KR20080028732A - Pipeline architecture and method for digital character motion making - Google Patents

Abstract

A system and a method for creating digital character motion are provided to create the motions of characters automatically by using an instinctive script such as a scenario to reduce the amount of producing a proto-type animation about a computer animation, thereby improving the entire work productivity and improving the quality of the computer animation. A motion analyzer module(3) analyzes a script(2) characterized by a grammatical expression and a logical structure for motion creation, inputted by a user(1), to enable a motion command to embody instructions for reflecting an individual characteristic and a current feeling state applied to a character in the script. A motion organizer module(5) analyzes the result of the motion analyzer module to divide the result into parallel motion and sequential motion and adjust the motion. A motion scheduler module(6) analyzes motion commands, applied by the result of the motion organizer module, to manage the motion commands so that the motion commands can be completed. A motion adaptation module(7) enables the motion commands, applied by the result of the motion scheduler module, to instruct a motion change method according to an adaptation method for surroundings. A motion performance module converts middle motion data corresponding to the motion commands, applied by the result of the motion adaptation module, into result motion data and applies motion to the character.

Description

Pipeline Architecture and Method For Digital Character Motion Making}

1 is a structural diagram showing a configuration of a motion generation pipeline of a digital character and a flow of generation of motion data.

2 is a table showing a state classified according to the level of the digital character motion according to an embodiment of the present invention.

<Brief description of symbols for the main parts of the drawings>

1: user 2: script

3: motion interpreter 4: application motion database

5: motion planner 6: motion planner

7: motion adaptor 8: compound motion database

9: Environment Database 10: Action Performer

11: motion data database 12: result motion data

13: character skeleton

The present invention relates to a digital character motion generation pipeline used for computer animation, and more particularly, to a digital character motion generation pipeline structure for defining and representing a pipeline for motion generation of a digital character used in computer animation. It is about a method.

The main task in computer animation is the production and presentation of digital characters. Once a character is produced, it is distinguished from traditional animation in that it can be recycled, but in the end there is no change in the manual processing of the expression of the character throughout the animation.

While characters of traditional animation exist only as images, the digitized characters themselves have skeletons and skins that cover the skeletons similar to real humans. The motion expression of digital characters is to control this skeletal structure. In traditional animation, drawing the character images of each frame was the role of the animator. In digitized animation, the role was changed to determine the location and orientation of each segment (Bone or Segment) that forms the character skeleton of each frame.

All of the motion representation of the digitized characters can be done manually, but much of the work can be generated automatically by the computer. This content increases the productivity of professional animators, lowers the threshold of animation production, and provides the opportunity for various works to be produced. Allows you to concentrate on performing your work.

There are several ways to automatically generate the motion of a digital character. Next, we will describe some ways to automatically generate the behavior of a character.

<Key frame>

The name keyframe system comes from the "Hierarchical Production System" originally developed by Walt Disney. This system allows a skilled animator to design a sequence of specific frames by drawing frames to animate a keyframe. Lack of skill animators then complete the entire frame sequence by creating intermediate frames. This work replaced the task of drawing intermediate frames using interpolation using a computer. This task is also more generalized by providing a high level of control by allowing interpolation of all parameters affecting motion. Almost all animated movies and computer animations are produced using this technique.

Kinematics Animation

This method expresses motion using segment position and velocity acceleration irrespective of the force generating motion. This method is divided into two parts.

* Forward Kinematics- X = f (Q)

The movement of every segment of a character is determined by the animator. End movements, such as hands and feet, are determined by the accumulation of values associated with all segments that exist from the center to the end of the character. That is, the position X of each segment is determined by the rotation value Q of the intermediate segments.

Inverse Kinematics- Q = f ^-1 (X)

When the animator determines the position of the distal end, it automatically calculates the rotations of the middle segments. That is, when the position X is determined, the rotation value Q is determined.

In the case of inverse kinematics animation, it is very intuitive to determine the end position, which is very convenient for the animator to deal with, but in the case of kinematic animation, it is very intuitive to predict the future motion by adjusting the speed and acceleration.

Dynamic Animation

This method expresses a motion by considering all the forces that generate the motion. And, like kinematic animation, it is divided into two parts.

* Forward Dynamics- F = MA

The force given to the segment is calculated by giving the position, velocity, and acceleration values for the segment.

Inverse Dynamics-A = F / M

When you apply a force to the segment, you find the position, velocity, and acceleration values for the segment.

Partial dynamics may be used to enhance reality by using only gravity effects.

<Motion Capturing>

It uses technology to record the motion of a real human actor as data and to generate motion from this data. In the method of acquiring human motion, optical recognition, magnetic sensor or ultrasonic sensor is used. The disadvantage of this method is that you cannot control the motion in the middle of the motion creation. In other words, the present invention is not merely an operation control method but merely a replay of an operation. If the desired action has not been obtained, then action creation is impossible. To address this, there are attempts to transform the motion data captured by "Motion Transition" little by little to show other similar motions.

Some of the features of the motion creation pipeline applied to avatars similar to digital characters are described in the paper ("The real-time motion generation of human avatars" Proc. Of 13 ^th symposium on Human Interface, 1997, pp435-438) (" The Avatar's Behavior and Interaction for Virtual World "Proc. Of VRSJAC'97, pp242-249).

Therefore, there is a need to create a character's motion more quickly and conveniently by constructing a standardized pipeline for generating the motion of the digital character in a plurality of methods for automatically generating the motion of the digital character described above.

The present invention has been proposed to meet the above-mentioned needs, and its object is to use a script written as a scenario for generating motions written by the user instead of a method of generating all the motions by controlling the segment of the character one by one. It is to provide a digital character motion creation pipeline structure and method that automatically generates the motion of characters easily.

Another object of the present invention is to provide a digital character motion generation pipeline that handles the transformation and generation of data from the higher language used by the user to the lower motion generation.

It is another object of the present invention to achieve various intentions and purposes in everyday motions and everyday motions in order to express various aspects of the character motions. It's about providing a pipeline of digital character motion creation that's been structured to make an impact.

The motion generation pipeline structure for the digital character of the present invention analyzes a script characterized by a grammatical expression and a logical structure for motion generation input from a user, so that motion commands reflect individual characteristics or current emotional state given to the character. Analyzes the motion interpreter module that embodies the instructions in the script, the motion composer module that separates and adjusts parallel and sequential motions by analyzing the results of the motion interpreter module, and analyzes the operation commands given in the results of the motion composer module. A motion planner module that manages the motion commands so that the motion commands can be completed, a motion adaptor module for instructing how motion commands given in the results of the motion planner module are adapted to the environment, and Corresponds to the operating instructions given in the output of the adaptor module. Characterized in that made in fact include the operation suhaenggi applying operation after converting the operation data into the intermediate operation result data character.

The method for generating a motion generation pipeline for a digital character of the present invention includes an analysis step of interpreting a motion to a character by referring to an application motion database in which individual characteristics given to the character or current emotional state information are stored, and a character to process a motion command. The composition step of constructing the character motion by referring to the complex motion database in which necessary rule information is stored, the adaptation step of adapting the character to the surrounding environment by referring to the environment database in which the information about the environment is stored, and for generating the result motion data. And generating a result motion data by referring to the motion data database in which the intermediate motion data are stored.

Hereinafter, the digital character motion generation pipeline of the present invention will be described in detail with reference to the accompanying drawings.

1 is a structural diagram showing a configuration of a motion generation pipeline of a digital character and a generation flow of motion data.

Referring to Figure 1, the motion generation pipeline proposed in the present invention is a hierarchical structure of three stages (lower stage is element operation, intermediate stage is complex operation, upper stage is application operation) in order to satisfy the above requirements. Configure and describe each step in more detail.

<Element operation>

This refers to the basic movements that can be expressed by the character, and more specifically the movements that can be represented by a single 'verb' in grammar. That is, 'walk', 'run' and 'sit' belong to this category.

In grammar, verbs are divided into intransitive verbs and transitive verbs, which have a meaning in itself, such as 'walk'. In contrast, transitive verbs rely on something different, so one motion can be expressed in various ways. For example, a movement meaning 'to dance' may include various movements depending on the type of dance. In the case of 'opening something', there may be an infinite way of opening depending on the kind or location of something. 'Sit' may be an action to sit in place as an intransitive verb, but if the suffix is attached to the chair, there will be various ways to sit as before. Although the operations represented by the transitive verbs seem complicated, it can be seen that they can be expressed by the selection and modification of the intransitive verbs.

Element behavior is defined as a set of automatic private behaviors and other passive behaviors implemented by processing these behaviors in a manner of selection and modification.

<Composite action>

While element operations can only do one thing at a time, combining them can do a wide variety of common tasks. For example, you can do things like "go out and do gymnastics" using "walk" and gymnastic moves. (Motion Composition)

If element movements are completely independent of each other, only one movement can be performed at a time. In reality, however, the walk and wave movements are performed together to do things like "walk and shake hands." Also, if the 'walk' action and the 'turn direction' action are separate from each other, you must repeat one step, change direction, take another step and turn. However, if a digital character can perform two or more element movements together, it is only necessary to walk a little bit in a moment while walking. Therefore, it should be possible to perform two or more element operations together at a time. Of course, there are combinations of actions that can be performed simultaneously such as 'walk' and 'hand shake', while there are combinations of actions that should not be performed simultaneously such as 'walk' and 'sit'.

Let's say you run a "walk and run" situation using element actions. Since the two actions, 'walk' and 'run' are not a combination of simultaneous executions, first of all, the action of 'walk' must be completed and stopped. In reality, however, it does not stop in the middle of walking and running. Therefore, in order to perform the element operation more realistically, it is necessary to generate an intermediate operation that smoothly connects the connected element operations.

If you want to move the character to the place you want, there is a way to control element movements such as 'walk' and 'turn direction'. There is a high level of control. This method should basically provide a way to solve changes in the environment that are in the direction of progress, that is, sudden appearance of obstacles. (Motion Planning)

On the other hand, if the motion of walking the plane and the motion of the terrain other than the plane exist separately from each other, it becomes difficult to provide all the element motions required for the character. Therefore, the environmental element associated with this operation is considered for one operation, so that even if the environment changes, if the operation is generated by adapting to it, the number of required element operations is reduced as well as the creation of a more natural operation. (Motion Adaptation)

In this way, an operation that can perform more general work by combining or modifying element operations in various ways is defined as a compound operation.

<Application operation>

In reality, even if humans perform the same movements, we can observe that they are performing slightly different movements. As such, there are factors that affect behavior and give slightly different results.

The first cause is in personal features. This is a very comprehensive concept. First of all, the physical characteristics include unnatural legs. Next, there are mental characteristics, such as violent or docile cases.

The second cause is the current emotional state. If human emotions are pleasant, they will walk vigorously when walking, and if they are sad, they will walk without power.

In order to create more realistic and human motions, we define motions that reflect personal characteristics or emotions of characters as application motions.

A preferred embodiment for implementing each hierarchical structure of the operation pipeline defined as described above will be described in detail as follows.

<Application Action Layer>

In the application operation processing layer 20, there is an operation interpreter module 3 to analyze the script input from the user 1 to determine detailed directions in which subsequent operations will be performed. This refers to the application database (4), which stores information on the personal characteristics or emotional state of the current character, and generates detailed instructions for future reference.

The motion planner module 6, which will be described later, is an instruction given by the motion interpreter module 3 with respect to a problem of selecting a rule of the motion plan to complete a given motion or a solution for solving a problem occurring in the middle. Are referenced.

In addition, the instructions given by the motion interpreter module 3 are referred to in the motion adaptor module 7 in terms of a problem in which an emotional state or the like is considered in the deformation method added when the resultant motion data 12 are generated.

<Complex Motion Processing>

The composite motion processing layer 30 requires three modules: the motion configurator module 5, the motion planner module 6, and the motion adaptor module 7.

The operation configurator module 5 analyzes the operation instruction content transmitted from the application operation processing layer 20 to separate and adjust the parallel operation and the sequential operation. The composite motion database 8, in which the rule information related to the motion configuration and the motion plan used by the motion configurator module 5, is stored, includes rules for parallel execution between each registered motion command and execution order. For example, "shake" and "walk" can be done in parallel, but "walk" and "run" must be performed sequentially. In addition, in the case of a "stand up" operation, an action such as "sit" or "lie down" must be performed in advance.

The motion planner module 6 analyzes the motion instruction content from the application motion processing layer 20 and configures detailed operations so that the instruction can be completed successfully. For example, the "go out door" action requires that the three actions "walk through the door", "open the door" and "walk through the door" be performed sequentially. Even a simple motion of "walk to that position" shows that two movements, "to swing to the position" and "to walk", need to be planned. The motion planner module 6 needs a database of rules for a detailed sequence of action commands to perform this for one action command. Therefore, the motion planner module 6 uses the compound motion database 8, which contains rules information on the sequence of detailed action commands, to plan the action for the compound motion command and to adapt the motion adaptor module 7 to it. Send to the element operation processing layer through.

The motion planner module 6 has two distinctive features, the first being that various schemes are possible for carrying out one operation command. For example, the "walk to the lower position" action command can walk after turning on the body as before, but if the angle of the body is not big, start walking first and perform the body movement in parallel while walking. can do. This means that the rules for action plans can include logical aspects, and that personal preferences of application actions can be applied here.

The second feature is that the sequences of the first generated operation instructions can be modified continuously until the original operation instructions are completed. For example, if an obstacle appears in the middle of walking while performing the "walk to low position" operation, a new operation plan to avoid it must be inserted in the middle. The action planner module (6) does not end with generating the first action command sequence and passing the result to the next module. Instead, the action planner module (6) continuously investigates the completion and performs a method for resolving it when affected by the environment. Should be.

The motion adaptor module 7 performs a task of specifying a method of modifying element motion to adapt to the surrounding environment. For example, in the case of a "walk" motion, it is necessary to use motion data that walks a plane, but to transform it into a form adapted to it according to a given surrounding environment to generate the result motion data 12. Therefore, the motion planner module 6 and the motion adaptor module 7 generate an operation adapted to the surrounding environment by using the surrounding environment database 9 in which the surrounding environment information is stored.

There is a need for an operator executor module 10 that generates resultant motion data 12 from operation instructions sent from the motion adaptor module 7. As described above, the motion executor module 10 includes the motion interpreter module 3, the motion configurator module 5, the motion planner module 6 and the motion adaptor module described above to generate the result motion data 12. (7) The robot receives the input character motion command and generates the result motion data 12 using the motion data database 11 in which intermediate motion data corresponding to the motion command are stored.

<Element action processing layer>

The operation executor module 10 is required for the processing layer 40 of the element operation. The operation data processed here are divided into five levels of processing, and each level is described in more detail in FIG. Let's do it.

2 is a table showing a state classified according to the level of processing the element operation in the operation executor module according to an embodiment of the present invention.

Referring to FIG. 2, such operation data may be classified according to the following FIG. 2. There are various types of motion data that can be used for each element motion. These element motion data may be generated using the techniques listed in the aforementioned method for automatically generating motion.

First, level 1 means an operation command that is one-to-one matching with element operation. Therefore, the motion data is directly expressed as 'walk', 'sit' and the like. Level 2 refers to sequences of positional data of the ends at which inverse kinematics techniques are used. Level 3 refers to the velocity and acceleration values of each segment of the character skeleton in which kinematics techniques are used. Level 4 means sequences of position data of each segment in a keyframe. Level 5 means a sequence of position data of each segment in every frame. The motion data obtained by motion capture corresponds to level 5.

On the other hand, when the resultant motion data 12 is output from the motion executor module 10, this value is applied to the character skeleton 13 to reflect the movement.

An animation is represented by a lot of motion data at level 5, but can be simply expressed by a few words at level 1. The data expressed in level 5 can be played directly, but the data expressed in level 1 can be converted to level 5 through a number of operations and then played.

In order for these motion data to be used in other motions, compound motions, and application motions in the future, they must be able to be modified. In this regard, the motion data are analyzed as follows.

In the case of level 2, it is possible to change to level 5 using a control point for the end trajectory as a parameter.

In level 3, the transition to level 5 is possible by using the velocity, acceleration, or force values of each segment as parameters.

In the case of level 4, the transformation to level 5 is possible using the data of the keyframe as a parameter.

In the case of level 5, there is nothing that can be used as a parameter since the resulting motion data itself. Only motion blending and motion warping on motion capture data can be used to study the deformation of motion.

If the operation instructions expressed by the intransitive verb expressed in level 1 are directly changed to the operation data of level 5, the operation conversion becomes difficult and the operation expression becomes difficult. Therefore, there is a need for one-to-one motion data matching the intransitive motion data while supporting such a transformation in the middle. As the operation data, operation data between levels 2 and 4 may be used. Level 2 motion data is suitable for manipulating certain objects. Level 3 has a relatively large number of parameters and is easy to automate because it is easy to control through the process. However, there remains a problem of generating kinematic motion data, which can be easily obtained from motion capture data.

In kinematic motion data, the velocity and acceleration values mean the derivative of the slope and the slope of the curve that the resulting motion data draws. Therefore, the velocity and acceleration data can be obtained by obtaining a piecewise cubic polinomials as a result of interpolation that approximates a curve drawn by the resulting motion data. The problem of the interval can be obtained by calculating the poles and inflection points of the curve.

So far, level 3 has been described as kinematic motion data, but kinetic motion data can also belong to level 3. In this case, since each segment has information on weight in advance, parameters such as velocity and acceleration can be replaced by force. There is also a problem of obtaining dynamic motion data, which can be solved by calculating the dynamics from the result data that can be obtained through motion capture and obtaining the motion data for the force given in each segment.

Although the present invention has been described in more detail with reference to some embodiments, the present invention is not necessarily limited to these embodiments, and various modifications can be made without departing from the spirit of the present invention.

The present invention provides the user with the ability to automatically generate the movements of characters using more intuitive scripts such as scenarios, resulting in faster prototyping for computer animation and dramatically reducing the content that animators need to work on, thereby improving overall productivity. There is an effect that can contribute to higher quality by assigning the surplus time that can be obtained here and the detailed emotional expression of the character.

In addition, the present invention may be implemented as an independent software application, and may be implemented as a plug-in for a software application related to computer animation, and thus, the present invention may be applied to various fields such as games and special effects of movies.

In addition, the present invention can be configured to level 1 to level 5, in the case of level 1 is simply composed of verbs to simplify the operation, and again, level 1 is converted to level 5 through level 2 to level 4 to the actual result motion data. On the contrary, the resultant motion data can be easily converted into verb form by converting the level 5 from the level 5 back to the level 1 in the opposite process, so that the motion can be easily described.

Claims (16)

In the motion creation pipeline for digital characters, A motion interpreter module for interpreting a script characterized by a grammatical expression and a logical structure for generating a motion input from a user, and incorporating instructions in the script to reflect the individual characteristics given to the character or the current emotional state; An operation configurator module analyzing the results of the operation interpreter module to separate and adjust the parallel operation and the sequential operation; An operation planner module for analyzing operation instructions given in the result of the operation configurator module and managing operation instructions so that the operation instructions can be completed; A motion adaptor module for instructing a motion transformation method according to an adaptation method to the surrounding environment with the operation commands given in the result of the motion planner module; And And a motion executor module for converting intermediate motion data corresponding to motion commands given from the result of the motion adaptor module into result motion data and actually applying motion to the character. The method of claim 1, And the motion interpreter module interprets the character motion with reference to an application motion database in which individual characteristics given to the character or current emotional state information are stored. The method of claim 1, And a motion controller module, the motion planner module, and the motion adaptor module, configured to refer to a complex motion database storing rule information necessary for processing a motion command. The method of claim 1, And the motion planner module and the motion adaptor module adapt to the surrounding environment by referring to the surrounding environment database in which information on the surrounding environment necessary for processing the operation command is stored. The method of claim 1, And the motion executor module generates result motion data for performing motion by referring to a motion data database in which intermediate motion data is stored for generating motion motion data. The method of claim 1, The action creation pipeline In order to embody the operation instructions, the lower level is divided into element operations, the intermediate level is compound operations, and the upper level is divided into application operations. Digital character motion generation pipeline, characterized in that each processing in the application motion processing layer processing. The method of claim 6, And the element motion processing layer is divided into levels 1 to 5, and the element performer module processes the element motion of the character according to the level 1 to level 5. The method of claim 7, wherein The element operation processing layer, Level 1, which indicates an operation command that has a one-to-one correspondence with element operation; Level 2 meaning sequences of positional data of the ends at which inverse kinematics technique is used; Level 3 meaning a velocity and acceleration value of each segment constituting the character skeletal structure in which kinematic technique is used; Level 4 meaning sequences of position data of each segment in the keyframe; And A digital character motion creation pipeline comprising level 5, which represents a sequence of position data of respective segments in every frame. The method of claim 8, In the case of the level 2, the control point (Control Point) for the trajectory of the end as a parameter is transformed to the level 5, and in the case of the level 3 using the value of the velocity, acceleration or force of each segment as a parameter the level 5 In the case of the level 4, the level 4 is transformed into a level 5 using the data of a keyframe as a parameter, and the level 1 is converted into motion data between the second and fourth levels. Digital character motion generation pipeline, characterized in that the conversion to 5. The method of claim 9, In the level 3, the speed and the acceleration data are converted into the level 5 of the kinematic motion data by obtaining a differential third-order polynomial as a result of interpolation approximating a curve drawn by the resulting motion data. Digital character motion creation pipeline. The method of claim 5, The motion data obtained by motion capture is included in the level 5 digital character motion generation pipeline. The method of claim 11, And transform the motion by using motion blending or motion warping on the level 5 motion capture data. As a method of generating a motion generation pipeline for digital characters, An analysis step of interpreting an operation on the character with reference to an application operation database in which individual characteristics given to the character or current emotional state information are stored; A configuration step of configuring the character motion by referring to a complex motion database in which rule information necessary for processing the motion command is stored in the character; An adapting step of adapting the character to the surrounding environment with reference to the surrounding environment database in which information about the surrounding environment is stored; And And generating a result motion data by referring to a motion data database in which intermediate motion data are stored for generating result motion data. The method of claim 13, In the constructing step, a digital character motion generation pipeline further comprises a processing step of processing the sub-steps into elementary operations, intermediate steps of complex operations, and upper steps of application operations to specify the operation instructions. How to produce. The method of claim 7, wherein The element motion is composed of a level 1 for processing an operation command that is one-to-one matching with the element motion, a level 2 for sequences of positional data of the ends at which inverse kinematics technology is used, and a character skeleton structure at which kinematics technology is used. Level 3 means the speed and acceleration value of each segment, level 4 means the sequence of position data of each segment in key frame, and level 5 means the sequence of position data of each segment in every frame. Digital character motion generation pipeline, characterized in that. The method of claim 15, Converting at the level 1 into operational data between the second and fourth levels; Transforming at level 2 to level 5 using a control point for the trailing edge as a parameter; Deforming to the level 5 using the value of the velocity, acceleration or force of each segment at the level 3 as a parameter; Transforming to level 5 using data of a keyframe as a parameter at level 4; And And processing said level 5 data sequence.

Images