Depth-Based Sweeping for Noise Reduction in Stochastic Rendering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current graphics rendering technologies, such as hardware rasterization pipelines and ray tracing programs, produce noisy images due to difficulties in reproducing physical camera effects like motion and defocus blur, and stochastic techniques at interactive frame rates are insufficient to produce visually smooth images.
Innovation Solution
A method that identifies noisy pixels and sweeps over neighboring samples in a depth-based order to reduce noise, using filter weights based on blur levels to reconstruct pixel values, effectively addressing noise from transparent objects and blurry regions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If stochastic techniques are used for interactive rendering, then camera effects like motion blur and defocus blur are produced, but the images become noisy and visually unsmooth due to insufficient random samples
Solution Approach 1:
The patent segments the sampling process by depth order, dividing samples into depth-based groups. This allows selective processing where samples from transparent and blurry regions are separated and handled differently, enabling noise reduction in specific depth regions while preserving camera effects.
Solution Approach 2:
The patent performs preliminary depth-based sorting and identification of transparent/blurry regions before the final compositing step. By pre-identifying problematic samples in transparent and blurry regions, the system can apply targeted noise reduction techniques without affecting the overall camera effect rendering.
2Productivity
If simple unbiased sample averaging is used, then the rendering process is computationally efficient, but visually smooth images cannot be produced with the available number of samples
Solution Approach 1:
The patent applies different processing strategies to different spatial regions based on their depth characteristics. Samples from transparent and blurry regions are identified and processed differently from standard regions, allowing local optimization of noise reduction without globally increasing sample count, thus maintaining rendering speed while improving local image quality.
3Manufacturing precision
If more random samples are used to reduce noise, then image smoothness improves, but the number of samples required exceeds what is available at interactive frame rates
Solution Approach 1:
The patent changes the processing parameter from uniform sample averaging to depth-based selective processing. By identifying and separately processing samples from transparent and blurry regions, the system achieves noise reduction through intelligent sample selection rather than increasing total sample count, thereby maintaining interactive frame rates while improving image smoothness.
Data Source
AI summary
A system, method, and computer program product are provided for reducing noise in an image using depth-based on sweeping over image samples. In use, each noisy pixel of an image having noise is identified. Additionally, for each noisy pixel, at least one sample included in each of a plurality of neighboring pixels to the noisy pixel is identified. Furthermore, the samples are swept over at least partially in a depth-based order to identify a value for the noisy pixel that reduces the noise.