2 minutes reading time (339 words)

Fake it 'til you make it: Deep Video experiments impersonate real people

Fake it 'til you make it: Deep Video experiments impersonate real people

If you saw the recent BuzzFeed promotion where Barack Obama talks about... well, all sorts of non-presidential stuff, then you may have wondered how it was produced. It's through a technique called ​Deep Learning​, demonstrated to quite an effect at SIGGRAPH 2018.

The technique reproduces a real face and "takes control" of it, simulating movement and allowing for fictitious events to take place featuring real people. The opportunities for this are quite incredible - in both positive and nefarious ways. Here's the blurb from Stanford University's Professor Michael Zollhöfer, who announces the project as part of a consortium including the University of Bath, TUM, MPI Informatics, and Technicolor.

In contrast to existing approaches that are restricted to manipulations of facial expressions only, we are the first to transfer the full 3D head position, head rotation, face expression, eye gaze, and eye blinking from a source actor to a portrait video of a target actor.

The core of our approach is a generative neural network with a novel space-time architecture. The network takes as input synthetic renderings of a parametric face model, based on which it predicts photo-realistic video frames for a given target actor. The realism in this rendering-to-video transfer is achieved by careful adversarial training, and as a result, we can create modified target videos that mimic the behavior of the synthetically-created input.

In order to enable source-to-target video re-animation, we render a synthetic target video with the reconstructed head animation parameters from a source video, and feed it into the trained network -- thus taking full control of the target. With the ability to freely recombine source and target parameters, we are able to demonstrate a large variety of video rewrite applications without explicitly modeling hair, body or background.

For instance, we can reenact the full head using interactive user-controlled editing, and realize high-fidelity visual dubbing. To demonstrate the high quality of our output, we conduct an extensive series of experiments and evaluations, where for instance a user study shows that our video edits are hard to detect.

Professor Michael Zollhöfer
Digital tumbleweed
AR in retail - almost always, never quite