What the firm's new facial performance capture tech can offer developers
As the experts in this month’s motion capture special discuss, facial performance is becoming more and more important to high quality characterisation in games.
And to convey the best performance, actors need to be untethered from cumbersome technology. While plenty
of full-body capture solutions have concentrated on making themselves lighter and less restrictive, Scottish firm Dimensional Imaging has poured just as much resource and effort into delivering the same experience for facial capture.
Due to be officially launched at this month’s Siggraph in Vancouver, the DI4D Head-Mounted Camera takes the company’s experiences with high-definition performance capture and crams it into a rig that allows actors to move freely while performing.
“Our previous DI4D capture systems have required the actor to sit in front of a fixed array of video cameras,” explains CEO Colin Urquhart. “Our new head-mounted camera system is different as it’s a lightweight head rig that the actor wears.
“It’s fitted with two small, synchronised, high resolution digital video cameras, allowing it to capture high quality stereo video of the actor’s performance. The stereo video streams are then post-processed using our DI4D software. The system is therefore able to combine the high fidelity of DI4D facial performance capture with the freedom of movement and convenience of untethered capture. The result is the capture of incredibly realistic, extremely dense, facial animation.”
Dimensional Imaging is also keen to emphasise that the new, more portable solution is still capable of what the firm refers to as ‘4D facial performance capture’.
“Most facial performance capture systems either capture a single video stream which is interpreted to drive a rigged character model, or capture the 3D trajectories of only a sparse set of discrete facial markers,” says Urquhart.
“By contrast, our 4D facial performance capture systems acquire a ‘3D scan’ per frame using passive stereo photogrammetry and then a dense mesh is tracked through the sequence using optical flow. Every vertex in the tracked mesh then effectively becomes a motion capture marker. The result is much higher fidelity data.”
The head rig is easily adjustable and lightweight for the actors’ comfort. It also features four small LED lights that illuminate the face, helping the DI4D HMC capture data without the need for markers, make-up or special lighting.
“The continuing demand for ever more true-to-life facial animation across the games industry is driving the need for ever higher fidelity facial mo-cap data,” says Urquhart. “However, developers require tools that enable them to acquire this data in cost-effective and time-efficient ways.”
To begin with, DI4D will be running an on-location service, managing the capture and processing of all data on behalf of the developers. Eventually, the firm plans to sell the device itself and licences for their software for studios to use in-house.
Devs can find out more or download sample data on the official website, or by contacting firstname.lastname@example.org.