[其它] Artist Friendly Facial Animation Retargeting

[复制链接]

晃晃

1023 主题	3 听众	359 积分

设计实习生

Rank: 2

纳金币: 335582
精华: 0

电梯直达

楼主

发表于 2011-12-28 10:34:50 |只看该作者 |倒序浏览

Artist Friendly Facial Animation Retargeting

Yeongho Seol J.P. Lewisx Junyong Noh

Jaewoo Seoy Weta Digital KAIST

Paul Hyunjin Kimz KAIST

Abstract

This paper presents a novel facial animation retargeting system

that is carefully designed to support the animator’s workﬂow. Ob-

servation and analysis of the animators’ often preferred process

of key-frame animation with blendshape models informed our re-

search. Our retargeting system generates a similar set of blend-

shape weights to those that would have been produced by an anima-

tor. This is achieved by rearranging the group of blendshapes into

several sequential retargeting groups and solving using a matching

pursuit-like scheme inspired by a traditional key-framing approach.

Meanwhile, animators typically spend a tremendous amount of

time simplifying the dense weight graphs created by the retarget-

ing. Our graph simpliﬁcation technique effectively produces ed-

itable weight graphs while preserving the visual characteristics of

the original retargeting. Finally, we automatically create GUI con-

trollers to help artists perform key-framing and editing very efﬁ-

ciently. The set of proposed techniques greatly reduce the time and

effort required by animators to achieve high quality retargeted facial

animations.

CR Categories: I.3.7 [Computer Graphics]: Three-Dimensional

Graphics and Realism—Animation;

Keywords: Animation, Face, Retargeting, Animator, Editing

1 Introduction

Creating facial animation that is both realistic and emotionally com-

pelling continues to be a challenge. Increasingly, the generation of

realistic facial animations starts with retargeting results from mo-

tion capture. This approach has been used for recent high-quality

character animations such as the ones in the ﬁlms King Kong, The

Curious Case of Benjamin Button, and Avatar. However, these are

extremely labor intensive efforts. Although motion capture and

retargeting are sometimes naively assumed to be automated pro-

cesses, as a ballpark ﬁgure current movie projects can require up

to a day of manual editing for each second of ﬁnished retargeted

animation.

Most current approaches do not produce facial animation retarget-

ing results that are directly usable for high quality applications.

Even in a relatively simple case of identical source and target

faces, the animator’s participation is inevitably needed to achieve

a high quality animation as evidenced by The Digital Emily Project

[Alexander et al. 2009]. Havaldar [2006] described the reasons why

manual editing is always necessary:

(a) The combination of artistically designed blendshapes cannot

perfectly match the actor’s motion.

(b) The proportions of CG face model and an actor’s face can be

signiﬁcantly different.

(c) Motion capture marker placements differ from day to day.

(d) The desired performance is not what the actor performed. Ei-

ther the required expression is not present in the motion capture

data or it needs to be exaggerated.

全文请下载附件：