New deepfake AI tech creates videos using one image

Samsung’s Artificial Intelligence Centre in Moscow has developed new AI software that can generate videos using just one image, known as deepfake technology.

In a paper published in the pre-publication journal arXiv, and in an accompanying video demo, the algorithm creates a video using one still image, such as the Mona Lisa painting or a photo of Salvador Dalí.

The Samsung algorithm was trained using the VoxCeleb database, which has more than 7000 images of celebrities from YouTube videos.

The new AI software is currently only able to produce ‘talking head’ videos from the shoulders up. However, the more images used, the better the quality.

Current AI systems usually require a large number of images of a person’s face to produce a moving picture based on it.

Artificial intelligence at museums and galleries

The Dalí Museum in St Petersburg, Florida is planning a $38 million expansion, with the addition of exhibits that use AI and augmented reality (AR).

Earlier this year, the Dalí Museum announced ‘Dalí Lives’ – a groundbreaking experience that brings Dali “back to life”.

The experience uses deepfake AI and machine learning to create an uncanny lookalike of the artist. Visitors will be able to interact with an engaging lifelike Salvador Dalí on a series of screens throughout the museum.

Elsewhere, China is working to build smart museums with the latest technology, including artificial intelligence (AI), augmented reality (AR), and virtual reality (VR)

Artainment fuses art and tech

Galleries and museums are evolving as they look to keep pace with visitor expectations and rapidly developing technology.

One key trend in the attractions industry is artainment, which fuses art and technology with spectacular results – check out some of the most eye-catching around.