We may come across claims that include a photo as evidence to back up what is being suggested, though we may question its accuracy. How do we determine a photo’s authenticity? This tutorial will provide an example of how to verify a photo’s authenticity, provenance and accuracy.
Say we come across a claim which includes an image allegedly showing bodies of COVID-19 victims lying in the streets in Italy. We can see the photo shows an aerial view of a street with bodies on the ground as pedestrians walk nearby (see Figure 1).
Figure 1: A claim suggests this is an image of COVID-19 victims, lying dead in the streets, shared on Facebook
First: Is this claim accurate? What part of it can be fact-checked?
To fact-check this image and the situation it portrays, we will try to answer the following questions:
- We want to know where the photo is from. How can we find its origin? Is this the first time the image has been posted online? Does it depict an actual event?
- Can we find out if the photo has been manipulated?
- Who shared the photo? How can we identify the original poster of this image?
- Can we identify the street in the photo? Where is the street located?
But before moving on, we suggest archiving the content that you are fact-checking. You can use tools like Archive.is or Perma.cc to archive the page/post to be able to refer to it later. Should it be deleted in the future, you want something you can go back to, but also have evidence that the claims were made.
Here is how we go about answering the questions posed above:
- How can we find its origin? Is this the first time the image has been posted online? Does it depict an actual event?
Using a reverse image search like Google Reverse Image Search, we can find out whether and where else this image has appeared online. Click on the grey camera icon on the right of the search bar to upload this image or paste in the URL (see Figure 2). The reverse image search finds that the same image has been shared across the web. Particularly, it has been reused on conspiracy theory websites as well as on Facebook in posts with anti-Chinese and Anti-Muslim sentiment.
Based on the previous search, we are also able to identify the original image. It is an Adobe and Reuters stock photo. The image was taken by Kai Pfaffenbach for Reuters on March 24, 2014. The photo shows people taking part in an art project in Frankfurt in remembrance of victims of the Katzbach Nazi concentration camp.
To confirm this, we can do a keyword search for Reuters photographer “Kai Pfaffenbach” and “Frankfurt art project,” and the first result that comes up shows us the Reuters Archive photo.
- Who shared the photo? How can we identify the original poster?
Now that we know where the photo originated, we have to establish the source of the COVID-related image. When was it changed from a photo of a demonstration in Frankfurt in remembrance of the 528 victims of the Katzbach Nazi concentration camp, to an alleged image of deceased COVID-19 victims in Italy?
To answer this question, we upload the photo into TinEye, another popular reverse image search engine. Following results from TinEye, we discover that the earliest use of the photo in connection with the coronavirus was in January 2020. The image was posted to Facebook by a user named “E-man Syg.” The post that accompanied it had an anti-Chinese sentiment and claimed the photo showed bodies on the street in Wuhan, China. It was fact-checked by Suara.com, one of the CekFacta news sites on January 29.
- Can we find out if the photo has been manipulated?
This is an important question to research because we regularly encounter images and videos in our online spaces that have been manipulated or changed to present an alternative reality for various reasons. Sometimes old images are manipulated and redistributed with new captions to present them as a new event.
Foto forensics is one tool to discover if an image you’re researching has been altered. It can identify hidden pixels and metadata details embedded in the image file. We will use the Error Level Analysis (ELA) provided in this tool to examine whether the compression level of the photo in question is consistent throughout the image. If it’s not, it may be a sign that it was modified. As per the tool’s tutorial: “Surfaces should be similar to other surfaces, and textures should be similar to other textures.” After reviewing the Compression Level Consistency map (see Figure 3), we don’t see any major inconsistencies in this photo, which is not surprising since we have already determined that the image is from an actual photoshoot that did happen (see Question 1 above).
- Where is the street located?
We want to confirm the location of the image. The claim suggests it is in Italy, but that may not be the case. This section outlines the steps we could take to determine the claim’s location and to find what street is pictured.
To find this out, we could browse Google Maps endlessly, but since we have a location (Frankfurt) and a date (March 2014) of the original event, it won’t be too difficult to find the street where this art project took place. The photographer’s caption (see Question 1 above) suggests that people were laying down in a “pedestrian zone,” so let’s search “pedestrian zone” and “Frankfurt” in Google (see Figure 4).
The Hauptwache and the Konstablerwache are large shopping and pedestrian zones in Frankfurt. Looking further at photos of the Hauptwache, you’ll notice a red brick building with white round windows on the side. That looks like the building in the photo. It’s a historical baroque building in the Hauptwache. If you wander around using Google Maps Street View, you’ll see the proximity between the two buildings in the claim photo (see Figure 5).
Across the plaza from this baroque building is a beige arched doorway to St.Catherine’s Church, which matches the building entrance in the photo we’re looking for. To double check our aerial view, we can look at the Aerial Photo Database. When we type in “frankfurt hauptwache” one of the photos we get in the second row of results is very close to the same angle as Kai Pfaffenbach’s original 2014 photo.
Based on our review, we determined this photo does not show bodies of COVID-19 victims in the streets of Italy.