Autostereoscopic 3D Display Rendering via Stochastic Jittering

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing autostereoscopic 3D display technologies face challenges in minimizing image distortions and achieving high image quality with real-time performance, often requiring significant hardware resources and resulting in high costs, while also struggling to evenly distribute discontinuities for free viewer movement and optimizing rendering for a set number of viewing positions.

Innovation Solution

A method for creating image data suitable for autostereoscopic displays involves taking a plurality of samples of a 3D scene, with the centre line of projection passing through a common point for each aperture, using stochastic jittering to distribute samples evenly, and calculating only samples that will be viewable, thereby reducing unnecessary calculations and hardware requirements.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If known rendering methods are used to optimize rendering for a set number of viewing positions, then rendering speed may be improved, but image distortions accumulate in certain observation positions and discontinuities are not evenly distributed

Engineering Contradiction:
Improverendering speedVSAvoidimage quality
Core Design Contradiction:
ProductivityVSManufacturing precision

Solution Approach 1:

The patent implements dynamic perspective rendering where the rendering system adapts to the viewer's actual position rather than pre-defining fixed viewing positions. The system continuously adjusts the perspective projection parameters based on detected viewer location, transforming the static rendering approach into a dynamic one that maintains image quality across all observation positions without accumulating distortions.

Inventive Principle:
Principle #15Dynamics

Solution Approach 2:

The patent changes the rendering parameters by using stochastic jittering to randomly distribute samples across the aperture rather than concentrating them at fixed viewing positions. This parameter change in the sampling distribution ensures that discontinuities are evenly spread throughout the viewing area, preventing accumulation of image distortions while maintaining real-time rendering performance.

Inventive Principle:
Principle #35Parameter changes

2Ease of manufacture

If known rendering methods are used with standard graphics processing units, then hardware cost may be reduced, but significant hardware resources are required and image update speed is slow

Engineering Contradiction:
Improvehardware costVSAvoidimage update speed
Core Design Contradiction:
Ease of manufactureVSProductivity

Solution Approach 1:

The patent extracts and eliminates unnecessary rendering calculations by determining which samples will actually be viewable before performing the full rendering process. By pre-calculating the visible sample set based on aperture geometry and viewer position, the system removes redundant computations that would otherwise consume significant GPU resources, thereby reducing hardware requirements while maintaining high update speeds.

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The patent applies partial action by rendering only the necessary subset of perspective samples that will be visible through the aperture, rather than rendering all possible viewing positions. This selective rendering approach uses stochastic jittering to distribute the limited samples effectively, achieving high-quality images with reduced computational load and faster update rates on standard graphics processing units.

Inventive Principle:
Principle #16Partial or excessive action

3Device complexity

If a continuous scene is represented with a discrete device, then device complexity is reduced, but image distortions occur including tearing and disappearance of thin features

Engineering Contradiction:
Improvedevice simplicityVSAvoidimage quality
Core Design Contradiction:
Device complexityVSManufacturing precision

Solution Approach 1:

The patent applies segmentation by dividing the continuous scene into multiple discrete perspective samples that are distributed across the aperture using stochastic jittering. Rather than attempting to represent the entire continuous scene directly, the system segments it into manageable discrete samples that, when combined, reconstruct the scene with high fidelity and minimize artifacts like tearing and feature disappearance.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent changes the sampling parameters by using stochastic jittering to randomly position discrete samples within the aperture rather than using a fixed grid. This parameter change in the sampling distribution helps to distribute discontinuities evenly and reduces aliasing artifacts, thereby improving image quality while maintaining the simplicity of discrete device representation.

Inventive Principle:
Principle #35Parameter changes

Data Source

PatentEP2665280B1Rendering improvement for 3D display
Publication Date: 2017.12.13 SETRED AS
  • EP2665280B1 patent drawingFigure 1~2
  • EP2665280B1 patent drawingFigure 3
  • EP2665280B1 patent drawingFigure 4

AI summary

A method of creating image data, the image data suitable for use with an autostereoscopic display which may be modelled as comprising a 2D display and a switchable aperture array, the method comprising taking a plurality of samples of a 3D scene using a display plane and an aperture plane modelling the autostereoscopic display, each sample taken for the combination of a pixel on a 2D display of the modelled autostereoscopic display and an aperture of the switchable aperture array of the modelled autostereoscopic display, wherein the line along which the sample is taken for a combination of a pixel and an aperture passes through substantially the same point of the aperture for all samples for a particular aperture and wherein the line along which the sample is taken for a combination of a pixel and an aperture passes through substantially the same point of the pixel for all samples for a particular pixel.