Embedded Gain Maps for Artifact-Free HDR and SDR Conversion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional image conversion techniques between standard dynamic range (SDR) and high dynamic range (HDR) images often result in inconsistent and undesirable visual artifacts, making it difficult to efficiently transition between these image types.
Innovation Solution
Generating a gain map by comparing a first image to a second image and embedding it into the first image to enable efficient reproduction of the second image using the first image and the gain map.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If conventional image conversion techniques are used to convert HDR images to SDR images, then the conversion process is simple, but visual artifacts such as banding are introduced into the resulting image
Solution Approach 1:
The patent introduces a gain map as an intermediary element that mediates between HDR and SDR images. The gain map stores pixel-wise gain values derived from comparing HDR and SDR versions, enabling accurate conversion without direct manipulation that causes artifacts. This intermediary structure allows the system to achieve both simple conversion processes and artifact-free results.
Solution Approach 2:
The patent performs preliminary action by pre-computing and storing gain maps that capture the relationship between HDR and SDR images. These gain maps are generated in advance and can be applied during conversion processes, eliminating the need for complex real-time calculations and preventing visual artifacts from appearing during conversion.
2Adaptability or versatility
If conventional image conversion techniques are used to upgrade SDR images to HDR images, then the conversion can be performed, but guesswork is involved which introduces uncorrectable visual artifacts
Solution Approach 1:
The gain map serves as an intermediary that contains pre-calculated gain values representing the precise relationship between SDR and HDR images. Instead of using guesswork-based conversion, the system applies these stored gain values to accurately reconstruct HDR images from SDR inputs, eliminating uncertainty and preventing visual artifacts.
Solution Approach 2:
The patent uses copying by storing actual pixel correspondence information in the gain map. Rather than attempting to guess HDR values from SDR images, the system copies the precise relationship data captured during gain map generation, ensuring accurate reproduction of HDR image characteristics without introducing artifacts.
3Measurement precision
If gain maps are generated and embedded into images, then accurate reproduction of images across different dynamic ranges is enabled, but additional processing steps are required
Solution Approach 1:
The patent merges the gain map data with the image data by embedding the gain map into the image file structure. This combining approach allows the conversion information to travel with the image, enabling accurate HDR-SDR transitions without requiring separate processing systems or complex external lookup tables.
Solution Approach 2:
The image file becomes self-sufficient by containing its own gain map information embedded within it. This self-service approach allows the image to carry all necessary conversion data internally, eliminating the need for external conversion tools or complex processing infrastructure while maintaining high reproduction accuracy.
Data Source
AI summary
Disclosed are various techniques for generating gain maps. According to some embodiments, one technique for generating a gain map includes the steps of (1) accessing a high dynamic range (HDR) image, (2) accessing a standard dynamic range (SDR) image, (3) for each pixel shared between the HDR and SDR images: plotting the pixel onto a graph, wherein an HDR value of the pixel is plotted on a Y-axis of the graph and an SDR value of the pixel is plotted on an X-axis of the graph, (4) establishing a first curve that represents an approximation of the plotted pixels, (5) inverting the first curve to establish a second curve, (6) applying the second curve against the plotted pixels to establish replotted pixels, (7) generating the gain map based on the replotted pixels, and (8) embedding the gain map into the HDR image.