Occluder Simplification via Planar Sections
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current occlusion culling methods are inefficient for complex 3D geometry, particularly in real-time rendering, as they fail to effectively simplify occluders without sacrificing error-boundedness or requiring specific geometric constraints, leading to performance issues and unnecessary rendering of non-visible objects.
Innovation Solution
The method involves voxelizing 3D graphics data, generating planes that intersect the voxelized data to identify slice polygons, and selecting a subset of planes to maximize the correctness of a simplified occluder, which is generated from the union of these planes, allowing for error-bounded and efficient occlusion culling.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If current occlusion culling methods (error-sensitive simplification or voxel-based simplification) are used, then occlusion culling can be performed, but the methods are unsuitable for producing simplifications of input geometry due to constraints on the input geometry (e.g., must be a mesh surface)
Solution Approach 1:
The patent changes the representation parameter from mesh surface to voxelized solid geometry. By converting the input geometry into a voxelized representation (a set of voxels forming a solid), the method can handle arbitrary input geometry types including non-mesh surfaces, while maintaining the error-boundedness guarantee through the conservative occluder property that ensures all rays intersecting the original geometry also intersect the occluder.
Solution Approach 2:
The patent segments the voxelized geometry into multiple planar sections (slices) at different depths. Each plane intersects the voxelized geometry to produce a polygon, and these polygons are used to construct the occluder. This segmentation allows the complex 3D voxelized geometry to be represented by a set of 2D polygons that maintain the occlusion properties while reducing the number of triangles needed.
2Productivity
If a simplified occluder is generated with fewer triangles, then rendering performance is improved, but the occluder may fail to accurately represent the original geometry's occlusion characteristics
Solution Approach 1:
The patent transitions from representing the occluder as a single complex 3D surface to representing it as multiple 2D planar sections at different depths. Each plane is positioned at a specific depth value, and the intersection of the plane with the voxelized geometry produces a 2D polygon. This dimensional approach allows the occluder to maintain geometric accuracy by capturing the geometry at multiple depth levels while using simpler 2D polygons that are easier to render.
Solution Approach 2:
The patent performs preliminary voxelization of the input geometry before generating the occluder. By first converting the input geometry into a voxelized representation, the method pre-processes the geometry into a form that is easier to manipulate and section. This preliminary action enables subsequent planar sectioning to produce accurate occluders with fewer triangles, as the voxelized geometry provides a discrete, manageable representation that can be efficiently processed.
3Productivity
If the occluder includes relatively few triangles, then the benefit to the rendering process is maximized, but some rays that do not intersect the occluder may still intersect the original geometry
Solution Approach 1:
Instead of trying to create a minimal occluder that perfectly matches the original geometry's surface, the patent inverts the approach by using a conservative occluder definition. The occluder is designed to be slightly larger than the original geometry in terms of ray intersection, ensuring that all rays that intersect the original geometry also intersect the occluder. This inversion of the traditional simplification approach allows for fewer triangles while maintaining correctness, as the occluder's purpose is to be a safe upper bound rather than a precise match.
Data Source
AI summary
Techniques are disclosed for simplifying an occluder representing 3D graphics data. Graphics data corresponding to one or more 3D objects is voxelized. A plurality of planes is generated to intersect the voxelized graphics data. Intersections between the planes and the voxelized graphics data are evaluated to identify corresponding slice polygons from the intersections. A subset of the planes is selected to maximize a correctness of a simplified occluder. The simplified occluder representing the object is generated from a union of the selected subset of the planes.