Real-time Character Motion Alignment via Behavioral Gridmaps
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing rendering applications face challenges in generating realistic animation sequences for non-player characters (NPCs) by transitioning from locomotion to performed motions, often resulting in motion artifacts such as sliding or jumping, due to the limitations of motion graphs in aligning and orienting characters properly.
Innovation Solution
A method is introduced to generate a behavioral gridmap that aligns a locomotion sequence with a performed motion sequence by creating a behavioral motion space from locomotion and performed motion spaces, using registration curves and parametric motion datapoints to define motion plans and rotations, allowing for real-time alignment and orientation of characters without storing multiple behavioral gridmaps.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If motion graphs are used to transition from locomotion to performed motion, then character motion can be generated, but motion artifacts such as sliding, jumping, or skipping appear that look unnatural
Solution Approach 1:
The system pre-computes alignment data (behavioral gridmaps) that store optimal transition parameters between locomotion and performed motions. By preparing alignment information in advance rather than calculating it in real-time during motion playback, the system eliminates motion artifacts while maintaining natural transitions.
Solution Approach 2:
The invention replaces the traditional motion graph blending mechanism with a registration curve-based alignment system. Instead of linearly blending motion clips in a motion graph, the system uses pre-computed registration curves that define precise positional and rotational alignments, substituting the mechanical blending process with a mathematically rigorous alignment approach.
2Measurement precision
If multiple behavioral gridmaps are stored to support multiple rotations of performed motion alignment, then real-time alignment accuracy improves, but storage requirements increase substantially
Solution Approach 1:
The system generates a single behavioral gridmap that can be reused for multiple performed motions and rotations by applying transformation operations. Instead of storing separate gridmaps for each rotation, the same gridmap structure serves multiple purposes through mathematical transformations, making the storage system universal and multi-functional.
Solution Approach 2:
The invention changes the approach from storing multiple complete gridmaps to storing a single gridmap with transformable parameters. By modifying parameters (rotation angles, transformation matrices) rather than duplicating entire gridmap structures, the system achieves multiple alignment scenarios with minimal storage overhead.
Data Source
AI summary
A method for real-time, goal-directed performed motion alignment for computer animated characters. A sequence of periodic locomotion may be seamlessly aligned with an arbitrarily placed and rotated non-periodic performed motion. A rendering application generates a sampling of transition locations for transition from a locomotion motion space to a performed motion space. The sampling is parameterized by control parameters of the locomotion motion space. Based on the location and rotation of a goal location at which the performed motion is executed, a particular transition location may be selected to define a motion plan to which a performed motion sequence may then appended. Advantageously, by utilizing a look-up of pre-computed values for the control parameters of the motion plan, the rendering application may minimize the computational cost of finding the motion plan to move the character to a location to transition to a performed motion.