Dynamic HDR Metadata Optimization via Iterative Appearance Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for generating metadata for video sequences, particularly for high-dynamic-range (HDR) content, are inadequate in ensuring optimal display rendering across different target displays with varying characteristics, leading to visibility differences and suboptimal image quality.
Innovation Solution
A processor-driven method that applies a display mapping process, compares input and mapped images using an appearance matching metric, and iteratively optimizes metadata to reduce visibility differences, generating updated metadata to ensure accurate rendering on target displays with distinct characteristics.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If current metadata generation methods are used for HDR content, then the process is simple and fast, but the image quality and rendering accuracy deteriorate across different target displays
Solution Approach 1:
The patent applies preliminary action by pre-calculating display mapping characteristics and storing them in a lookup table before actual metadata generation. The system pre-processes the complex mapping between reference display and target display parameters, so that during runtime, metadata can be generated quickly by referencing pre-computed values rather than performing complex calculations in real-time.
Solution Approach 2:
The patent introduces an intermediary lookup table that mediates between the reference display characteristics and target display characteristics. This lookup table serves as an intermediate data structure that stores pre-computed mapping relationships, allowing the system to translate metadata between different display spaces without direct complex calculations, thus improving both accuracy and efficiency.
2Manufacturing precision
If metadata is optimized for one target display, then rendering accuracy improves for that display, but adaptability to other displays with varying characteristics deteriorates
Solution Approach 1:
The patent applies universality by creating a display mapping process that works across multiple target displays with different characteristics. The lookup table stores mapping relationships that can accommodate various display types (HDR, SDR, different brightness levels, different color spaces), making the metadata generation process universally applicable to diverse display devices while maintaining optimization for each specific display type.
Solution Approach 2:
The patent utilizes parameter changes by dynamically adjusting metadata parameters based on the target display characteristics. The system modifies luminance ranges, color space parameters, and tone mapping curves according to the specific target display's capabilities, allowing the same base metadata to be adapted for different displays while maintaining optimal rendering accuracy for each.
3Manufacturing precision
If complex display mapping processes are applied to ensure accuracy across different displays, then rendering precision improves, but processing time and computational complexity increase
Solution Approach 1:
The patent applies preliminary action by pre-calculating and storing display mapping relationships in a lookup table before runtime. Complex mapping calculations between reference display and various target display characteristics are performed in advance, allowing the system to retrieve pre-computed mapping data during actual metadata generation without performing time-consuming calculations in real-time.
Solution Approach 2:
The patent uses copying by creating a lookup table that contains copied and stored mapping relationships from complex display characterization processes. Instead of重新 performing complex measurements and calculations for each metadata generation task, the system copies previously established mapping data into the lookup table for rapid retrieval and application, significantly reducing processing time while maintaining precision.
Data Source
AI summary
Methods and systems for generating dynamic picture metadata are presented. Given an input picture generated on a mastering display, characteristics of a target display which is different than the mastering display, and an initial set of dynamic metadata for the input picture, an iterative algorithm: maps the input image to a mapped image for the target display according to a display management process and the image metadata, compares the input image to the mapped image according to a visual appearance-matching metric, and updates the image metadata using an optimization technique until a visibility difference value between the input image and the mapped image according to the visual appearance-matching metric is below a threshold.