3D Avatar Soft-Body Interaction Using Compressed Point Clouds
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for simulating clothing and other soft body objects on moving 3D avatars in virtual environments are computationally intensive and often impossible in real-time, especially on mobile devices, due to their limited processing power, and fail to provide realistic interactions among multiple objects.
Innovation Solution
A method involving offline pre-computation and compression of physics states for soft body objects, followed by real-time decompression and simplified collision modeling on client devices, enabling real-time object-to-object interaction simulations with artist-controlled fine-tuning.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If high quality physics simulation of soft bodies is performed, then simulation accuracy is improved, but computational resource requirements increase significantly
Solution Approach 1:
The patent pre-computes physics simulation data offline before runtime, storing pre-simulated clothing and soft body interactions in advance. This allows the system to avoid performing complex physics calculations during real-time execution, thereby achieving high simulation accuracy without requiring significant computational resources during runtime on mobile devices.
2Reliability
If real-time simulation of multiple interacting soft body objects is performed, then interaction realism is improved, but processing power requirements exceed mobile device capabilities
Solution Approach 1:
The system performs offline pre-computation of interactions between multiple soft body objects (such as layered clothing items) before runtime. The pre-simulated interaction data is stored and retrieved during real-time execution, enabling realistic multi-object interactions without requiring mobile devices to perform computationally intensive calculations.
Solution Approach 2:
The patent uses point cloud representations as simplified copies of complex soft body geometries. These point cloud models preserve the essential interaction characteristics of the original objects while requiring minimal computational resources for real-time simulation and rendering on mobile devices.
3Manufacturing precision
If complex physics states are computed and stored, then simulation quality is improved, but data storage and transmission requirements increase
Solution Approach 1:
The patent represents complex soft body physics states using simplified point cloud data structures instead of full mesh representations. This copying approach preserves the essential geometric and physical characteristics needed for realistic simulation while dramatically reducing the storage space and bandwidth requirements for transmitting simulation data.
Data Source
AI summary
There are provided methods, systems and non-transitory computer-readable mediums for performing simulations of object-to-object interaction in an animation of a 3D avatar, Compressed point clouds representing first and second objects individually animated with respect to the 3D avatar are received and decompressed, the first object being located on a layer below the second object. For each time frame, the first point cloud of the first object is projected onto the animation to obtain a projected point cloud. The projected point cloud is regrown towards the first point cloud and collisions with the second point cloud are detected to obtain a regrown point cloud. The regrown point cloud is solved using a set of constraints associated with the first object to obtain a final point cloud of the first object representing a frame of the first object interacting with the second object. The frame of the animation can then be rendered.