Floating Point Video Codec for HDR Precision
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding standards, such as MPEG-2 and H.264/AVC, are primarily integer-based and support only fixed point operations, which limits their ability to efficiently encode and decode floating point data, particularly for high dynamic range content, resulting in poor coding efficiency and limited support for dynamic range variations.
Innovation Solution
The development of a floating point video codec that supports both floating point input data and operations, enabling efficient compression and decoding of floating point data by converting between floating point and fixed point representations during encoding and decoding processes, while signaling precision in the bitstream to adapt to varying dynamic ranges.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If floating point operations are implemented in video coding, then coding efficiency and accuracy for high dynamic range content is improved, but device complexity increases
Solution Approach 1:
The patent changes the numerical representation parameter from fixed point to floating point, enabling precise representation of high dynamic range content while maintaining coding efficiency through standardized floating point operations
Solution Approach 2:
The patent creates a universal video coding framework that can handle both standard dynamic range and high dynamic range content using the same floating point operations, making the system multi-functional and adaptable to different content types
2Adaptability or versatility
If floating point video coding is implemented, then support for dynamic range variations is improved, but compatibility with existing integer-based standards deteriorates
Solution Approach 1:
The patent introduces dynamic precision control where the precision of floating point operations can be adjusted based on content requirements, allowing adaptation to different dynamic range scenarios while maintaining a consistent coding framework
Solution Approach 2:
The patent uses floating point representation as an intermediary that can represent both integer values (for compatibility) and high precision fractional values (for HDR), serving as a bridge between legacy and advanced content types
3Productivity
If floating point to fixed point conversion is performed, then coding efficiency is improved, but processing time increases
Solution Approach 1:
The patent performs floating point to fixed point conversion at predetermined stages in the coding process (such as after transformation and quantization), allowing optimization of when conversions occur to balance efficiency and processing time
Solution Approach 2:
The patent extracts only the necessary precision from floating point representations at each stage, converting to fixed point only when and where needed, rather than maintaining full floating point precision throughout the entire coding process
Data Source
Figure 1~2
Figure 3
Figure 4
AI summary
Floating point video coding is described. In particular, a video encoder/decoder and methods for floating point video coding are discussed. Floating point video coding can be used in the video encoder and decoder to support video content, such as HDR content. Specifically, a video codec structure such as H.264/AVC can be modified in order to support floating point video coding. Within any component of a codec system, some operations can be performed at floating point precision whereas other operations can be performed at fixed point precision.