Cloth Simulation Collision Detection Using Cell-Based Collider Acceleration
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Cloth simulation in video games is computationally intensive, particularly due to mesh collisions, leading to slower frame rates and decreased performance, and existing acceleration structures introduce visual artifacts or require excessive subdivision, complicating collision detection and resolution.
Innovation Solution
Utilize collider objects divided into cells as acceleration structures to project cloth features onto their surfaces, performing collision tests only within identified cells, reducing the number of collision tests needed and optimizing for GPU processing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If full collision detection using body mesh is performed, then collision accuracy is improved, but computational cost increases significantly
Solution Approach 1:
The patent segments the collision detection process into two stages: first using a simplified collider object (bounding volume) to perform a coarse collision test, and only if that detects a collision, proceeding to the second stage of detailed mesh collision detection. This segmentation reduces the number of expensive full mesh collision tests performed.
Solution Approach 2:
The patent introduces a collider object (such as a bounding sphere or capsule) as an intermediary between the cloth mesh and the body mesh for collision detection. This intermediary provides a computationally cheaper approximation that filters out non-colliding cases before performing expensive full mesh collision detection.
2Productivity
If spatial acceleration structures are used, then collision detection speed is improved, but visual artifacts are introduced
Solution Approach 1:
The patent dynamically adjusts the level of detail used for collision detection based on the distance between objects and the current frame's collision state. For distant objects or non-colliding cases, simplified colliders are used. For close or potentially colliding objects, the full mesh is used, thus adapting the detection accuracy to the situation.
Solution Approach 2:
The patent changes the parameter of collision detection detail by using different representations (simplified collider vs. full mesh) depending on the context. The collider object serves as a low-detail approximation that can be switched to high-detail mesh detection when needed, changing the detection parameter based on spatial and temporal conditions.
3Measurement precision
If three-dimensional subdivisions are increased, then collision detection accuracy is improved, but device complexity increases
Solution Approach 1:
The patent uses simplified collider objects that are geometric copies or approximations of the actual body mesh (such as bounding spheres, capsules, or simplified meshes). These copies retain the essential collision properties while having far fewer polygons and simpler geometry, reducing the complexity of the acceleration structure needed.
Data Source
AI summary
A body mesh to be collided with a cloth mesh are received, together with collider objects (that correspond to or approximate the body mesh) divided into cells. Polygons of the body mesh are projected onto the surface of the collider objects from a location within the collider object to identify cells of the collider object that overlap the projection of the polygons. A set of cloth features that collide with the collider object are projected onto the surface of the collider object to identify cells onto which the cloth features are projected. For each cell that includes a projection of a cloth feature, collision tests are performed between the cloth feature and the polygons whose projections also overlap the same cell. Using the collider object as an acceleration structure allows for cloth simulation to be performed while reducing collision tests for each cloth feature to a limited number of polygons.