Video Coding Tool Conditional Disabling for Pipeline Complexity
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video encoding and decoding systems face challenges in efficiently managing interactions between various coding tools, leading to increased pipeline complexity and dependency, which can degrade performance.
Innovation Solution
The proposed method involves disabling specific coding tools, such as Local Illumination Compensation (LIC) or Generalized Bi-prediction (GBI), for certain conditions, such as small block sizes or non-default weights, to reduce pipeline dependency and complexity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If multiple coding tools (LIC, GBI, BIO, DMVR) are enabled simultaneously, then coding efficiency and prediction accuracy are improved, but pipeline complexity and inter-tool dependency increase
Solution Approach 1:
The patent implements dynamic conditional enabling/disabling of coding tools based on block characteristics. The decoder adaptively selects which tools to apply (LIC, GBI, BIO, DMVR) depending on block size, prediction mode, and other parameters, making the pipeline flexible rather than static. This resolves the contradiction by allowing high efficiency when needed while reducing complexity when tools would provide minimal benefit.
Solution Approach 2:
Different coding tools are applied selectively to different blocks based on local characteristics. For example, LIC is disabled for small blocks where it provides little benefit, while GBI is selectively applied based on block size and prediction mode. This local differentiation allows the system to maintain high coding efficiency for suitable blocks while avoiding the complexity overhead for blocks where tools would be ineffective.
2Reliability
If coding tools are applied without conditions, then comprehensive prediction coverage is achieved, but processing overhead and computational load increase
Solution Approach 1:
The patent applies coding tools partially rather than universally. Instead of applying all available tools (LIC, GBI, BIO, DMVR) to every block, the system applies only the subset of tools that are beneficial for each specific block based on its characteristics. This partial application maintains prediction accuracy where needed while significantly reducing the overall computational load across the entire video stream.
Solution Approach 2:
The patent changes the operational parameters of coding tools based on block characteristics. For example, GBI is enabled only for specific block sizes and prediction modes, while LIC is disabled for small blocks. These parameter-based conditional applications ensure that tools operate effectively when their parameters are favorable while avoiding unnecessary computational expenditure when parameters indicate limited benefit.
3Adaptability or versatility
If coding tools are cascaded in sequence, then multiple prediction hypotheses are evaluated, but inter-tool dependency and pipeline complexity increase
Solution Approach 1:
The patent segments the coding tool application into independent, conditionally-executable modules. Each tool (LIC, GBI, BIO, DMVR) is treated as a separate segment that can be independently enabled or disabled based on block characteristics. This segmentation breaks the rigid cascaded dependency chain, allowing tools to be selectively applied without forcing all tools to process every block, thereby reducing inter-tool dependency while maintaining prediction flexibility.
Data Source
AI summary
Decoding and encoding methods and apparatuses are described for at least one first block level parameter for a current block. Bi-directional Optical flow mode allowing refining a motion of samples of a block based on an optical flow for the current block is disabled responsive to the at least one first block level parameter indicating generalized bi-prediction weights are not equal to default weights. The current block is then decoded or encoded accordingly.