Sub-Mesh Quantization Bitdepth for Boundary Distortion Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In complex scenes, the largest object, often a background feature, can tolerate a higher quantization step size, while main objects suffer significant quantization errors due to differing quantization bitdepths across connected meshes, leading to distortion at boundaries.
Innovation Solution
Adaptive bitdepth quantization is applied to each sub-mesh based on its area, with a maximum bitdepth determined for all sub-meshes, and additional vertices are added to separate connected sub-meshes to minimize distortion.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If different quantization bitdepths are applied to different sub-meshes, then coding efficiency is improved for individual objects, but quantization errors increase at the boundaries of connected meshes
Solution Approach 1:
The patent changes the quantization bitdepth parameter based on the area of each sub-mesh. Larger sub-meshes (typically background) use lower bitdepths to save bits, while smaller sub-meshes (typically main objects) use higher bitdepths to maintain precision. This area-based parameter adaptation resolves the contradiction by optimizing the trade-off between coding efficiency and quantization accuracy for different regions.
Solution Approach 2:
The patent applies different quantization qualities to different local regions (sub-meshes) based on their area characteristics. Each sub-mesh receives a customized bitdepth assignment, allowing high-quality encoding for important small objects while using lower quality for large background regions, thus achieving both coding efficiency and localized precision.
2Stability of the object's composition
If a uniform quantization bitdepth is applied to all sub-meshes, then quantization consistency is maintained across mesh boundaries, but coding efficiency decreases for complex scenes with varying object scales
Solution Approach 1:
Instead of using a fixed uniform bitdepth, the patent dynamically changes the quantization parameter based on sub-mesh area. This adaptive approach maintains sufficient consistency at boundaries while optimizing the overall bit allocation to match the scene complexity, thereby improving coding efficiency without sacrificing too much boundary consistency.
Solution Approach 2:
The patent introduces dynamics into the quantization process by making the bitdepth assignment variable rather than static. The quantization parameters are dynamically adjusted according to the area of each sub-mesh, allowing the system to adapt to different scene compositions and object scales, thus improving coding efficiency while maintaining acceptable consistency.
3Manufacturing precision
If smaller objects are encoded with higher bitdepth, then quantization error for main objects is reduced, but overall bit consumption increases
Solution Approach 1:
The patent changes the bitdepth parameter adaptively based on sub-mesh area, assigning higher bitdepths only to smaller sub-meshes that require more precision, while using lower bitdepths for larger sub-meshes. This area-driven parameter adaptation ensures that increased bit consumption is concentrated only where necessary for small objects, rather than uniformly across the entire mesh.
Solution Approach 2:
The patent applies high quantization quality locally only to small sub-meshes (main objects) that require it, while using lower quality for large sub-meshes (background). This localized quality assignment reduces overall bit consumption compared to uniform high-bitdepth encoding, while still maintaining high precision for the important small objects.
Data Source
AI summary
A method comprises receiving an input mesh comprising a plurality of sub-meshes; determining, for a respective sub-mesh in the plurality of sub-meshes, a sub-mesh bounding box that encloses the respective sub-mesh; determining, for the respective sub-mesh, a coordinate in the respective sub-mesh, wherein the coordinate is a minimum coordinate that has a smallest value in the respective sub-mesh; determining, for the respective sub-mesh, a first bitdepth based on the coordinate and the sub-mesh bounding box, wherein the first bitdepth is an adaptive bitdepth that varies according to an area of the respective sub-mesh; determining a second bitdepth that is a maximum of the first bitdepth for the respective sub-mesh and a first bit depth determined for each of the remaining sub-meshes from the plurality of sub-meshes; quantizing the respective sub-mesh based on the second bitdepth to generate a respective quantized sub-mesh; and encoding the respective quantized sub-mesh.