3D Grid Volumetric Sampling for Stable High-Quality Rendering

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Rendering volumetric effects in video graphics applications, such as video games and movies, is computationally expensive and often results in low quality with poor temporal coherence and flickering, which is exacerbated by the need to balance computational overhead with spatial and temporal resolution.

Innovation Solution

An image processing method that uses a 3D grid to sample volumetric effect data at a low resolution, followed by bicubic interpolation to upscale the results, improving spatial and temporal resolution with minimal computational overhead.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Manufacturing precision

If volumetric effects are rendered at high resolution, then spatial quality is improved, but computational cost increases significantly

Engineering Contradiction:
Improvespatial qualityVSAvoidcomputational cost
Core Design Contradiction:
Manufacturing precisionVSProductivity

Solution Approach 1:

The volumetric effect rendering is divided into two separate stages: a first pass at low resolution to capture temporal coherence, and a second pass at high resolution to achieve spatial quality. This segmentation allows each pass to be optimized for its specific purpose, reducing overall computational cost while maintaining both temporal and spatial quality.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The first rendering pass is performed preliminarily at low resolution to establish temporal coherence and base lighting information. This preliminary action provides a foundation that reduces the computational burden of the subsequent high-resolution pass, as some calculations have already been performed in the first pass.

Inventive Principle:
Principle #10Preliminary action

2Productivity

If volumetric effects are rendered at low resolution to reduce computational overhead, then processing speed is improved, but spatial quality deteriorates with blockiness

Engineering Contradiction:
Improveprocessing speedVSAvoidspatial quality
Core Design Contradiction:
ProductivityVSManufacturing precision

Solution Approach 1:

The rendering process is segmented into two resolution levels: a fast low-resolution pass for temporal coherence and a high-resolution pass for spatial quality. This allows the system to achieve both processing speed and spatial quality by performing different operations at different resolutions rather than choosing one or the other.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The results from the low-resolution first pass and the high-resolution second pass are merged together. The first pass provides temporal coherence and base lighting, while the second pass adds spatial detail. By combining these results, the system achieves both processing efficiency and high spatial quality that would be difficult to obtain with a single pass.

Inventive Principle:
Principle #5Merging (Combining)

3Stability of the object's composition

If multiple frames are blended to smooth temporal discontinuities, then temporal coherence is improved, but temporal resolution is sacrificed

Engineering Contradiction:
Improvetemporal coherenceVSAvoidtemporal resolution
Core Design Contradiction:
Stability of the object's compositionVSLoss of time

Solution Approach 1:

The rendering is segmented into two passes where the first pass at low resolution handles temporal coherence through blending, while the second pass at high resolution restores temporal resolution. This segmentation allows temporal smoothing without permanently losing temporal detail, as the high-resolution pass recovers fine temporal variations.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

Temporal coherence is established preliminarily in the first pass through frame blending, but the second pass then performs a preliminary restoration of temporal resolution by rendering at high resolution. This two-stage approach allows the system to benefit from temporal smoothing while recovering temporal detail that would otherwise be lost.

Inventive Principle:
Principle #10Preliminary action

4Manufacturing precision

If full resolution sampling is performed, then spatial quality is improved, but computational overhead increases

Engineering Contradiction:
Improvespatial qualityVSAvoidcomputational overhead
Core Design Contradiction:
Manufacturing precisionVSDevice complexity

Solution Approach 1:

The sampling process is segmented into two resolution levels: low-resolution sampling in the first pass to reduce computational overhead, and high-resolution sampling in the second pass to achieve spatial quality. This segmentation allows the system to perform full resolution sampling only when necessary, reducing overall computational overhead while maintaining spatial quality.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

Instead of performing full resolution sampling throughout the entire rendering process, the system applies partial action by sampling at low resolution for the majority of the work (first pass) and only applies excessive action (high-resolution sampling) where needed for the final quality pass. This reduces overall computational overhead while achieving the desired spatial quality.

Inventive Principle:
Principle #16Partial or excessive action

Data Source

PatentEP4687106A1Image processing method and system
Publication Date: 2026.02.04 SONY INTERACTIVE ENTERTAINMENT LLC
  • EP4687106A1 patent drawingFigure 1~2a
  • EP4687106A1 patent drawingFigure 2b~3
  • EP4687106A1 patent drawingFigure 4~5

AI summary

There is provided an image processing method for generating images including a volumetric effect. The method comprises: sampling, using a three-dimensional grid, computer-generated volumetric effect data for a virtual scene at a first sampling resolution, to obtain a first set of sampling results; upscaling at least part of the first set of sampling results by interpolating between samples of the at least part of the first set of sampling results, to obtain a second set of sampling results having a second, higher, sampling resolution; and generating one or more display images for the virtual scene at least partly in dependence on the second set of sampling results