![](https://csdnimg.cn/release/download_crawler_static/10204961/bg13.jpg)
i
i
i
i
i
i
i
i
1.2. A Brief Historical Sketch of F acial Animation 5
In 1987, Waters reported a new muscle model approach to facial ex-
pression animation [Waters 87]. This approach allowed a variety of fa-
cial expressions to be created by controlling the underlying musculature
of the face. In 1988, Magnenat-Thalmann and colleagues also described
an abstract muscle action model [Magnenat-Thalmann et al. 88]. In 1987
Lewis [Lewis and Parke 87] and in 1988 Hill [Hill et al. 88] reported tech-
niques for automatically synchronizing speech and facial animation.
Another groundbreaking animation short was Tin Toy, which received
an Academy Award. Produced by Pixar, Tin Toy was an example of the
capabilities of computer facial animation. In particular, a muscle model
was used to articulate the facial geometry of the baby into a variety of
expressions [Parke 90].
The development of optical range scanners, such as the Cyberware
TM
optical laser scanner, provides a new wealth of data for facial animation
[Cyberware Laboratory Inc. 90]. In 1990, Williams reported the use of reg-
istered facial image texture maps as a means for 3D facial expression anima-
tion [Williams 90b]. By the late 1990s, large data sets of high quality laser
scanned data were being used to create detailed morphable facial models
by Blanz and Vetter [Blanz and Vetter 99].
The new wave of enhanced image processing and scanning technology
promised to usher in a new style of facial animation. In 1993, Lee, Ter-
zopoulos, and Waters described techniques to map individuals into a canon-
ical representation of the face that has known physically based motion at-
tributes [Lee et al. 93].
Another growth area was in medicine, with a focus on surgical plan-
ning procedures and accurate simulation of face tissue dynamics. In 1988,
Deng [Deng 88] and later Pieper [Pieper 91] in 1991, used a finite-element
model of skin tissue to simulate skin incisions and wound closure. More
recently the finite-element approach has been applied to highly detailed
biomechanical models of muscle and skin tissue derived from the Visible
Human Project by Sifakis, Neverov, and Fedkiw [Sifakis et al. 05].
Through the late 1990s there was a surge of interest in facial analy-
sis from video cameras. This interest was twofold: first, to provide the
ability to track the human face to create lifelike characters, and second,
to develop the ability to detect facial expression and thereby derive emo-
tional states. There has been some success in both areas. Two popular
techniques are model-based [Yuille et al. 89, Blake and Isard 94] and opti-
cal flow-based [Black and Yacoob 95, Essa and Pentland 94] techniques.
The late 1990s and early 2000s became a threshold for high-fidelity face
capture and rendering for the film industry. Landmark films such as The
Lord of the Rings (New Line Cinema 2002), The Matrix Reloaded (Warner
Bros. 2003), The Polar Express (Warner Bros. 2004), Monster House (Sony
Pictures 2006) required face motion capture sessions of actors using mark-