Deepfake AI Tool Lets You Become Anyone in a Video Call With Single Photo

Deep-Live-Cam
X user Joao shows how he was able to use Deep-Live-Cam to transform into Republican Vice Presidential candidate J.D. Vance (left) and Meta CEO Mark Zuckerberg (right).

A free deepfake AI tool that lets you become anyone during a video call with a single photo is going viral online.

An open-source software package called “Deep-Live-Cam” is gaining significant traction online for its ability to extract a person’s face from a single photo and apply it to another individual during a live webcam video.

According to a report by ArsTechnica, the surge of interest in Deep-Live-Cam led the open source project to briefly take the number one spot in GitHub’s trending repositories list.

Deep-Live-Cam, which is available to download for free, has been in development since late 2023.

However, Deep-Live-Cam has recently amassed popularity due to a series of viral videos circulating on X (the platform formerly known as Twitter) which reveal the software’s capabilities.

Deep-Live-Cam can take one photo of a person’s face and instantly overlay it onto a live webcam video feed. The AI software replicates the individual’s pose, lighting, and expressions on the webcam in real-time.

Although the results may have some flaws, this technology reveals the significant progress in remote deception methods.

In one of the most popular videos circulating online, X user Joao is seen transforming into Republican Vice Presidential candidate J.D. Vance, Meta CEO Mark Zuckerberg, as well as actors Hugh Grant and George Clooney with the help of Deep-Live-Cam.

In another viral clip posted by X user Matthew Berman, a Deep-Live-Cam user immediately morphs into billionaire X owner Elon Musk in real-time on a live webcam video feed.

How Deep-Live-Cam Works

Deepfake technology, which involves swapping faces in videos using AI, has been around since 2017.

According to NewsBytes, Deep-Live-Cam works by detecting faces in both the source and target images and then applying a pre-trained AI model called “inswapper” to perform the face swap.

To improve the quality of the swapped faces and correct any artifacts, another model called “GFPGAN” is used.

The inswapper model, trained on a large dataset of millions of facial images featuring thousands of people with various expressions and lighting conditions, can predict how a person might look from different angles and with different expressions.

However, some online users have commented on how Deep-Live-Cam can be easily misused by deepfake fraudsters. In February, a Hong Kong-based finance worker was scammed into paying $200 million Hong Kong dollars ($25.6 million) to criminals after a virtual meeting with deepfakes.


 
Image credits: Header photo via X/@jay_wooow.
 

Discussion