Light-Field HMD Optics for Vergence-Accommodation Conflict
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current head-mounted display (HMD) technologies face challenges in resolving the tradeoff between eyes focus accommodation and convergence, leading to user fatigue and non-uniform perception resolution, and lack effective holographic display capabilities.
Innovation Solution
The HMD apparatus employs a 2D display panel with non-uniform resolution, a light modulator with non-uniformly distributed dot pitch, and an eye lens to create a light field image, combined with a light screen and aperture to project and filter light rays, adjusting distances and locations to enhance 3D image generation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If 3D image information is generated using related art methods, then the processing capability is limited by existing algorithms, but the accuracy and naturalness of the generated 3D image data are insufficient
Solution Approach 1:
The patent introduces a pre-trained diffusion model as an intermediary component between the captured 2D image data and the final 3D image generation. This diffusion model acts as a mediator that transforms the input images into accurate 3D image information through learned statistical relationships, rather than using direct geometric processing methods. The intermediary model enables high-accuracy 3D reconstruction without requiring complex manual processing pipelines.
Solution Approach 2:
The patent replaces traditional mechanical or algorithmic 3D reconstruction methods with a data-driven diffusion model approach. Instead of using complex geometric computations and multi-view stereo algorithms, the system uses a pre-trained neural network model that has learned 3D structure from data, substituting the mechanical processing system with an intelligent computational system that achieves better accuracy.
2Reliability
If conventional 3D reconstruction methods are used, then the system structure is relatively simple, but the generated 3D image information lacks naturalness and accuracy
Solution Approach 1:
The patent applies preliminary action by using a pre-trained diffusion model that has already been trained on large datasets before deployment. This pre-training phase performs the complex learning work in advance, so that during actual 3D reconstruction, the system can directly apply the learned transformations without performing complex computations in real-time. The pre-trained model ensures natural and accurate 3D image generation while keeping the operational system relatively simple.
3Productivity
If AI technology is integrated into the head-mounted display, then the processing capability and image quality are improved, but the device complexity and computational requirements increase
Solution Approach 1:
The patent reduces real-time computational complexity by performing the heavy AI processing work in advance during the pre-training phase of the diffusion model. The model learns complex patterns and transformations beforehand, so that during actual operation in the head-mounted display, it can generate 3D images efficiently using the pre-learned knowledge, thereby improving productivity without proportionally increasing operational device complexity.
Data Source
Figure 1~2
Figure 3~4
Figure 5~7
AI summary
Provided is a head-mounted display (HMD) apparatus, including a light modulator and eye lens, where the light modulator is configured to perform integral imaging on an image output from a 2D display panel, to form a light field image, the eye lens are configured to magnify the light field image formed by the light modulator.