Translucent Object Rendering via Voxel Light Transport

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Conventional image rendering methods fail to accurately represent the internal structure of translucent objects, resulting in a significant difference between rendered images and actual images, leading to poor user experience.

Innovation Solution

An image rendering method and apparatus that obtain target camera and light source association information to determine voxel association information using a pre-trained object attribute determination model, allowing for accurate rendering of the internal structure of translucent objects by processing light source and camera position information.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If conventional multi-view technology is used to synthesize virtual perspectives, then the rendering process can be completed, but the rendered image does not match the actual object image, resulting in low accuracy

Engineering Contradiction:

Improverendering accuracyVSAvoidimage fidelity

Core Design Contradiction:

Measurement precisionVSReliability

Solution Approach 1:

The patent transforms the rendering approach by changing from surface-based multi-view synthesis to volume-based voxel representation. Each voxel stores density and color information, allowing light transport simulations to accurately model how light interacts with translucent materials, thereby improving both rendering accuracy and image fidelity

Inventive Principle:

Principle #35Parameter changes

Solution Approach 2:

The patent replaces the conventional geometric rendering mechanism with a physics-based light transport simulation. By using voxel density information and solving light transport equations, the system naturally simulates subsurface scattering and internal reflections, achieving accurate translucent rendering without manual shading models

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

2Manufacturing precision

If conventional rendering methods are used, then the process is simple, but the internal structure of translucent objects cannot be accurately represented

Engineering Contradiction:

Improveinternal structure representationVSAvoidrendering system complexity

Core Design Contradiction:

Manufacturing precisionVSDevice complexity

Solution Approach 1:

The patent segments the translucent object into a three-dimensional voxel grid, where each voxel represents a small volume element with specific density and color properties. This segmentation enables precise representation of internal structures by capturing variations in material properties at the voxel level

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent introduces voxel density information as an intermediary between the object's geometric representation and the final rendered image. This density field acts as a mediator that encodes internal structural information, allowing the rendering system to accurately simulate light interaction with translucent materials

Inventive Principle:

Principle #24Intermediary (Mediator)

Data Source

PatentUS20250014261A1Image rendering method and apparatus, electronic device, and storage medium

Publication Date: 2025.01.09 DOUYIN VISION CO LTD

AI summary

The disclosure provides an image rendering method and apparatus, an electronic device, and a storage medium. The image rendering method includes: obtaining target camera association information and target light source association information; determining target voxel association information of each voxel in a target object based on the target camera association information, the target light source association information, and a pre-trained target object attribute determination model; and rendering a target structure image corresponding to the target object based on the target voxel association information of each voxel.