Dynamic Ray-Traced Scene Denoising With Dual History Buffers
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image generation techniques, particularly in ray tracing, suffer from temporal lag, ghosting, and increased computational complexity due to challenges in adapting to dynamic scenes, leading to compromised denoising quality and noise levels.
Innovation Solution
Implementing a denoising system that utilizes both normal and responsive history buffers with different convergence rates, applying historical acceleration and reset techniques based on temporal and spatial variance to adapt to changing lighting conditions, enhancing responsiveness and reducing artifacts.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional temporal accumulation techniques are used in ray tracing, then denoising quality is improved, but temporal lag and ghosting artifacts occur
Solution Approach 1:
The patent implements a dynamic history buffer system that adapts its behavior based on scene conditions. The responsive history buffer dynamically adjusts its convergence rate based on detected scene changes, allowing the system to switch between aggressive denoising (when scene is static) and rapid adaptation (when scene changes), thereby reducing temporal lag while maintaining denoising quality.
Solution Approach 2:
The patent divides the history buffer into two separate components: a normal history buffer for standard temporal accumulation and a responsive history buffer for rapid adaptation to scene changes. This segmentation allows each buffer to specialize in different aspects of denoising, with the responsive buffer specifically addressing temporal lag issues by converging faster to new pixel values.
2Measurement precision
If traditional temporal accumulation techniques are used in ray tracing, then denoising quality is improved, but ghosting artifacts occur
Solution Approach 1:
The system dynamically adjusts the convergence behavior of the responsive history buffer based on scene activity detection. When scene changes are detected, the responsive buffer accelerates its convergence rate, preventing the accumulation of outdated pixel values that cause ghosting artifacts, while still maintaining denoising quality through coordinated use with the normal history buffer.
Solution Approach 2:
The patent implements a feedback mechanism that monitors scene changes and adjusts the history buffer convergence rates accordingly. This feedback loop detects when ghosting artifacts are likely to occur (during scene transitions) and modifies the denoising behavior in real-time to prevent artifact formation while preserving denoising quality in stable regions.
3Measurement precision
If traditional temporal accumulation techniques are used in ray tracing, then denoising quality is improved, but computational complexity increases
Solution Approach 1:
The patent segments the computational workload into two distinct history buffer processing paths. The normal history buffer handles standard temporal accumulation with lower computational overhead, while the responsive history buffer handles adaptive convergence only in regions where scene changes are detected, reducing overall computational complexity compared to applying aggressive denoising uniformly across the entire image.
Solution Approach 2:
The system applies different convergence rates and denoising strategies to different regions of the image based on local scene activity. The responsive history buffer applies accelerated convergence only to pixels affected by scene changes, while other regions use the normal history buffer with standard processing, thereby reducing total computational complexity while maintaining denoising quality where needed.
Data Source
AI summary
In various examples, systems and methods are disclosed relating to historical acceleration. One method includes determining a plurality of history buffers for a frame, the plurality of history buffers comprising a responsive history buffer and a normal history buffer, the responsive history buffer comprising a first pixel value at a pixel location of the frame, and the normal history buffer comprising a second pixel value at the pixel location of the frame, and the normal history buffer including a second pixel value at the pixel location of the frame. The method further includes determining at least one difference between the first pixel value of the responsive history buffer and the second pixel value of the normal history buffer and updating at least one of the first pixel value or the second pixel value based on the at least one difference and a tuning parameter.