Linear Blend Skinning Rig with SSDR for Realistic Avatar Animation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for creating avatar heads in virtual experiences require manual skinning techniques that are computationally intensive and inefficient, particularly due to the large number of matrix multiplications needed to determine joint transforms and skinning weights, which can lead to unrealistic animations.
Innovation Solution
The use of a Smooth Skinning Decomposition for Rigid Bones (SSDR) technique that automatically computes linear blend skinning rigs by determining vertex-joint pairs and optimizing joint transforms and skinning weights using fewer matrix multiplications, reducing computational complexity and improving animation realism.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If manual skinning techniques are used to create avatar heads, then animation realism can be achieved, but computational complexity and time consumption increase significantly
Solution Approach 1:
The patent segments the skinning computation by identifying and processing only the relevant vertex-joint pairs that actually influence each other, rather than computing all possible combinations. This segmentation reduces the computational scope while maintaining animation realism by focusing resources on meaningful interactions.
Solution Approach 2:
The patent extracts and removes unnecessary matrix multiplications from the traditional skinning pipeline by determining a subset of influential vertex-joint pairs. This extraction eliminates redundant computations while preserving the essential deformation relationships needed for realistic animation.
2Reliability
If traditional matrix multiplication methods are used to determine joint transforms and skinning weights, then complete vertex transformation can be achieved, but computational resources are excessively consumed
Solution Approach 1:
The patent applies local quality by assigning different computational treatments to different vertex-joint relationships. Instead of uniform matrix multiplications for all pairs, it identifies influential pairs that require accurate transformation and excludes non-influential pairs, thereby reducing overall computational resource consumption while maintaining transformation accuracy where needed.
Solution Approach 2:
The patent performs partial action by computing transformations only for the necessary subset of vertex-joint pairs rather than all possible pairs. This partial computation approach reduces energy consumption while still achieving sufficient transformation accuracy for realistic animation by focusing on the most influential relationships.
3Manufacturing precision
If comprehensive vertex-joint pair computation is performed, then animation accuracy is improved, but device complexity and processing load increase
Solution Approach 1:
The patent performs preliminary action by pre-determining the influential vertex-joint pairs before the actual animation rendering. This preliminary identification of relationships simplifies the subsequent processing by establishing a reduced computation graph, thereby reducing device complexity while maintaining animation accuracy through careful selection of influential pairs.
Data Source
AI summary
The present disclosure provides techniques for automatically computing a linear blend skinning (LBS) rig. Most mesh vertices do not change positions from neutral pose to another pose (e.g., the mesh vertices at the top and back of the head). Thus, the present Smooth Skinning Decomposition for Rigid Bones (SSDR) technique determines joint transforms and skinning weights using vertex-joint sets in which a vertex i is influenced by joint j.