Multiview Compression via Image Difference for VR Bandwidth Reduction

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing VR/AR graphics pipelines face challenges in further reducing memory bandwidth and power consumption despite current compression techniques.

Innovation Solution

A method and system for multiview compression/decompression in graphics pipelines, where the difference between two images is calculated and compressed, allowing for storage and retrieval with reduced bandwidth and power usage, utilizing a multiview compression module and decompression module within a DDR memory system.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Loss of energy

If separate compression is applied to each view in multiview rendering, then compression coverage is complete, but memory bandwidth and power consumption are not sufficiently reduced

Engineering Contradiction:
Improvepower consumptionVSAvoidcompression complexity
Core Design Contradiction:
Loss of energyVSDevice complexity

Solution Approach 1:

The patent merges the compression of two views by calculating the difference between views and compressing only the residual difference data rather than compressing each view separately. This combining approach reduces the total amount of data to be compressed and stored, thereby reducing memory bandwidth requirements and power consumption while maintaining the ability to reconstruct both views.

Inventive Principle:
Principle #5Merging (Combining)

Solution Approach 2:

The patent extracts only the essential difference information between views rather than storing complete view data. By calculating view differences and storing only the residual data, the system removes redundant information that exists across multiple views, achieving significant bandwidth reduction while preserving the ability to reconstruct original views through difference accumulation.

Inventive Principle:
Principle #2Taking out (Extraction)

2Loss of energy

If multiview compression is implemented, then memory bandwidth is reduced, but additional processing steps are required

Engineering Contradiction:
Improvememory bandwidthVSAvoidprocessing complexity
Core Design Contradiction:
Loss of energyVSDevice complexity

Solution Approach 1:

The patent performs preliminary difference calculation between views during the compression phase, storing the residual difference data along with reference view data. This preliminary processing establishes a compression structure that reduces memory bandwidth requirements for subsequent operations, even though it adds initial processing steps.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent introduces difference calculation as an intermediary process between view generation and compression. By computing view differences and using these differences as intermediate data for compression, the system achieves reduced memory bandwidth utilization while the intermediary processing steps manage the trade-off between compression efficiency and processing complexity.

Inventive Principle:
Principle #24Intermediary (Mediator)

Data Source

PatentUS10497090B2Systems and methods for reducing memory bandwidth via multiview compression/decompression
Publication Date: 2019.12.03 QUALCOMM INC
  • US10497090B2 patent drawing
  • US10497090B2 patent drawing
  • US10497090B2 patent drawing

AI summary

Systems, methods, and computer programs are disclosed for reducing memory bandwidth via multiview compression/decompression. One embodiment is a compression method for a multiview rendering in a graphics pipeline. The method comprises receiving a first image and a second image for a multiview rendering. A difference is calculated between the first and second images. The method compresses the first image and the difference between the first and second images. The compressed first image and the compressed difference are stored in a memory. The compressed first image and the compressed difference are decompressed. The second image is generated by comparing the first image to the difference.