Gamut Border Clipping for Stable HDR-to-SDR Color Conversion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing gamut clipping methods in video conversion from HDR to SDR cause artifacts like banding effects and unstable color patches due to quantization errors at color gamut boundaries, leading to inconsistent and visually unacceptable results.
Innovation Solution
A gamut clipping method that modifies negative components to positive values based on neighboring pixels within the target color gamut, using a clipping curve determined by averaging corrected components from multiple virtual pixels, and approximated by models like straight lines or polynomial functions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If basic gamut clipping is applied to convert RGB values from first color gamut to second color gamut, then the conversion process is simple and fast, but quantization effects and artifacts like banding appear at color gamut boundaries
Solution Approach 1:
The patent applies preliminary action by performing gamut clipping on neighboring pixels before converting the target pixel. Specifically, it identifies pixels at color gamut boundaries, clips their RGB components to ensure they fall within the target gamut, and uses these clipped neighboring pixels to determine appropriate clipping values for the target pixel. This preliminary processing of surrounding pixels prevents quantization artifacts while maintaining conversion efficiency.
Solution Approach 2:
The patent applies local quality by treating pixels at color gamut boundaries differently from interior pixels. It detects whether a pixel lies on or near the boundary of the color gamut and applies specialized clipping logic only to these boundary pixels. The clipping process adjusts RGB components locally based on the specific boundary conditions and neighboring pixel values, rather than applying uniform clipping to all pixels.
2Productivity
If clipping values are determined without considering neighboring pixels, then the processing is faster, but unjustified variations and unstable color patches occur
Solution Approach 1:
The patent implements feedback by using the clipped values of neighboring pixels to inform the clipping decision for the target pixel. The process clips neighboring pixels first, then uses those clipped results as reference feedback to determine appropriate clipping values for the target pixel. This feedback mechanism ensures consistency across adjacent pixels and eliminates unjustified variations, while the localized nature of the feedback keeps processing efficient.
3Device complexity
If quantized YUV pixels are converted directly without correction, then the conversion is straightforward, but banding effects and artifacts appear in the output
Solution Approach 1:
The patent converts the harmful quantization effect into a benefit by using the quantized neighboring pixels to define clipping boundaries. Instead of treating quantization errors as purely harmful, the invention leverages them to establish appropriate clipping values that prevent artifacts in the target pixel. The quantization-induced variations in neighboring pixels are transformed into useful information for determining optimal clipping thresholds, thereby eliminating banding effects.
Data Source
Figure 1A~1B
Figure 1C
Figure 2A
AI summary
A method comprising: obtaining a first YUV picture in a first color gamut; applying a conversion process to the first YUV picture to obtain a second YUV picture in a second color gamut, the first color gamut encompassing the second color gamut, the conversion process comprising for a first YUV pixel of the first YUV picture: obtaining (712) a component of a first RGB pixel in the second color gamut obtained from the first YUV pixel; and, applying (713) a gamut clipping process to the component, the gamut clipping process consisting in clipping the component to a value defined by a clipping curve responsive to the component fulfil a clipping criterion based on the clipping curve; wherein, the clipping curve is determined using a clipping curve determination process based on at least one second YUV pixel in the first color gamut with a value neighboring the value of the first YUV pixel and that provide a value of the component not fulfilling the clipping criterion when the second YUV pixel is converted in a second RGB pixel in the second color gamut.