NAFO’s Info Ops class presents: How to spot manipulated and deceptive photos

(Viimeksi päivitetty 10/01/2023.)

Article_image-1

Intro

Welcome to NAFO’s Info Ops class. This class is not sponsored by the CIA because there is no CIA. Today you will learn how to check if a photo is deceptive or not. In the context of information war and propaganda a photo can be deceptive because a) it has been manipulated or b) because it’s used in a way that is misleading.

When should you check if a photo is deceptive or fake? Only if it’s posted by evil people who are for example supporting Putin and his attack on Ukraine? No. You should check every photo that you are going to share with others for the purpose of telling them about something important and newsworthy. Even photos you got from good and trustworthy people. Because they might not have checked the photo, especially if it seems to verify things they already believe are true.

Lesia
In December 2022, Lesia Vasylenko (Ukrainian MP) tweeted this photo that she probably thought was the real thing. In reality, it was an old photo from years back that had been used also as a book cover. The Russian propaganda channels now use captures of that tweet to prove Ukrainians are liars.

Real photos, deceptive usage

Let’s start with real photos used in deceptive ways. These photos are for example cropped in a way that leaves important things in the original photo out and makes the photo look like it is telling a completely different story. A deceptive photo can also be the original photo, but it has been bundled together with a news headline or other text (like a tweet) that lies about what the photo is about.

cliff
A cliff in Pedra do Telegrafo, Brazil. It is a popular tourist attraction that is frequently captured in photographs that are taken using deceptive cropping and angles.

So how to check what’s the story behind a photo, its origin story? Your first go-to tool is Google Lens. Copy and paste your image in its search window or select the image from your computer/mobile device (on the computer some browsers let you use Google Lens by right-clicking on an image).

Google Lens will show you images that match yours and it also has a “Find image source” -button. If the image has text on it, Google Lens can translate the text to a language of your choosing. If the image is stolen from some website or it is news or social media famous, Google lens can probably find its origin. Another similar free image search is TinyEye and Russia’s own Google, Yandex (some say it is better at finding people than Google).

Bucha
A photo and the results with the “Find image source” -button. Even if Google Lens does not find the original image, it might find other images related to the same event, like in this example.

Finding a person

If the photo has a person in it and you want to check who that person is, first try the before-mentioned image searches. If they give you nothing, it’s time to go to the (mostly) free search engines that specialize in finding people. Let’s start with a Bellingcat favourite named FindClone (also known as SearchFace and some other names). If you are searching for Russian people this one will let you dig deep into the photos of people using VKontakte (VK), Russia’s number-one social media platform. Unfortunately, FindClone wants you to register with a phone number. Because the site’s origin is as shady as anything else that is Russian, use a number from a burner phone and keep your VPN on at all times.

If you want to find people worldwide, try PimEyes, Betaface or Pinterest image search. They are not as good as FindClone, (probably because they take privacy a bit more seriously than the Russians) but they are a nice addition to the big image search engines.   

Finding a location

One of the best ways to check if a photo is really about what it’s supposed to be is to check where it has been taken. If a tweet with a photo talks about dead Ukrainian soldiers in Bucha but the photo location points to Belarus, you can be pretty sure you are being lied to.

To find out where a photo was taken, you can use its metadata/EXIF data. To make the most out of hidden location data, head over to Pic2Map. If the photo has the info, the site will show you on a map where it was taken. Once you found the location, you can use Wikimapia, Google Maps and Google Street View to find out more about that exact place.

If the photo does not have any metadata but includes for example a big building, statue or mountain, you should try to select that part of the image in Google Lens and see what it can find.

place
The results in Google Lens when searching with a photo of Helsinki Cathedral.

Google Lens is also good a finding smaller things. If a photo has for example a box of cereal in it, Google Lens can probably show you who is the maker of that cereal. Same with furniture, cars and many other things.

I hope these tricks will help you with finding the origin of a photo. Next let’s see how to spot a manipulated photo, one of the most common ways of spreading propaganda.

Photo manipulations

These days it’s hard to find photos that have not been manipulated in any way. If nothing else, the brightness, contrast or colours of the photo has been changed. With photos used for propaganda we are mostly interested in if something has been added or taken out from a photo. To check this, there are some very good services to use for help.

For most photos, FotoForensics will be enough to find out if it has been manipulated. FotoForensics uses Error Level Analysis (ELA) to identify areas within an image that are at different compression levels. With basic JPEG images, the entire picture should be at about the same error levels. If a section of the image is at a significantly different error level, then it could indicate manipulation. FotoForensics also has some nice other features like finding hidden pixels and viewing the photo metadata. If FotoForensics is not enough for you, check out Forensically and Ghiro.

Pixels
The photo on the left looks completely real, but an ELA check with FotoForensics shows us that there is something fishy going on in the middle of the pic.

If the photo manipulation tools are not showing anything, it does not hurt to check images the old-school way. Below is a list of common ways to check a photo.

Check the edges

If something has been superimposed into a scene, you can sometimes tell by looking at the edges of subjects. They might be jagged or way too smooth around the subject.

Check the sharpness of things

Many times when people add things to images, they are of a different resolution than the original image. Because of that, the superimposed object is often sharper or blurrier than the rest of the image.

Check if you can find warping

Often photo manipulators want to make something in their image bigger or smaller. While doing that, the people bad at their job easily edit something else too. Commonly, the background around the edited object or person will be warped/distorted.

Examine the shadows

Mistakes in shadows often reveal when an image has been manipulated. The shadow might be missing, too light/dark or pointing in the wrong direction.

Check for missing reflections

Look for shiny objects and mirrors in the photo. Whoever manipulated the photo might have forgotten to account for the reflections.

Look for signs of cloning

Image editors like Photoshop make it easy to copy (clone) one part of a photo into another part. Check the photo for clones of anything.

Try zooming In

Some errors in manipulated photos can only be seen when you zoom in. What looks real at normal viewing size may clearly appear fake when viewed close up.

Look for inconsistent lighting

Unless there are clearly multiple light sources in a photo, the lighting generally comes from one direction. If people or objects are close to each other in the photo but have light and dark areas on different sides of them, your image might be fake.

Outro

What to do if you happen to spot a photo used in a way that might deceive a large number of people? First, send a message to the person that put the picture live. If they refuse to do anything about it, report it to the service they have posted it to. If this does not help, post the image and a warning text written over it on all your social media accounts. Hopefully, the search engines will pick up on your post and the next time somebody is searching for pictures for their post, they spot your warning next to the original manipulated photo.

That’s all for now. This is the first version of this Info Ops class, I will update this whenever I find something interesting to add to it.

Sources

FindClone.ru: https://findclone.ru/

TinEye: https://tineye.com/

Bellingcat: https://www.bellingcat.com/

Using the New Russian Facial Recognition Site SearchFace: https://www.bellingcat.com/resources/how-tos/2019/02/19/using-the-new-russian-facial-recognition-site-searchface-ru/

Google Lens: https://lens.google/

HOW-TO6 Cool Search Engines to Search for Faces: https://www.groovypost.com/howto/cool-search-engines-to-search-for-faces/

Yandex Images: https://yandex.com/images/

Pimeyes: https://pimeyes.com/

Betaface: https://betaface.com/demo.html

Pinterest: https://www.pinterest.com/

How to Use Google Lens to Identify Landmarks in Your Images: https://smartphones.gadgethacks.com/how-to/google-photos-101-use-google-lens-identify-landmarks-your-images-0184647/

Forensically: https://29a.ch/photo-forensics/

FotoForensics: https://fotoforensics.com/

Hidden pixels: https://www.hackerfactor.com/blog/index.php?/archives/894-PNG-and-Hidden-Pixels.html

Ghiro: http://www.getghiro.org/

Lesia Vasylenko: https://twitter.com/lesiavasylenko

Tweet related to Lesilies tweet with an old photo: https://twitter.com/MB_Rockwell/status/1607820047569031168

El hombre en el olvido (Spanish Edition): https://www.amazon.com/hombre-olvido-Spanish-Christina-McKenna/dp/8415996012

Error level analysis (ELA): https://en.wikipedia.org/wiki/Error_level_analysis

Photograph of Cliff Hanger Isn’t Quite What It Seems: https://www.snopes.com/fact-check/photograph-cliff-hanger-isnt-quite-seems/

11 Ways to Easily Identify Manipulated Images: https://www.makeuseof.com/ways-to-easily-identify-manipulated-images/

Keywords

Send this NAFO Info Ops class to a friend