Will A Formalized Scale For Photoshopping Do Any Good?

Illustration for article titled Will A Formalized Scale For Photoshopping Do Any Good?

At this point, we know that most, if not all, of the commercial, non-news images we see are digitally enhanced. Still, studies show that women — of all sizes — feel bad and/or think negative thoughts about their bodies after viewing pictures of models in magazine ads for just three minutes. And many magazine readers are not yet women; they are girls, with still-developing physiques, self-esteem and brains. Yet extensive Photoshopping persists. But what if you could measure just how much a photograph had been altered? Would it help in countering the effect of these misleading images? We may find out: A proposal for a program is in the works.


The New York Times reports that Dr. Hany Farid, a professor of computer science and a digital forensics expert at Dartmouth, and Eric Kee, a Ph.D. student in computer science at Dartmouth, are proposing a software tool that can measure just how much fashion and beauty photos have been altered, on "a 1-to-5 scale that distinguishes the infinitesimal from the fantastic." The research that Farid and Kee have done has been published in The Proceedings of the National Academy of Sciences.

According to The Economist:

Illustration for article titled Will A Formalized Scale For Photoshopping Do Any Good?

Dr. Farid says Mr. Kee gathered roughly 450 examples of before and after photos that show retouching, finding them at sites that document fashion-photography blunders (which the magazines typically defend as legitimate) and retouchers advertising their services. He provides a sampling of such images on a website, including ones that show modified bodies that are anatomically impossible.

The researchers' algorithms look at two separate forms of change: geometric, in which retouchers enlarge breasts, trim legs, elongate necks and the like; and photometric alterations, which involve changing skin tone, removing blemishes and wrinkles and smoothing the skin. The output of these algorithms is a number from 1 to 5, with 1 being the least changed and 5 the most.

The Economist points out that Dr. Farid's work may be computer science, "but his interest is social change."

He proposes that magazines voluntarily adopt a code in which his algorithm's result would be shown alongside modified photographs, possibly with explanatory text that details the sorts of changes found.

Farid thinks self-regulation could happen. "Models, for example, might well say, ‘I don't want to be a 5. I want to be a 1,' " he tells the Times. Wishful thinking, considering the fact that most models (and celebs) are digitally sculpted without their permission. Plus, it seems incredibly unlikely that a single woman's magazine would agree to using or printing such a scale. And advertisers? No way. Most try to get away with as much fakery as possible — whether it's Jennifer Lopez pretending to be in the Bronx when she's actually in L.A., mascara being shilled with the help of false lashes, or merely forcing already-thin models to be even thinner.


Still, the program could be beneficial. Students, teachers, media critics and parents could use the tool to for their own watchdog purposes. And the more we learn to gaze upon "perfect" images with a critical eye, the better. Because even though here, on this site, we talk about digital enhancements and Photoshop of Horrors all the time, there are plenty of folks who consider the images in glossy ads and magazine covers to be "real" photographs — representations of the truth. When, in fact, they are forgeries, compositions, pastiches — a collaborative effort between a model/actress, her trainer, her aesthetician, her plastic surgeon, her makeup artist, the photographer, the camera, the lighting, the photo editor, the editor in chief, and the post-production peeps who do color correction and fiddling. The opening line of this story in the Economist reads, "The pretty people pictured in popular periodicals aren't real." But doesn't the average Jane tend to think so? And, more important: Why aren't those pretty people real? And why do ladymags and advertisers get away with disseminating falsehoods?

Photoshopped or Not? A Tool to Tell [NY Times]
Physical Implausibility [The Economist]
A Perceptual Metric For Photo Retouching [PNAS]
Retouching Examples [Hany Farid]


Kat Callahan

I wonder if black models are closer to 5s. The first picture looks like a difference in make up, George Clooney looks about ten years younger, but is easily recognisable, but that black model looks completely different. Not even like the same person.

White beauty standards = nonwhite models are photoshopped even more?