DENOISING DYNAMICALLY RAY-TRACED SCENES USING TEMPORAL AND SPATIAL VARIANCES OF HISTORICAL PIXEL VALUES

Smart overview of the Invention

The patent application describes a method and system for enhancing the quality of dynamically ray-traced scenes by reducing noise and artifacts. This is achieved through the use of historical pixel data, which is analyzed both spatially and temporally to determine an optimal pixel value for rendering. The approach involves calculating spatial and temporal components of accumulated pixel values to improve visual output in applications like video games and animations.

Methodology

The method involves determining a history buffer for each frame, which includes accumulated pixel values at specific locations. Spatial and temporal domains are analyzed separately; spatial analysis involves calculating a mean and variance of pixel values in a region, while temporal analysis focuses on changes in pixel values over time. These analyses help in defining a pixel value range, which is crucial for determining the historical reset amount to apply to the accumulated pixel value.

Implementation

The process includes updating accumulated pixel values based on the historical reset amount, which is influenced by the current pixel value and the defined pixel value range. A linear interpolation factor (LERP) is applied to blend the accumulated pixel value with the current pixel value, ensuring a smooth transition and reducing visual artifacts. This method allows for adjustments based on spatial and temporal tolerances, which are tailored to specific ray-tracing implementations.

System Components

The system comprises several components, including a temporal accumulator, a history system, and a spatial filterer. These elements work together to manage pixel data, calculate necessary parameters, and apply spatial filters to optimize the final visual output. The history system is responsible for updating the accumulated pixel values during the ray-tracing process, ensuring that the rendered content is both accurate and visually appealing.

Applications and Benefits

This technology can be integrated into various applications through an API, enabling efficient processing of pixel data in real-time rendering environments. By improving the quality of ray-traced scenes, this method addresses common challenges such as temporal lag and ghosting, ultimately enhancing user experience in digital media. The approach allows for scalable adjustments, making it adaptable to different rendering needs and technological advancements.