Object-Based Layer Stacks for Multi-View Image Rendering

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing multi-view image formats like MPI and MSI require large data storage and transmission due to uniform layer placement, leading to inefficiencies and potential ghost objects, especially in real-time broadcasting scenarios.

Innovation Solution

Generate stacks of layers for each object in a virtual scene based on their actual positions, reducing pixels with 100% transparency and using object detection or graphics engines to determine layer positions, with optional machine learning for improved accuracy.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Ease of manufacture

If uniform layer placement is used in MPI/MSI formats, then the rendering process is simple, but data storage and transmission requirements increase significantly

Engineering Contradiction:
Improverendering process simplicityVSAvoiddata storage and transmission requirements
Core Design Contradiction:
Ease of manufactureVSQuantity of substance

Solution Approach 1:

The patent applies local quality by transitioning from uniform layer placement to non-uniform layer placement where layers are concentrated around object positions in the scene. This means different regions of the virtual scene have different layer densities - areas with objects have more layers while empty spaces have fewer or no layers, optimizing the balance between rendering quality and data efficiency.

Inventive Principle:
Principle #3Local quality

Solution Approach 2:

The patent segments the virtual scene into multiple stacks of layers, where each stack corresponds to a specific object or region. This segmentation allows the system to process and transmit data for only the relevant regions rather than the entire scene uniformly, reducing overall data requirements while maintaining rendering effectiveness.

Inventive Principle:
Principle #1Segmentation

2Reliability

If layers are placed everywhere in the virtual scene, then coverage is complete, but ghost objects are introduced and data efficiency decreases

Engineering Contradiction:
Improvescene coverage completenessVSAvoidghost objects
Core Design Contradiction:
ReliabilityVSObject-generated harmful factors

Solution Approach 1:

The patent eliminates ghost objects by implementing local quality through non-uniform layer placement. Instead of placing layers uniformly throughout the entire virtual scene (which creates ghost objects in empty spaces), the system places layers locally around detected object positions, ensuring complete coverage only where objects actually exist.

Inventive Principle:
Principle #3Local quality

Solution Approach 2:

The patent performs preliminary object detection and position determination before generating the layer structure. By detecting objects and their positions in advance, the system can pre-configure layers only in the regions where objects are present, preventing ghost objects from being generated in empty spaces while maintaining complete scene coverage.

Inventive Principle:
Principle #10Preliminary action

3Quantity of substance

If sparsity is enforced on layers, then data requirements are reduced, but visual quality of objects deteriorates

Engineering Contradiction:
Improvedata requirementsVSAvoidvisual quality of objects
Core Design Contradiction:
Quantity of substanceVSManufacturing precision

Solution Approach 1:

The patent resolves this contradiction by implementing local quality through non-uniform layer distribution. The system dynamically adjusts layer density based on scene content - using sparsity (fewer layers) in empty regions to reduce data requirements while maintaining sufficient layer density around objects to preserve visual quality, thus achieving both data efficiency and rendering quality simultaneously.

Inventive Principle:
Principle #3Local quality

Data Source

PatentUS12488531B2Multi object surface image (MOSI) format
Publication Date: 2025.12.02 KONINKLIJKE PHILIPS NV
  • US12488531B2 patent drawing
  • US12488531B2 patent drawing
  • US12488531B2 patent drawing

AI summary

A method for processing multi-view image data. The method comprises obtaining source view data from a plurality of sensors, the source view data containing source texture data and source depth data of a scene with one or more objects. The positions of one or more of the objects in the scene are obtained and a stack of layers is generated in a virtual scene for at least one of the objects, wherein the position of a stack of layers in the virtual scene is based on the position of the corresponding object in the scene. Generating a stack of layers comprises generating a plurality of layers, wherein each layer comprises texture data and transparency data for the corresponding object.