Samsung AI Center’s Moscow research team have combined with Skolkovo Institute of Science and Technology to publish a new document that shows how terrifying new software can be used to create 3-dimensional talking heads. The images are created via still images. The software’s creations are nothing short of stunning and scary. It can take a regular, common image of anyone and produce what appears to be a video communication from that person. Of course, the implications of such technology are unsettling. This means rival governments will be capable of serving up a completely fake video of politicians. It also means countries such as the United States could leverage such technology for its own purposes.
The researchers published document can be found here. It claims that as little as one image is needed to create the life-like videos. Although, the software tends to work most accurately when given a series of images, rather than a single image. “Crucially, only a handful of photographs (as little as one) is needed to create a new model, whereas the model trained on 32 images achieves perfect realism and personalization score in our user study (for 224p static images).”