Stereo Rendering With Pixel Reuse for Reduced Computational Load
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for rendering stereo views in head-mounted displays are inefficient as they require rendering the scene twice, especially for objects that appear similar in both views, leading to redundant processing and resource wastage.
Innovation Solution
The method involves rendering a first eye view and reusing pixels from the first eye view for the second eye view based on disparity indicative characteristics such as depth, distance, material type, and view-dependent effects, reducing redundant rendering by identifying regions of pixel similarity or small disparity and copying pixels from the first view.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If the scene is rendered twice for stereo view, then both left eye view and right eye view can be displayed, but computational resources are wasted on redundant rendering of similar distant objects
Solution Approach 1:
The patent applies the copying principle by copying pixel data from the first eye view to the second eye view for regions with small disparity. Instead of rendering the same distant background objects twice, the system copies the rendered pixels from one view to the other view, significantly reducing computational resources while maintaining stereo visualization quality.
Solution Approach 2:
The patent applies segmentation by dividing the image into different depth regions and processing them differently. Distant objects with small disparity are handled through pixel copying, while closer objects with larger disparity are rendered separately. This segmentation of the rendering process optimizes computational efficiency by applying the most appropriate method to each region.
2Manufacturing precision
If distant objects are rendered with high complexity, then detailed visualization is achieved, but rendering time increases significantly when rendering both views
Solution Approach 1:
For distant background objects that appear similar in both views, the system copies pixel data from the first view to the second view. This eliminates the need to re-render these distant objects, significantly reducing total rendering time while preserving visualization quality for the regions that do require rendering.
Solution Approach 2:
The system applies partial rendering by only rendering objects and regions that have sufficient disparity or view-dependent effects. Distant objects with minimal disparity are excluded from the rendering process and handled through copying, applying rendering resources only where necessary to achieve the required visualization quality.
3Productivity
If pixel copying is used for regions with small disparity, then computational load is reduced, but rendering accuracy may decrease due to reprojection errors
Solution Approach 1:
The system applies local quality by differentiating the treatment of different depth regions. Distant objects with small disparity are processed through copying (lower accuracy requirement), while closer objects with larger disparity and view-dependent effects are rendered with full accuracy. This localized approach optimizes the balance between rendering speed and accuracy based on the specific requirements of each region.
Data Source
AI summary
Various implementations disclosed herein include devices, systems, and methods that generates a stereo view of a three-dimensional (3D) environment by rendering a first eye view and reusing pixels from the first eye view for a second eye view based on a disparity indicative characteristic (e.g., identifying regions of pixel difference/similarity in the views). For example, an example process may include rendering a first view for a first viewpoint (e.g., a left eye view) of a three-dimensional (3D) environment. The process may further include estimating differences between the first view and a second view for a second viewpoint (e.g., a right eye view) of the 3D environment. The process may further include rendering the second view for the second viewpoint based on the estimated differences.