Adaptive Denoising Filter Radius for Recurrent Blur Artifacts
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing denoising techniques for ray-tracing applications, particularly those using recurrent blurs, often result in over-blurring due to the feedback of blurred outputs, leading to artifacts in rendered frames.
Innovation Solution
The proposed solution involves adapting the radius of denoising filters based on the number of successfully accumulated frames for each pixel, using a hierarchical approach to account for rejected samples, and employing anisotropic filters that adjust dimensions based on the angle of the view vector and the direction of reflected specular lobes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If recurrent blur is used to mix low and high frequency inputs by iteratively accumulating and blurring frames, then denoising effectiveness is improved, but over-blurring occurs due to feedback of blurred outputs
Solution Approach 1:
The patent applies dynamics by making the blur radius adaptive rather than fixed. The blur radius is dynamically adjusted based on the number of accumulated frames, using a formula that reduces the blur radius as more frames are accumulated. This prevents over-blurring while maintaining the benefits of temporal accumulation for denoising.
Solution Approach 2:
The patent changes the parameter of blur radius based on the accumulation state. By modifying the blur radius parameter dynamically according to the number of frames accumulated, the system optimizes the balance between denoising effectiveness and preventing over-blurring at different stages of temporal accumulation.
2Device complexity
If a fixed blur radius is used in recurrent blur, then computational simplicity is maintained, but over-blurring artifacts are introduced
Solution Approach 1:
The patent transitions from a static to a dynamic blur radius system. The blur radius is no longer fixed but varies based on the number of accumulated frames, providing adaptability without excessive computational complexity. The dynamic adjustment prevents over-blurring while maintaining efficiency.
Solution Approach 2:
The patent implements parameter changes by making the blur radius frame-dependent. The radius is calculated based on the number of accumulated frames, allowing the system to adapt to different temporal accumulation states and prevent over-blurring artifacts.
3Reliability
If temporal accumulation is used to increase effective sample count, then denoising performance is improved, but artifacts result from not mixing low and high frequency inputs
Solution Approach 1:
The patent merges temporal accumulation with spatial blurring in a unified process. By combining these operations and using a feedback mechanism where the blurred output is fed back to update the accumulation, the system effectively mixes low and high frequency inputs while maintaining denoising performance.
Solution Approach 2:
The patent introduces feedback by using the blurred output of one iteration to update the temporally accumulated data for the next iteration. This feedback mechanism ensures that low and high frequency components are properly mixed, improving render quality while maintaining denoising effectiveness.
4Productivity
If isotropic filters are applied to pixels, then computational efficiency is maintained, but reflections do not elongate along axes under glancing angles
Solution Approach 1:
The patent applies asymmetry by using anisotropic filters instead of isotropic filters. The filter characteristics are made direction-dependent, allowing reflections to elongate along specific axes based on the view vector angle. This provides more accurate reflection rendering while maintaining reasonable computational efficiency through structured filter application.
Data Source
AI summary
Recurrent blurring may be used to render frames of a virtual environment, where the radius of a filter for a pixel is based on a number of successfully accumulated frames that correspond to that pixel. To account for rejections of accumulated samples for the pixel, ray-traced samples from a lower resolution version of a ray-traced render may be used to increase the effective sample count for the pixel. Parallax may be used to control the accumulation speed along with an angle between a view vector that corresponds to the pixel. A magnitude of one or more dimensions of a filter applied to the pixel may be based on an angle of a view vector that corresponds to the pixel to cause reflections to elongate along an axis under glancing angles. The dimension(s) may be based on a direction of a reflected specular lobe associated with the pixel.