Leave it to the braniacs at Disney Research to pioneer an automated way to blend the facial expressions from two takes into one new performance.
It's been a solid year since we've last covered new technology from Disney Research. Back during the Holiday months of 2014, we featured two intriguing examples of Disney Research's future technologies: automatic multi-camera editing and making high dynamic range content work with conventional non-HDR televisions.
Now, we're thrilled to update our readers on the next exciting innovation from the Mouse (and this one is a doozy). FaceDirector is a program that can analyze an actor's facial performance and, using facial and audio cues, remap the timing so the two takes are synchronous. The software would be impressive if its functionality ended there, but you can also blend the two performances together. Of course, there is an adjustment, allowing you to vary the degree of the blend, but, for the most part, the process is hidden from the user (in other words, magical).
There are some limitations. For extreme difference in performance, there seems to be some weirdness around the morphing of teeth, as seen at the 3:45 mark of the video below. But the video also demonstrates how well the programs blends even handheld takes and performances with line variations. Check out the video below and let us know what you think about the FaceDirector in the comments!
[Thanks to Engadget for alerting us to this story.]