Replacing Flat Surface Points with Painted Planes in 3D Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing techniques for generating 3D representations of physical environments are inadequate in capturing and sharing detailed views, often resulting in granular appearances that are not representative of flat surfaces, leading to inefficient communication and rendering.
Innovation Solution
The method involves generating a 3D representation by identifying points representing flat surfaces and replacing them with planar elements 'painted' using image data, while maintaining point-based representations for non-flat objects, thereby reducing the size of the 3D model and enhancing rendering efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a point cloud representation is used to represent all surfaces, then the representation captures detailed geometry, but the file size becomes large and rendering becomes inefficient
Solution Approach 1:
The patent segments the 3D environment representation into two distinct types: point cloud representations for non-planar surfaces and planar surface representations for flat surfaces. This segmentation allows each surface type to be represented using the most efficient method, resolving the contradiction by applying different representation strategies to different portions of the environment.
Solution Approach 2:
The patent applies local quality by using high-detail point cloud representations only where necessary (non-planar surfaces) and simplified planar representations where sufficient (flat surfaces). This localized approach to representation quality maintains geometry detail where needed while improving overall rendering efficiency.
2Loss of information
If a point cloud representation is used to represent all surfaces, then the representation is comprehensive, but the data size increases and communication becomes slower
Solution Approach 1:
The patent divides the environment data into two segments: point cloud data for non-planar surfaces and planar surface data for flat surfaces. This segmentation reduces the total data size by representing large flat areas with compact planar definitions rather than dense point clouds, thereby reducing communication time while preserving essential visual information.
Solution Approach 2:
The patent uses planar surface representations as simplified copies or approximations of flat surfaces, replacing the need to transmit actual point cloud data for these surfaces. This copying approach maintains the visual appearance of flat surfaces while dramatically reducing the data that needs to be communicated.
3Device complexity
If planar elements are used to replace points, then the representation becomes more compact, but the granular appearance of non-flat surfaces may be lost
Solution Approach 1:
The patent applies local quality by selectively applying planar representations only to surfaces that are determined to be planar, while maintaining point cloud representations for non-planar surfaces. This ensures that simplification is applied only where appropriate, preserving the granular detail of non-flat surfaces while achieving compact representation for flat areas.
Data Source
AI summary
Various implementations disclosed herein include devices, systems, and methods that generate a 3D representation of a physical environment by generating a point cloud and selectively replacing some of the points. For example, points representing flat surfaces (e.g., walls and ceilings) may be replaced with planar elements that are “painted” using image data. In contrast, points representing non-flat portions (e.g. furniture, curtains, wall hangings, etc.) are left as points in the model or altered in a different way. Selectively altering the 3D representation may provide (a) a cleaner feeling and/or (b) a more compact 3D representation for more efficient and faster communication and/or rendering.