Synthetic Imaging via Multi-Camera Viewpoint Synthesis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Users of video content often desire views from different perspectives than what is currently available, leading to unsatisfactory experiences when the desired viewpoint is not provided by the available camera array.
Innovation Solution
A system that generates synthetic imaging data to create images from synthetic viewpoints by combining overlapping imaging data from multiple digital cameras, allowing for unique perspectives without the need for physical camera movement or additional cameras, and can handle obstructions and gestures to adjust the view.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If multiple digital cameras are deployed to provide different viewpoints, then the variety of available perspectives is improved, but the device complexity and cost increase
Solution Approach 1:
The patent creates synthetic copies of camera viewpoints through computational processing. Instead of deploying physical cameras at every desired viewpoint, the system generates synthetic images from existing camera data, effectively copying the visual information from multiple angles through software rather than hardware multiplication.
Solution Approach 2:
The patent replaces the mechanical system of physical camera deployment and positioning with a computational system. Rather than mechanically adding more cameras to achieve different viewpoints, the system uses image processing algorithms to synthesize virtual camera positions, substituting mechanical complexity with computational processing.
2Adaptability or versatility
If physical cameras are moved to change viewpoints, then the perspective variety is improved, but the response time and operational complexity increase
Solution Approach 1:
The patent replaces mechanical camera movement with computational image processing. Instead of physically moving cameras to change viewpoints, the system processes existing images through algorithms that generate new perspectives, eliminating mechanical movement time and enabling instantaneous viewpoint switching.
Solution Approach 2:
The patent captures images from multiple camera positions in advance, storing them for later processing. This preliminary capture of data from various angles allows the system to quickly generate synthetic viewpoints from pre-captured images, rather than waiting for cameras to physically move to desired positions.
3Adaptability or versatility
If the camera array is expanded to cover all desired viewpoints, then the viewpoint coverage is improved, but the cost and device complexity increase
Solution Approach 1:
The patent makes each physical camera serve multiple functions by using its captured images to generate multiple synthetic viewpoints. A single camera's image data is processed to create virtual images from various positions, allowing each physical camera to effectively represent multiple camera positions and expanding viewpoint coverage without adding physical cameras.
Solution Approach 2:
The patent merges data from multiple physical cameras and combines them through computational processing to create a unified set of synthetic viewpoints. By merging the information from limited physical cameras and processing it through image synthesis algorithms, the system achieves comprehensive viewpoint coverage that would otherwise require many more individual cameras.
Data Source
AI summary
Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for synthetic imaging. In one aspect, a method includes receiving from each digital camera respective imaging data, each digital camera having a viewpoint that is different from the viewpoints of each other digital camera and having a field of view that is overlapping with at least one other digital camera; for a synthetic viewpoint that is a viewpoint that is within a geometry defined by the viewpoints of the digital cameras, selecting respective imaging data that each has a field of view that overlaps a field of view of the synthetic viewpoint and generating, from the selected respective imaging data, synthetic imaging data that depicts an image captured from a virtual camera positioned at the synthetic viewpoint.