’t. But I think it might be a while. So the jobs of the portraitists are, I think, safe for now. This also suggests that, in order to add most value, portraitists should focus on their comparative advantage—what the camera fails at.