Recently, we get news from foreign media, Samsung AI center in Moscow and researchers Skolkovo Institute of Science and Technology published a new paper detailing generate 3D animated portrait from a single static portraits created. Different realistic portrait with previous artificial intelligence capable of generating photo-AI system, this new technology can produce a motion picture moving portrait talk, although not perfect, but very realistic.
Samsung's new AI functions: static portraits to 3D moving map
the process of creating analytic 
study newly published explained, "real-world scenario" requires system can only use a few or even a single person be generated, rather than a wide range of image data sets. To meet this requirement, the researchers created a system, despite the need to adjust the parameters of tens of millions of dollars, but "based on a small amount of training can quickly complete the picture and only."
By using the generated against the network, the researchers were able to make painted portraits in addition to the images, you can generate the Mona Lisa conversation, the mobile version. As shown in the detailed study of video, final results vary in terms of quality and realistic, and some may be indistinguishable from the real video (at least at low resolution).
Reproduced in: https: //my.oschina.net/u/3008585/blog/3060379