New Software Creates 3D Face Models with Smartphone
Face scans are extremely useful. They are used in advanced bio-metric security, and they are used to create hyper-realistic video game and animation characters. Entertainment is using face-scanning tech even more. A new trend in Los Angeles, VFX and gaming companies pay emerging actors to get their faces scanned and then they use their scans as "background actors" in various project. Scans often require hours of work, and expensive equipment. Now, a team at Carnegie Mellon University have developed a way to complete a 3D scan with only a smartphone.
Using the camera at 120 frames per second (or "slow motion") the phone can acquire enough points on one's head to create a scan with sub-millimeter accuracy - much more accurate than any other camera-based scanning system. This is important for most applications, as explained by Simon Lucey - professor at the school's Robotics Institute.
"Building a 3D reconstruction of the face has been an open problem in computer vision and graphics because people are very sensitive to the look of facial features... even slight anomalies in the reconstructions can make the end result look unrealistic."
The steps are simple for an effective scan.
First, 20 seconds of slow motion video are taken of the subject.
Second, using visual simultaneous localization and mapping (SLAM) the software creates a map of the face using the video.
Third, deep learning algorithms find the key components of the face - eyes, nose, ears - and then fill in and remove extra data.
It takes over 30 minutes to execute, but it can be completed with only a smartphone.
Read more at the school's website.