Tool Reveals Which Celebs, Models Have Been Photoshopped

Researchers have finally come up with the answer to a question that’s plagued us all for years: “Just how much Photoshopping did that magazine cover model get?”

Researchers at the Department of Science at Dartmouth College have developed a software tool that can rate photographs based on how much they have been digitally altered with programs such as Adobe Photoshop. The proposed tool is part of an effort to bring truth to advertising and restore the perception of natural beauty.

“If the tool is used in the future, magazines and advertisements could have a warning label on images similar to a nicotine ad,” Dr. Hany Farid, a professor of computer science and digital forensics at Dartmouth College, told Mashable. “There could be a number at the bottom of the photo revealing just how much it’s been altered.”

Farid and Eric Kee, a Ph.D student in computer science at Dartmouth, published their research this week in the journal National Academy of Sciences.

The tool would work on a rating scale of one to five. Farid and Kee created a base metric by analyzing and statistically measuring results from various before-and-after photos. They then correlated these findings with a study group that was asked to rank the amount of photo alteration on a scale of one (very similar) to five (very different). This numbered metric could then be algorithmically applied to photos of, say, celebrities and models to reveal just how much photo-manipulation took place.

“The ubiquity of these unrealistic and highly idealized images has been linked to eating disorders and body image dissatisfaction in men, women, and children,” the paper said. “In response, several countries have considered legislating the labeling of retouched photos.”

Although some countries in Europe such as Norway and France have already proposed the concept of labeling altered images, it has yet to be approved. “If legislation to label altered photos is passed then it will be important to have the right technology to determine the extent by which a photo has been altered,” Farid said.

“If a magazine is only publishing pictures with ratings of five and you’re a model with a high score, there may be incentive for editors to lean back on excessive altering and not be so extreme,” Farid said.

But the rating system would only work if all parties were on board, from the publishing world to legislation. “It’s not something you want to rush into,” Farid said. “Everyone would need to be in agreement and the right metrics and technology would have to be in place.”