Virtual Liquid Surface Rendering Without GPU Read-Back
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing rendering technologies experience performance fluctuations due to the read-back of GPU data for rendering liquid surfaces and interacting virtual objects, leading to instability and resource wastage.
Innovation Solution
A virtual scene-based rendering method that utilizes a GPU for vertex fluctuation data and a CPU for grid fluctuation data to separately handle the rendering of liquid surfaces and virtual objects, optimizing resource utilization and stability.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If GPU data is read back to render virtual objects interacting with liquid surfaces, then rendering completeness is improved, but rendering performance stability deteriorates
Solution Approach 1:
The patent segments the rendering workflow into two independent parallel processes: (1) GPU resolves vertex fluctuation data for liquid surface rendering, (2) CPU divides models into grids, resolves grid fluctuation data, and renders virtual objects. This segmentation eliminates the need for GPU data read-back while maintaining complete rendering functionality, thus improving both stability and efficiency
Solution Approach 2:
The CPU performs self-service by independently resolving grid fluctuation data for virtual object rendering without relying on GPU data. The system uses two separate fluctuation data sets: vertex-level data for liquid surfaces (GPU) and grid-level data for virtual objects (CPU), allowing each processor to work autonomously and eliminate performance fluctuations caused by data transfer
2Manufacturing precision
If GPU resolves all fluctuation data for both liquid surfaces and virtual objects, then rendering quality is improved, but resource consumption increases
Solution Approach 1:
The patent segments the computational workload by data granularity: GPU handles fine-grained vertex fluctuation data for liquid surfaces requiring high visual fidelity, while CPU handles coarser-grained grid fluctuation data for virtual objects. This segmentation maintains rendering quality for both elements while optimizing resource distribution according to each processor's strengths
Solution Approach 2:
The patent applies different data resolution qualities to different rendering targets: vertex-level precision for liquid surfaces (where visual quality is critical) and grid-level precision for virtual objects (where functional accuracy suffices). This local quality differentiation maintains overall rendering quality while reducing total computational burden on the GPU
Data Source
AI summary
Provided are a virtual scene-based rendering method and apparatus, an electronic device, a computer-readable storage medium, and a computer program product. The virtual scene-based rendering method includes: resolving, by a graphics processing unit (GPU), vertex fluctuation data corresponding to each vertex to be rendered in a model to be rendered to obtain a vertex fluctuation data set corresponding to the model to be rendered; dividing, by a central processing unit (CPU), the model to be rendered into a plurality of grids to be rendered, and resolving grid fluctuation data corresponding to each of the grids to be rendered to obtain a grid fluctuation data set corresponding to the plurality of grids to be rendered; rendering, by the GPU, a liquid surface to be rendered in a virtual scene-based on the vertex fluctuation data set; and rendering, by the CPU, a virtual object for interacting with the liquid surface to be rendered based on the grid fluctuation data set.