Faceware Technologies has revealed more details of Retargeter 4.0, the forthcoming update to the solving half of its two-part markerless facial performance capture pipeline.
The software takes an IMPD file generated by Analyzer, Retargeter’s sister package, from footage of an actor’s face, and maps the tracking data to a CG character.
New AutoSolve option
Retargeter 4.0 will feature a number of improvements designed to reduce the time taken to reach a first pass at a facial animation: notably, an AutoSolve feature and Expression Sets.
The former, described by Faceware as “one-button retargeting” can be applied to a character’s entire face, or individual groups such as eye, jaw or brow movements.
The resulting rough solve gives artists actual facial animation to refine, rather than having to begin work from a static character. “It’s a much more intuitive way to decide what needs clean-up,” said Faceware VP Peter Busch.
More automated set-up
The process can be assisted by Retargeter 4.0’s new Expression Sets feature. During set-up, users pose their character to match 45 standard reference expressions, enabling the software to achieve a more accurate solve.
The set covers a range of common expressions seen in dialogue or facial performances (blinks, eye darts, brow and jaw movements). According to Faceware, it typically takes around 20 minutes to recreate.
Other new features include batch scripting via Python and embedded languages like MEL and MAXScript; and the option to store thumbnail images of a character’s face along with pose data, making it easier to browse large data sets.
New features in Analyzer 1.3
Artists will also be able to use Analyzer to export an image sequence of an actor’s face, complete with the landmark tracking points, along with the track data itself.
The resulting snapshots can then be displayed in Retargeter alongside a CG character to make it easier to resolve problems with the retargeting.
Other features in Analyzer 1.3, which will be released alongside Retargeter 4.0, include support for a neutral reference frame to improve the accuracy of a solve; and batch processing and production-management tools.
A quicker way to get 80% there
Faceware describes its technology’s role as to get users ‘80% there’: to accelerate the process of generating a blocking pass of a facial animation that can then be refined by hand.
The company claims that a pass that would have taken six hours to create with Retargeter 3.0 can now be generated in 20 minutes with Retargeter 4.0.
“It’s by far and away the biggest improvement we’ve made to the underlying technology, and to the speed of the workflow [in six years],” commented Busch.
In beta: due out in April
Retargeter 4.0 and Analyzer 1.3 are currently in beta: Faceware will be accepting more users into the beta program at GDC next week. The commercial release is scheduled for April.
Retargeter 4.0 will be available as a plugin for 3ds Max, Max, Softimage and MotionBuilder 2010 and above.
Both software packages remain free: instead, Faceware charges to generate IMPD files from Analyzer – an operation performed on the company’s servers – according to the number of minutes of footage processed.
Read more about Retargeter 4.0 on Faceware’s website
(Includes official press release and additional demo videos)