Inertial field generator: a method for controllably coupling kinematic character motions to dynamically simulated elements

Abstract

A method is described to let animators control the extent by which kinematically scripted character motions affect dynamically simulated objects' motions. The dynamic objects are connected to the kinematic character, such as clothing or hair, and the motion of the dynamic objects is simulated based on the motion of the kinematic character. Such control is important to produce reasonable behavior of dynamic objects in the presence of physically unrealistic kinematic character motion. An Inertial Field Generator (IFG) is employed to compensate for the unreasonable behavior of dynamic objects when the kinematic character undergoes unrealistic motion.

Description

BACKGROUND OF THE INVENTION

[0001] 1. Field of the Invention

[0002] The present invention relates generally to animation and more specifically to computer animation of characters. The present invention is also directed to providing animators with greater control over aspects of the character to produce reasonable changes in those aspects even when the character undergoes unrealistic motion.

[0003] 2. Description of the Related Art

[0004] Traditional animation techniques allow the animator to create the apparent motion of animated characters, to be viewed by a viewer of the animation. The use of computers to simplify the animation process has provided many benefits to the traditional hand-drawn process. Computer animated characters are well known in the prior art and have been used in many different capacities. Such characters are utilized in traditional movies, videos and online streaming of moving pictures, as well as interactive movies where the motion of characters is often initiated by a user.

[0005] Often times in the animation of characters, the characters have “secondary elements”, such as clothing and hair, that are responsive to main motion of the characters. The motions of some secondary elements in computer graphics imagery are often too complex for an animator to directly control. Instead of a human animator determining the motion of these secondary elements, computer programs use physically-based numerical methods that simulate the motion of these secondary elements (such as hair or cloth) over time.

[0006] This is accomplished by modeling the physical properties of these dynamic elements (how the cloth bends due to forces or collisions with solid objects, how the hair deforms or collides with itself), the external forces on these elements (gravity, wind) and the motions of the kinematic elements (the characters that cloth rests on or that the hair grows out of). The animation of the kinematic elements is provided by the animator and is independent of and unaffected by anything that the dynamic elements do.

[0007] For concreteness, consider an animation of a superhero wearing a long cape: The superhero (the kinematic element, Captain IFG), FIG. 1a, is directly animated by a skilled human animator, while the motion of the superhero's cape (the dynamic element) in response to the superhero's animation is simulated using physically-based numerical techniques.

[0008] If the physical properties and external forces acting on the dynamic elements are accurately modeled, the resulting motion will be plausible and seemingly realistic: cloth will hang down and fold naturally, hair will droop or blow in the wind. However, it is often the case that the kinematic elements (the primary characters of the scene) may be animated in a physically exaggerated manner, yet this requires the dynamic elements to essentially ignore this exaggeration and continue to behave in a realistic manner.

[0009] As an example assume the superhero stands motionless for a moment, then abruptly accelerates upward at 100G's (98,000 cm/s2). See FIG. 1b. His cape, tied around his neck, drapes down his back while the superhero is motionless, but must then react to the superhero's enormous upward motion. The result of the simulation program in this situation is to accelerate the portions of the cape around the superhero's neck along with the superhero.

[0010] The rest of the cape, unfortunately, does not fare so well; the physical properties of the cloth dictate that the superhero will have moved a large distance before the bottom edge of the cape “realizes” that the top edge around his neck has moved. This delay is caused because forces do not propagate instantly through cloth. Equivalently, one can think of the cape suddenly weighing 100 times more than it used to, so it initially stretches to many times its normal length.

[0011] While physically correct, this is not the desired behavior of the cape. Because of this effect, an animator must change the response of the dynamic elements to achieve the desired result. As discussed above, the dynamic elements are oftentimes complex and difficult for the animator to control. This is often frustrating for the animator, since the animator has achieved the proper motion of the kinematic elements and should not have to deal with unrealistic motion of the dynamic elements that are simulated by physically-based numerical techniques.

SUMMARY OF THE INVENTION

[0012] Accordingly, it is a principal object of the present invention to provide a method to allow an animator to control how kinematic elements' motions affects the simulation of the dynamic elements. The present invention provides a compensation for dynamic elements whose motion is unrealistic because of the motion of the kinematic elements and does not require individual manipulation of the dynamic elements on a per frame basis to achieve the desired motions.

[0013] The present invention is directed to a method of simulating relative motion of objects in computer animation. That process includes providing a motion of a kinematic object, providing at least one dynamic object associated with the kinematic object, where motions of the at least one dynamic object are based on the motion of the kinematic object, and selectively manipulating the motions of the at least one dynamic object to simulate physical motion.

[0014] The method selectively manipulates the dynamic objects by compensating for unreasonable motions of the dynamic objects when the kinematic object undergoes exaggerated motion. More particularly, that compensation occurs when the exaggerated motion includes accelerations that are unrealistic for humans.

[0015] In addition, the method can also compensate for the unreasonable motions of the dynamic objects when the kinematic object undergoes accelerated motions above a predetermined limit. Thus, the compensation need not occur until the acceleration of the kinematic object is greater than a certain limit.

[0016] The method is applicable to when the kinematic object is an animated character and the dynamic elements are coupled to the animated character. Additionally, the dynamic elements may represent hair or clothing attached to the animated character.

[0017] Also, the dynamic objects may include a first set of dynamic objects and a second set of dynamic objects and in the step of selectively manipulating the motions of the dynamic object, each set of dynamic objects is manipulated with respect to separate reference points on the kinematic object. In addition, the dynamic objects may selectively manipulated with reference to a plurality of reference points coupled to those dynamic objects.

[0018] The above and other objects, features and advantages of the invention will become apparent from the following description of the preferred embodiment taken in conjunction with the accompanying drawings.

BRIEF DESCRIPTION OF THE DRAWINGS

[0019] FIG. 1 illustrates a kinematic character. FIG. 1a is an illustration of a kinematic character having dynamic elements. FIG. 1b presents the illustration of FIG. 1a when the kinematic character undergoes unrealistic motion.

[0020] FIG. 2 is a flow chart illustrating the general method of the present invention.

[0021] FIG. 3 is a schematic showing the connections between a kinematic element and dynamic elements, and the response of the dynamic elements based on the acceleration of the kinematic element.

[0022] FIG. 4 schematic showing the connections between a kinematic element and dynamic elements, and the application of the inertial field generator to the dynamic elements to compensate for the reaction of the dynamic elements to the acceleration of the kinematic element.

DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0023] The present invention is directed to a method for controllably coupling kinematic character motions to dynamically simulated elements. This control is accomplished through the use of an Inertial Field Generator (IFG). The use of the IFG allows an animator to control how kinematic element's motions affects the simulation of the dynamic elements.

[0024] In the context of the present application, kinematic and dynamic objects, or elements, can be, in the preferred embodiments, collections of data points having connections between the data points. These collections of data points are used to form elements of computer animation images. These data points are entered into a computer system and those data points are used to form the images. Through manipulation of the data points, the images are changed on a per frame basis to simulate motion. The kinematic objects described herein are preferably characters that are animated and interact with other elements in the animation scene.

[0025] In the superhero example illustrated in FIGS. 1a and 1b, an IFG is used so that the cape would “feel” only a small fraction (or even none at all) of the superhero's upward acceleration, and would thus not stretch appreciably. Thus, the use of the IFG would allow the superhero's cape to appear as it does in FIG. 1a, even though the superhero is being simulated as moving through unrealistic motion.

[0026] Another example would be a character wearing a shirt who jumps off a cliff. In that case, the animator might exaggerate the character's motion so that the fall is faster than gravity allows for. Yet, visually the animator may wish to avoid the shirt flying up over the character's head as would realistically happen. Likewise, a fur-covered character might wave its arms extremely energetically, but the animator may want the dynamic fur's motion to reflect only some portion of this vigorous motion.

[0027] The methodology employed to apply the IFG will now be discussed with respect to FIG. 2. FIG. 2 illustrates the general method used. First, the motion of a kinematic object is provided. This is supplied by the animator and is usually provided on a per frame basis. Dynamic objects are associated with the kinematic object, where the motions of dynamic objects are based on the motion of the kinematic object. The kinematic object can be an animated character and the dynamic objects are attributes of that character, such as hair and/or clothing. Lastly, the motions of the dynamic objects are selectively manipulated to simulate physical motion. The IFGs allow the motions of the dynamic objects to be controlled so that their motion does not become unrealistic even when the kinematic object moves in ways that are not possible in the unanimated world.

[0028] A common component of all physically-based simulation is that forces acting on the dynamic elements are part of the input to the simulation. An IFG is a specifically constructed force-field that is added to the simulation, in such a way as to compensate for unwanted accelerations in a kinematic character's animation. An IFG works by measuring acceleration in the kinematic character, and adding some portion of that acceleration to the dynamic elements. Essentially then, an IFG is a coupling device between a kinematic character, and a dynamic element whose motion is likely to be highly dependent on the kinematic character's motion.

[0029] For simplicity, it is assumed that the dynamic element consists of n particles, with the ith particle having mass mi. Such dynamic elements consisting of particles 22 are illustrated in FIGS. 2 and 3. While the examples provided herein discuss the representation of the dynamic elements as a series of particles, IFGs that act on continuum objects are just as easily defined. Each particle i in the dynamic element is associated with a predetermined motion of a reference point piεR3, that is, at time t, the reference point for the ith particle has location pi(t). In FIGS. 2 and 3, the reference point is denoted by element 21.

[0030] The motion of the reference points defines an acceleration field, some portion of which the animator wants to pass along to the dynamic particles. Thus, for each particle i a filtering function Gi is defined whose input and output is an acceleration in R3. Given these definitions, at time t the IFG imparts on particle i a force fi(t) defined by:

f i (t)=miGi({umlaut over (p)}i(t))  (1)

[0031] where 1${\ddot{p}}_i\left(t\right)=\frac{d}{d{t}^2}{p}_i\left(t\right)$

[0032] is the acceleration of pi. This force is illustrated by the white arrows 23 shown in FIG. 3. Examples of different IFGs and their use will now be described.

[0033] In a first embodiment, IFG uses a single reference point. In the example of the superhero, FIG. 1, assume that the superhero is accelerating upward abruptly. His cloth cape consists of a number of dynamic particles. For each particle, the same reference point Pi=Pc is chosen, where Pc is the location of a point approximately centered in the superhero's body.

[0034] If G, is defined as:

Gi({umlaut over (p)}c(t))=α{umlaut over (p)}c(t)  (2)

[0035] then for α=1, the motion of pc is completely canceled out. That is, the entire cape feels an accelerating force that exactly matches the motion of pc over time. When the superhero accelerates upward at 100G's, a compensating 100G upward acceleration is imposed on the entire cape. As a result, the cape does not stretch appreciably—in fact, it does not react to the superhero's upward bound except to move exactly upward with the superhero. By varying α, the amount of the superhero's acceleration that is imparted to the cape can be controlled.

[0036] Additionally, the output force from an IFG need not act in the same direction as the input acceleration. For example, if we wish the IFG to only affect vertical accelerations, then we could write

G i ({umlaut over (p)}c(t))=α({umlaut over (p)}c(t)·{circumflex over (z)})){circumflex over (z)}  (3)

[0037] where {circumflex over (z)} is a unit vector along the vertical axis.

[0038] IFGs can also be more selective in their filtering ability. For example, suppose the animator wants the IFG to only compensate for accelerations that are unrealistic for humans. A normal human is limited to accelerations of only a few G's (say three): so that one might wish to compensate for the superhero's acceleration only to the extent that it exceeds 3·980=2940 cm/s2. Setting αM=2940, the filter can be written: 2$\begin{array}{cc}{G}_i\left({\ddot{p}}_c\left(t\right)\right)=\left\{\begin{array}{cc}0& \mathrm{if}\&LeftDoubleBracketingBar;{\ddot{p}}_c\left(t\right)\&RightDoubleBracketingBar;<a_M\\ \left(1-\frac{a_M}{\&LeftDoubleBracketingBar;{\ddot{p}}_c\left(t\right)\&RightDoubleBracketingBar;}\right){\ddot{p}}_c\left(t\right)& \mathrm{if}\&LeftDoubleBracketingBar;{\ddot{p}}_c\left(t\right)\&RightDoubleBracketingBar;\ge a_M\end{array}\right\}& \left(4\right)\end{array}$

[0039] For input accelerations {umlaut over (p)}c(t) with magnitude less than αM, the output force is zero since Gi is zero. As the magnitude of {umlaut over (p)}c(t) reaches and exceeds αM, the IFG begins outputting a greater and greater fraction of {umlaut over (p)}c(t) onto the cloth. In this way, unrealistic motions of the kinematic character that cause problematic effects in the dynamic elements can be avoided.

[0040] In another embodiment, an IFG may use multiple reference points. Greater flexibility can be achieved by allowing each pi to track a different point on the kinematic character. Consider a character with long dynamic hair on its arms. The character moves it arms in different directions very energetically. To prevent the hair on the arms from stretching due to high accelerations of the arms, an IFG is employed.

[0041] However, since the right arm might move independently from the left arm, the hair on the right arm has to use a different reference point from the hair on the left arm. In fact, for this example, all the points in a given strand of hair should use pi(t)=pr(t) where pr(t) is the root point of the hair (specifically, the point on the body that the hair is attached to). In general, each strand of hair is coupled to the motion of a different point pr(t). Obviously, the motions of points on the left arm can be quite different than the motion of points on the right arm. Using multiple independent reference points allows an IFG to take into account simultaneous motions by parts of the body that differ greatly in their direction and or intensity.

[0042] In another embodiment, the IFG uses a varying filter function. In the examples above, each filter function Gi was the same. If one wished the response of the IFG to vary over a character, it is a simple matter to let the filter function Gi vary from point to point. This might be done if the animator wished to smoothly vary the effects of the IFG over a character. For example, an animator might wish to have the effects of an IFG applied to the furry character's arms, but taper off in intensity for the fur on the torso.

[0043] IFGs greatly reduce excessively unrealistic motions from dynamic elements. Even when a kinematic character's motion is completely plausible, the IFG is still invaluable because it allows the animator to control how much of the character's base motion is transferred to the dynamic element. IFGs are easily applied to any sort of dynamic element (including, but not limited to, dynamic clothing, hair, fur, appendages, water, and vapor).

[0044] Although the embodiments discussed above provide for filtering function with certain parameters, the filtering functions Gi can be specified in terms of an arbitrary number of control parameters. Additionally, both the filtering functions Gi and the reference points can be functions of which dynamic particles they apply to. Thus, individual portions of the dynamic elements can be controlled to a fine degree.

[0045] Although embodiments of the present invention have been described in detail, it will be understood that the present invention is not limited to the above-described embodiments, and various modifications in design may be made without departing from the spirit and scope of the invention defined in claims.

Claims

1. A method of simulating relative motion of objects in computer animation comprising the steps of: providing a motion of a kinematic object, where the kinematic object is an element of a computer animation display; providing at least one dynamic object associated with said kinematic object, where said at least one dynamic object is another element of the computer animation display and where motions of said at least one dynamic object are based on the motion of the kinematic object; selectively manipulating the motions of said at least one dynamic object to simulate physical motion; and displaying the elements of the computer animation display, including associated motions of said elements.

2. A method of simulating relative motion of objects according to claim 1 wherein said step of selectively manipulating comprises compensating for unreasonable motions of said at least one dynamic object when the kinematic object undergoes exaggerated motion.

3. A method of simulating relative motion of objects according to claim 2 wherein said exaggerated motion comprises accelerations that are unrealistic for humans.

4. A method of simulating relative motion of objects according to claim 2 wherein said step of selectively manipulating comprises compensating for the unreasonable motions of said at least one dynamic object when the kinematic object undergoes accelerated motions above a predetermined limit.

5. A method of simulating relative motion of objects according to claim 1 wherein said kinematic object is an animated character and said at least one dynamic object is coupled to the animated character.

6. A method of simulating relative motion of objects according to claim 5 wherein said at least one dynamic object is a representation of hair attached to the animated character.

7. A method of simulating relative motion of objects according to claim 5 wherein said at least one dynamic object is a representation of clothing attached to the animated character.

8. A method of simulating relative motion of objects according to claim 1 wherein said at least one dynamic object comprises a first set of dynamic objects and a second set of dynamic objects and said step of selectively manipulating the motions of said at least one dynamic object comprises selectively manipulating motions of said first set of dynamic objects with respect to a first reference point on said kinematic object and selectively manipulating motions of said second set of dynamic objects with respect to a second reference point on said kinematic object.

9. A method of simulating relative motion of objects according to claim 1 wherein said at least one dynamic object comprises a plurality of dynamic objects coupled to a plurality of reference points on said kinematic object and wherein said step of selectively manipulating the motions of said at least one dynamic object comprises manipulating the motions of each of said plurality of dynamic objects with respect to said plurality of reference points coupled thereto.

10. A method of simulating relative motion of objects according to claim 9 wherein said kinematic object is an animated character and said plurality of dynamic objects are coupled to the animated character and said plurality of reference points are different points on the animated character.

11. A method of simulating relative motion of objects according to claim 9 wherein said step of selectively manipulating comprises compensating for unreasonable motions of said plurality of dynamic objects when the kinematic object undergoes exaggerated motion.