Skip to main content
Author: Patrick Wong

Detecting fake images

Updated Tuesday, 25th February 2020
Is seeing believing, or have we all become a bit cynical with the rise of fake news? Read on as Patrick Wong shares some tips on how to detect a fake image.

A moment ago, my smartphone buzzed and I received a social media message with a beautiful image from a friend. The image shows a cute little girl happily hugging a large lion, as shown below. It is a touching image but isn’t it too dangerous to hug a wild animal?

Would you let your daughter take part in a photo like this? I normally wouldn’t, but I would this time as I know the girl was not really hugging the lion. It is a fake image. The lion was superimposed onto the photo of the girl.

Photo of a young child (girl) hugging a roaring lion in the wild (Fake imagery) A digital manipulated image showing a girl hugging a lion.  

The idiom seeing is believing is no longer true in this digital era. With the proliferation of powerful camera phones, which often equipped with multiple cameras, they give everyone chances to take, edit and create spectacular images without using a professional camera and image editing software. Furthermore, with a few touches and wipes, the images can be shared with many people in seconds.

Photo of the Loch Ness monster (Fake imagery) The hoaxed photo of the Loch Ness monster Why do people distribute fake images?

In the above case, it is just a friend who wanted to share a photo she found interesting. However, fake images have been distributed for various reasons, ranging from benign purposes such as to impress friends to malicious motives such as to influence people’s political view and blackmail.

The impact of fake images on society can be very significant. It can make fake news more convincing. One well-known example is the hoaxed photo of the Loch Ness monster. The photo has misled thousands (if not millions) of tourists, believers and scientists visiting the loch. Another example is the Cambridge Analytica saga, embroiled in reportedly extracting Facebook users’ data for companies seeking to influence national politics. 

Characteristics of fake images

Fake images can be defined differently. For the purpose of this article, I consider fake images as those that have been manipulated such that they can deceive people in believing something untrue.

These images usually feature things, scenes or people’s behaviours that are unbelievable but look real. Because of the outrageousness or spectacular nature of these images, they can be spread rapidly through social media online and attract lots of attention.

Make a fake image yourself

To experience how a fake image is produced, follow the link below to the’s website, which allows you to use artificial intelligence to create fake images of celebrities by swapping their faces, and create one or more fake images.

The website allows you to upload a photo of a person and swap their face with one from a celebrity. However, I would not recommend you to do so as we do not know how your uploaded images will be stored and used by the website. LINK:[P1] 

Similar technologies have been used to make fake videos of politicians speaking something outrageous and fake pornographies of celebrities (Lee 2018) 

Detecting fake images

Although fake images can look very realistic, there are sometimes tell-tale signs that may help us to discover the deception. However, detecting fake images is a very complex business and many new kinds of research are being undertaken.

To spot a fake image, sometimes you just need common sense

A fake image often looks unbelievably and extraordinarily spectacular. When being presented such an image, we should ask ourselves whether the contents of the image can be real or whether it agrees with the character of the subject. This often needs lots of common sense and judgement.

Continuity of edges

When a part of an image is modified, it is likely the colour, brightness and texture around the edges of the modified part do not match with that of the original image. By carefully inspecting the image, it may be possible to detect the non-continued edges.

Noise analysis

The noise I refer here is not the unwanted audio sound, but the tiny small defects produced by cameras. Different cameras (or more strictly speaking the imaging sensors) produce different noises under different conditions. When images captured by different cameras are superimposed over one another, the noise patterns and levels of each image are likely to be different

By carefully analysing the noise pattern of the superimposed image, it is possible to detect where images are superimposed together.

Luminance Gradient

When a light source shines a light on an object, the luminance (brightness) of the object and its surrounding should be similar. If an image of an object is superimposed into another image, their luminance is likely to be different.

By carefully analysing the luminance differences of the superimposed image, it is possible to detect where images are superimposed together.

Spot a fake image yourself

It is very difficult to analyse and detect signs of a fake image without tools. To give you an experience of analysing a manipulated image, follow the link below to the Forensically’s website, which provides a set of useful tools for analysing an example manipulated image.

You can access the help page through the Help menu at the top left corner of the page. The help page briefly explains what each tool does.

Read the Help page before you start using the tools to analyse the image. Try and find any suspected manipulations in the example image -[P2] 

Seeing is no longer believing

As the technologies on detecting fake image advances, the same happens to the technologies for generating fake images. It is a race of two big forces.

In recent years, artificial intelligence has been used to automatically generate fake images that look almost flawless. The technologies have also been made available to the general public and made so easy to be used that anyone can generate an image almost effortlessly. The real impact of these technologies on society is to be seen. For sure seeing is no longer believing.

If you are interested in creating and editing digital images, check out the Open University’s Digital Photography module.



Kleinman, Z. (2018) Cambridge Analytica: The story so far. BBC News UK [online]. Available at: (accessed on 9/12/2019)

Lee, D. (2018) Deepfakes porn has serious consequences. BBC News UK [online]. Available at: (accessed on 9/12/2019)



Become an OU student


Ratings & Comments

Share this free course

Copyright information

Skip Rate and Review

For further information, take a look at our frequently asked questions which may give you the support you need.

Have a question?