Image Encoder Quality Adjustment via 3D Scene Information
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Centralized GPU systems face inefficiencies in rendering, encoding, and transmitting graphics content due to resource constraints, particularly in networks with insufficient capacity, where existing solutions either throttle frame rates or fail to distinguish content importance, leading to suboptimal throughput and increased CPU and GPU resource utilization.
Innovation Solution
A method and apparatus that determine regions for quality adjustment in a raster sequence of frames based on 3D scene information, allowing the image encoder to adjust quality according to resource targets, with the GPU providing 3D scene information to the encoder via shared memory, enabling differentiated rendering and encoding based on content characteristics.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If the GPU renders at high frame rates with full quality, then rendering quality is improved, but network bandwidth capacity is exceeded and throughput decreases
Solution Approach 1:
The patent applies local quality by differentiating encoding quality across different spatial regions within the same frame. Regions containing important content (cursor, text, icons) are encoded at high quality, while other regions use lower quality encoding. This resolves the contradiction by maintaining high rendering quality where needed while reducing overall bandwidth consumption to improve throughput.
Solution Approach 2:
The patent dynamically changes encoding parameters (quality level, bitrate) based on content analysis and network conditions. The encoder adjusts quality parameters selectively for different regions and frames, allowing the system to maintain high quality when necessary while reducing parameters to improve throughput when bandwidth is constrained.
2Speed
If the frame rate is regulated by throttling the v-sync signal, then frame rate control is achieved, but content importance is not distinguished and resource efficiency remains suboptimal
Solution Approach 1:
Instead of uniform frame rate throttling, the patent applies different quality and update rates to different regions based on content importance. High-priority regions (cursor, text) maintain higher frame rates and quality, while low-priority regions are updated less frequently, improving overall resource efficiency while maintaining control.
Solution Approach 2:
The system performs self-service by automatically analyzing content importance and adjusting encoding parameters without external intervention. The encoder identifies important regions and autonomously applies appropriate quality levels, eliminating the need for manual frame rate throttling and improving resource efficiency.
3Productivity
If CAD application software reduces the level of detail of 3D scene under motion, then rendering rate is sustained, but shaded surface artifacts are created that are inefficient to encode
Solution Approach 1:
The patent applies preliminary action by performing quality adjustment during the encoding stage rather than requiring pre-rendering LOD adjustments. The encoder receives full-quality rendered frames and selectively applies compression and quality reduction during encoding, eliminating the need for preliminary LOD adjustments that create artifacts.
Solution Approach 2:
The system creates multiple versions of the same content at different quality levels through the encoding process. Instead of rendering different LOD versions, the encoder creates compressed representations of the full-quality scene, maintaining rendering rate while improving encoding efficiency through intelligent compression rather than artifact-prone LOD reduction.
4Extent of automation
If centralized GPU rendering is implemented, then graphics processing is consolidated, but network bandwidth is insufficient to transmit full-resolution frame sequences at original frame rate
Solution Approach 1:
The patent enables centralized GPU processing while overcoming bandwidth limitations by applying local quality differentiation. Important regions (cursor, text, icons) are transmitted at high quality, while other regions use lower quality encoding, allowing centralized rendering to be maintained without exceeding network bandwidth capacity.
Solution Approach 2:
The system dynamically changes transmission parameters (quality, bitrate, resolution) based on network bandwidth availability and content importance. This allows centralized GPU processing to be maintained while adapting transmission capacity to match network constraints, improving overall productivity.
Data Source
AI summary
The present invention describes exemplary embodiments of a method and apparatus for transmitting raster graphics. The method comprises determining a region, from 3D scene information for quality adjustment in a raster sequence of frames, wherein the raster sequence is generated by a graphics processing unit (the GPU) and consumed by an image encoder and the 3D scene information is provided by the GPU to the image encoder via shared memory and adjusting, by the image encoder, quality for the region according to the 3D scene information and a resource target for encodings of the raster sequence of frames.