by Thierry Secretan.
as told by BBC Brasil
Eduardo Martins allegedly worked as a UN photographer whose experiences helped him connect with human suffering in some of the world’s most dangerous conflict zones. His work appeared in reputable international outlets such as Getty Images, The Wall Street Journal, Vice and BBC Brasil. In between trips to Mosul in Iraq, the Syrian city of Raqqa under the control of so-called Islamic State (IS) and the Gaza Strip, Eduardo Martins enjoyed surfing. He shared glimpses of his life with his almost 125,000 Instagram followers. Until it all came crumbling down when a BBC Brasil investigation found out that Eduardo Martins was a completely fictitious character.
For years, someone using that name had been stealing pictures taken by professional photographers who had risked their lives in conflict to get them. Eduardo Martins fooled journalists and picture editors by making slight alterations to the images, such as inverting them, just enough to elude software that scans pictures for plagiarism.
Who are we going to trust if The Wall Street Journal, the BBC or Getty Images’ sources become questionable? This is a tsunami which raises once again the question of the veracity of what we look at on our screens and newspaper pages…
In a few lines we are told by BBC Brasil — one of the victims — that by simply flipping images and adding some minor alterations the man acting under the name Eduardo Martins eluded all detection by Getty Images and some of its most vital customers. In addition the label of legitimacy conferred by a huge quantity of followers on Instagram becomes also questionable.
This fake-war-photographer-case comes a few months after the true-documentary-photographer-thief-case — Souvid Datta — who extracted faces from Mary Ellen Mark’s photographs and embedded them in his own work. As easily as Eduardo Martins fooled the heavyweights in the media, Souvid Datta fooled the heavyweights of photography obtaining a Getty grant, a Magnum under 30 Award and a Pulitzer grant. Both forgeries were not detected due to a digital process but by human beings who raised the alarm.
— TIME LightBox (@timepictures) 4 mai 2017
This unacceptable situation also comes two weeks after the weird move of Google who published a paper about its new algorithm able to erase visible watermarks such as those displayed across images by Getty Images and other agencies. Weird because Google collaborated at the same time with Shutterstock in finding and offering a cure to this weapon of mass destruction.
— PetaPixel (@petapixel) 18 août 2017
Who or what is to blame besides the forgers? Google for never showing image metadata? Getty Images for not double checking its sources? The publishers for the same reason? The similarity search used to detect plagiarized images on the web? Or ultimately the photographers whose images were so easily stolen due to lack of protection? In fact all of them. Let me tell you why.
Ironically early September at VISA, the world famous photojournalism festival in Perpignan, I was moderating a lecture called Photographers and the War of Metadata. Fifty persons were in the room.
— Could photographers please raise your hand? Fifteen did.
— Who informs rigorously its IPTC fields? Ten arms down.
No surprise. As chairman of PAJ (Photographers, Authors, Journalists) — one of the main professional photographers’ association in France — I know that for many of us the term metadata amounts to mere copyright and description. But what about those who duly inform the contact, image and conditions fields and use a watermark?
Sorry for you guys. A test conducted by LAMARK in March 2017 over 82,000 photos on more than 120 French media sites indicates that only 3,6% of the photos contained metadata (2,6% were credited). If this figure does not necessarily apply to German, British or American media sites it certainly does to Google Images in even more staggering numbers.
The overwhelming majority of photographs on Google Images have no metadata and, since Google gradually introduced its new image gallery layout early this year in Europe, the media source of a photograph no longer appears in the search results. The consequence is a huge collapse of media sites consultation. This also reminds us that the point of entry on internet is the image, not the word.
Finding an image on the internet and using it for publishing needs has never been easier. However, tracing the source of an image to credit the person who took it or to ask permission to use it becomes ever more difficult due to this loss of source data (photographer, agency, copyright, description, etc.). Truth be told, this information, be it in watermark or metadata form, generally disappears because of some editing modifications made on the internet such as compression, cropping, sharing via social media, screenshots, etc.).
If watermarks and metadata are easily removable and if a plagiarized image eludes detection by similarity search — just thanks to a mirroring — what are the solutions ?
In Europe several companies like Copyright Hub or IMATAG are at work on solutions. First it is important to make a distinction between visible and invisible watermarks. They do not aim at the same purpose. Visible watermarking is not a technical protection mean. Its purpose is to warn that this picture is copyrighted. Adding always the same watermark is a big flaw.
Invisible watermarking of the kind developed by IMATAG — spread all over the image to achieve excellent robustness — is a technical protection mean. It has the same statistical distribution than the visual content of the image to prevent any filtering. Each photo is protected by its own watermark. There is no repetition. If an attacker succeeds to estimate the invisible watermark signal of one photo, this information is useless for attacking other pictures. Even if the image file no longer contains the original metadata, IMATAG’s search engine enables viewers to identify the photographer and source. If the image is flipped (mirroring) the watermark signal remains identifiable. Eduardo Martins could not have stolen your photographs or mine once tagged with such a watermark.
You could legitimately think that I am writing this because I am the CSO of IMATAG. In fact it is the other way round. After using IMATAG as a professional photographer for over a year — having found no alternative watermarking of the type recommandable to PAJ members — I was satisfied by the simplicity of the IMATAG system to tag and send images, the automatic identification of infringers thanks to whom an out of court settlement proved easier to obtain. I widely shared my experiences with other photographers and with publishers and eventually joined the company in May. Late August the Eduardo Martins’ case demonstrated that IMATAG’s choice to develop an invisible/indelible watermark was the right one because it anticipated and addresses the type of attacks and distortions that “Eduardo Martins” cunningly used on stolen images. But that is not enough.
Above all I joined IMATAG for its vision. With 3 billion images shared daily on the web and 86% of these without metadata, photographers, producers of visual content and publishers alike have a major problem. Considering that current image search engines are not professional tools to check a source IMATAG is creating the first public reverse search engine indexing all published images and linking them to sources and metadata. Putting value back into images stripped on the web is the goal of IMATAG. It is a commendable goal and the way to the future.