Object Operating Device for Smooth 3D Motion Blending
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing techniques for blending motion data between two different actions in virtual three-dimensional spaces struggle with velocity features and short overlapping frame periods, often ignoring velocity features and requiring complex spline interpolation, which complicates the smooth transition between actions.
Innovation Solution
An object operating device that generates motion data for seamless transitions between actions by decomposing and regenerating rotations using scalar velocities and rotations, applying matrix operations like Poisson's equation, and employing exponential and logarithmic maps to handle non-linear data, ensuring natural and smooth connections between actions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Stability of the object's composition
If motion data blending is performed by gradually varying the mixing ratio during overlapping frame periods, then the transition between actions can be made smooth, but the method becomes complex and requires sufficient overlapping frame periods
Solution Approach 1:
The patent changes the parameter space by transforming rotation data into scalar rotations through decomposition, enabling linear interpolation in the scalar domain while maintaining rotational smoothness. This avoids the complexity of direct rotation blending and eliminates the need for sufficient overlapping frame periods.
Solution Approach 2:
The patent introduces scalar rotations as an intermediary representation between the original rotation data and the final blended motion. By decomposing rotations into scalar components, performing interpolation in this intermediate space, and then regenerating rotations, the system achieves smooth transitions without complex blending algorithms.
2Duration of action of moving object
If spline interpolation is applied to position and angle data to enable blending with short or no overlapping periods, then motion data blending can be performed even with minimal overlap, but selecting curve type and optimizing parameters becomes troublesome
Solution Approach 1:
The patent transforms the interpolation problem from a complex curve selection task into a simple linear interpolation task by changing the parameter representation to scalar rotations. This eliminates the need for spline curve selection and parameter optimization while enabling blending with minimal or no overlapping periods.
3Stability of the object's composition
If motion data blending is performed using traditional methods, then the transition can be made smooth, but velocity features in the former and latter actions are ignored
Solution Approach 1:
The patent preserves velocity features by performing interpolation in the scalar rotation domain, which maintains the temporal derivatives (velocities) of the motion. The linear interpolation of scalar rotations naturally preserves velocity transitions, avoiding the information loss that occurs when velocity features are ignored in traditional blending methods.
Data Source
AI summary
Motion data to be inserted between two different object actions that include a rotating action and are to be successively performed are generated so that the two actions are smoothly connected. Respective elements d required to define a transitional motion action are calculated from the respective frames after the end point of the former motion and before the start point of the latter motion. As for the three-dimensional rotation R(t) that defines the posture in each frame, independent linear values rx(t), ry(t), and rz(t) are calculated as values approximating the three-dimensional rotation R(t) by using exponential maps. The velocity components of the respective elements d are determined, and the largest value of the absolute values of the velocity components in each of the post-end frames and the pre-start frames is extracted as an interpolating velocity v(t). The respective elements d(t) in each of blended transitional motion frames can be determined.