3D Scene Rendering With ML Guide-Channel Denoising
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Path-tracing in 3-D graphics is computationally expensive due to the need for a large number of samples to reduce noise, making it impractical for devices with limited resources, and existing denoising algorithms do not efficiently improve quality without increasing computational load.
Innovation Solution
A method involving machine learning models to derive enhanced guide channels from initial guide channels and noisy images, using a denoising model that approximates the noisy image as a function of these channels, allowing for efficient denoising and rendering on devices with limited resources.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the number of samples per pixel is increased to reduce noise, then the quality of the path-traced image is improved, but the computational cost increases significantly
Solution Approach 1:
The patent pre-calculates and stores guide channels (depth, surface normals, albedo) during the rendering process before denoising. These guide channels are computed in advance and used to guide the denoising operation, avoiding the need to increase sample counts while achieving quality improvement through intelligent use of pre-computed structural information
Solution Approach 2:
The patent introduces guide channels as intermediary data structures that mediate between the noisy path-traced image and the final denoised output. These guide channels (depth map, surface normals, albedo) serve as intermediate representations that contain structural information about the scene, enabling quality enhancement without additional sampling
2Measurement precision
If denoising algorithms are applied to reduce noise, then image quality is improved, but computational resources and time are consumed
Solution Approach 1:
The patent segments the denoising process into distinct stages: guide channel computation, machine learning model inference, and parameter application. By dividing the denoising task into these separate computational stages, each can be optimized independently and executed in parallel where possible, reducing overall denoising time
Solution Approach 2:
The patent replaces traditional iterative denoising algorithms with a machine learning-based approach. Instead of using computationally intensive iterative optimization methods, the system uses a trained neural network model that can perform denoising in a single forward pass, significantly reducing computation time
3Manufacturing precision
If guide channels are used for denoising, then edge preservation is improved, but memory usage increases
Solution Approach 1:
The patent creates simplified copies of scene geometry information in the form of guide channels (depth map, surface normals, albedo). These are 2D projections that capture essential structural information without storing the full 3D scene data, enabling edge-preserving denoising with reduced memory requirements compared to storing complete geometric models
Solution Approach 2:
The guide channels are computed and stored in memory during the rendering phase before denoising begins. By preparing these structural references in advance, the denoising algorithm can access them efficiently without requiring additional memory allocations during the time-critical denoising operation
Data Source
AI summary
An image of a 3-D scene is rendered by rendering a noisy image at a first resolution; obtaining initial guide channels at the first resolution, and obtaining corresponding initial guide channels at a second resolution. When the two resolutions are the same, the initial guide channels at the first resolution and the corresponding initial guide channels at the second resolution may be provided by a single set of initial guide channels. Enhanced guide channels are derived from the initial guide channels and the noisy image, using machine learning models. For each of a plurality of local neighbourhoods, the parameters of a denoising model that approximates the noisy image as a function of the one or more enhanced guide channels (at the first resolution) are calculated, and the calculated parameters are applied to the one or more enhanced guide channels (at the second resolution), to produce a denoised image at the second resolution.