Deferred Decoupled Shading for Reduced Power Consumption
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current graphics rendering technologies face challenges in reducing processing and power demands, particularly in portable devices, due to the high resource requirements for shading graphics primitives, which are exacerbated by increasing pixel resolutions and the need for realistic CGI with effects like motion blur and depth-of-field.
Innovation Solution
The technique decouples shading from rasterization and applies effects in screen space, employing fine and coarse rasterization to identify visible geometry and primitives, and performs shading in a separate shading space with reduced pixel resolution, avoiding the use of complex data structures like G-buffers, and applying motion blur and depth-of-field effects during rasterization or post-processing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional rasterization and shading are performed at full screen resolution, then image quality is maintained, but processing demands and power consumption increase exponentially
Solution Approach 1:
The patent segments the rendering process into two distinct phases: a coarse pass that identifies visible geometry at reduced resolution, and a fine pass that applies shading only to visible surfaces at full resolution. This segmentation allows the system to avoid shading invisible geometry, thereby reducing power consumption while maintaining image quality for visible elements.
Solution Approach 2:
The patent applies partial action by performing shading operations only on a subset of primitives that are determined to be visible. Instead of shading all primitives at full resolution, the system performs coarse rasterization to identify visible geometry, then applies fine rasterization and shading only to those visible portions, reducing overall processing demands and power consumption.
2Reliability
If shading is performed on all graphics primitives, then complete coverage is ensured, but processing time and computational resources are wasted on occluded surfaces
Solution Approach 1:
The patent performs preliminary coarse rasterization before fine shading to pre-identify which primitives are visible. This preliminary action creates a set of candidate visible primitives that are then shaded in the fine pass, ensuring that no visible geometry is missed while avoiding waste of processing time on occluded surfaces.
Solution Approach 2:
The system uses feedback from the coarse rasterization pass to guide the fine rasterization pass. The coarse pass provides information about which primitives are potentially visible, and this feedback is used to limit the scope of the expensive fine shading operation only to those candidates, ensuring complete coverage of visible surfaces while minimizing processing time.
3Measurement precision
If high pixel resolution displays are used, then image detail is improved, but the amount of shading required increases exponentially
Solution Approach 1:
The patent applies local quality by using different resolution levels for different parts of the rendering process. Coarse rasterization is performed at reduced resolution to identify visible geometry, while fine rasterization is applied only to visible surfaces at full screen resolution. This local differentiation maintains image detail where needed while reducing overall shading throughput requirements.
Solution Approach 2:
The patent introduces a temporal dimension by separating rendering into multiple passes (coarse pass followed by fine pass). Instead of performing all shading operations in a single pass at full resolution, the system uses the first pass to gather visibility information and the second pass to apply shading, effectively trading spatial resolution for processing efficiency.
4Adaptability or versatility
If complex data structures like G-buffers are used to support deferred shading, then shading flexibility is improved, but device complexity and power requirements increase
Solution Approach 1:
The patent extracts and removes the complex G-buffer data structure from the rendering pipeline, replacing it with a simpler two-pass approach. By taking out the need for complex intermediate storage structures, the system achieves adequate shading flexibility through the coarse-fine pass methodology without requiring the additional hardware complexity and power resources that G-buffers would demand on portable devices.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Various embodiments are generally directed to techniques for reducing the processing demands of shading primitives in rendering a 2D screen image from a 3D model. A device includes a fine rasterizing component to employ multiple screen image samples per screen image pixel to perform fine rasterization to identify a set of visible primitives, a coarse rasterizing component to employ a single shading image sample per shading image pixel to perform coarse rasterization from the same perspective as the fine rasterization to identify at least one primitive along a path of a shading image pixel, and a shading component to shade a primitive identified by a screen image sample of a first screen image pixel as visible within the shading image pixel and identified as along the path of the shading image sample associated with the shading image pixel to derive a color value. Other embodiments are described and claimed.