1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135

6

directly manipulate the DOFs of an object [Mez68, BW71, Csu71, KB84, Stu84, MTT85b,

SB85, Las87].

Later systems allowed the animator to specify the position of specific points on

the objects being animated (such as a hand or foot ) and used inverse kinematics to determine the

appropriate values for the creature's internal DOFs [KB82] [GM85] [BMW87].

Procedural

descriptions of motion, often based on real-world data and observations, can be used to model

very specific classes of movement, effectively "programming" the animated movement [Zel82]

[GM85]. In all cases, the quality of the resulting motion is heavily dependent on the ability of the

animator

who

is

responsible

for

ensuring

that

the

perceived

dynamics

of

the

motion

are

appropriate.

This is a task which requires significant skill.

It potentially distracts the animator

from the primary task at hand, but it also allows him or her complete artistic control.


Rotoscoping and motion capture are techniques commonly used to obtain kinematic data from

real-world sources.

Directly recording a phenomenon to be animated guarantees realistic and

natural-looking motion.

Specialized hardware is generally required,

but

such

equipment

is

becoming more accessible.

A number of problems with this approach make the investigation of

other motion generation techniques desirable.

First, captured motions are limited to real-world

motions that can easily be recorded. Essentially, motion capture has many of the same restrictions

as live actors. Also, approaches to parameterizing captured motions often produce results that are

no longer fully realistic.

Physical constraint violations, such as ground interpenetration and

sliding are common examples of failure.

While solutions to enforcing such constraints for

particular classes of motion have been demonstrated [BMTT90] [KB93], no general solution

currently exists.

More recent parameterization approaches seem oriented toward more broadly

modifying captured motions and are likely to have similar problems [BW95] [WP95] [UAT95].

Finally, captured motions cannot easily be modified to respond realistically to environments

different from the one in which they are obtained. Varying terrain and collisions are two examples

of such potentially desirable changes. As the demand for fully interactive environments increases,

this issue becomes more important.

In recent years the interactive home-entertainment industry

[CONVERTED BY MYRMIDON]