Dynamic Max Transform Size Control in VVC Decoding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The latest Versatile Video Coding (VVC) draft fixes the maximum Transform Unit (TU) size at 64, limiting the ability to control this parameter, which affects hardware complexity and interactions with sub-block transform (SBT) and Intra sub-partitioning (ISP) schemes.
Innovation Solution
A method for controlling the maximum transform size in VVC is proposed, involving identifying high-level syntax elements, determining the maximum transform size, decoding the video sequence using this size, and transmitting the decoded video sequence. This approach allows for flexible control of the maximum transform size, addressing conflicts with SBT and ISP.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If the maximum transform size is fixed at 64 in VVC, then the standard implementation is simplified, but the hardware complexity cannot be optimized and flexibility in controlling transform size is lost
Solution Approach 1:
The patent introduces dynamic control of maximum transform size through syntax elements (log2_max_transform_size) that allow the parameter to be adjusted based on sequence, picture, slice, or tile level requirements. This enables the system to adapt the transform size dynamically rather than using a fixed value, resolving the contradiction between implementation simplicity and flexibility.
Solution Approach 2:
The patent changes the parameter of maximum transform size from a fixed constant (64) to a configurable parameter controlled by syntax elements. By introducing log2_max_transform_size and related flags, the system can adjust the maximum transform size parameter according to different coding scenarios, achieving both simplicity through standardized control mechanisms and flexibility through multiple levels of parameter adjustment.
2Productivity
If the maximum transform size is increased to improve compression efficiency, then larger blocks can be processed, but hardware complexity increases
Solution Approach 1:
The patent enables dynamic adjustment of maximum transform size allowing the system to use larger sizes (up to 64 or potentially larger) when compression efficiency benefits are needed, while using smaller sizes when hardware complexity needs to be reduced. The syntax elements provide runtime control to optimize the balance between compression efficiency and hardware requirements.
Solution Approach 2:
The patent allows different maximum transform sizes to be applied to different regions (sequences, pictures, slices, or tiles) based on local requirements. This enables the system to use larger transform sizes in regions where compression efficiency is prioritized while using smaller sizes in regions where hardware complexity is a concern, achieving local optimization.
3Device complexity
If the maximum transform size is reduced to decrease hardware complexity, then processing becomes simpler, but compression efficiency deteriorates
Solution Approach 1:
The patent provides dynamic control mechanisms that allow the system to reduce maximum transform size when hardware complexity needs to be decreased, while maintaining the capability to increase it when compression efficiency is more important. The syntax elements enable runtime adjustment to optimize the trade-off based on actual operating conditions.
4Adaptability or versatility
If the maximum transform size is made configurable, then flexibility is improved, but the interaction with SBT and ISP schemes becomes more complex
Solution Approach 1:
The patent introduces dynamic syntax elements (log2_max_transform_size, sps_sbt_max_size_64_flag, slice_sbt_max_size_64_flag) that control the interaction between maximum transform size and SBT/ISP schemes. These elements provide a standardized interface for managing the complexity of interactions while maintaining flexibility in transform size control across different coding scenarios.
Data Source
AI summary
In a method for video decoding in a video decoder, a maximum transform size associated with a video sequence is determined. The maximum transform size is indicated by a syntax element associated with the video sequence. Whether a coding unit (CU) size of a CU of the video sequence is larger than the maximum transform size is determined. When the CU size is less than the maximum transform size, a sub-block transform (SBT) mode is enabled. When the CU size is larger than the maximum transform size, the CU is split into a plurality of first transform units (TUs) and when intra sub partition (ISP) mode is enabled, one of the plurality of first TUs is split into a plurality of second TUs. The CU is reconstructed based on whether the CU size is larger than the maximum transform size.