Inverse Tone Mapping for Visual Effects HDR Quality
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing inverse tone mapping methods amplify noise, clipping, luminance compression, and lack of details in HDR images containing visual effects, particularly in legacy low dynamic range content, due to the limitations of quantization and precision in visual effects assets.
Innovation Solution
A method involving a first pass of inverse tone mapping on low dynamic range images with visual effects, followed by regeneration of visual effects using HDR lighting indications derived from both the original and tone-mapped images, to enhance the visual quality of HDR images by maintaining higher bit depth and contrast, thereby avoiding issues like clipping and noise amplification.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Illumination intensity
If inverse tone mapping is applied to LDR images containing visual effects, then HDR image is generated with extended luminance range, but noise, clipping, luminance compression and quantization artifacts are amplified
Solution Approach 1:
The patent segments the image processing into two distinct passes: first pass inverse tone mapping on the composite LDR image to generate HDR image, and second pass selective re-rendering of visual effects assets at HDR precision. This segmentation allows different processing strategies for different image components, preserving VFX quality while maintaining HDR benefits for natural content.
Solution Approach 2:
The patent performs preliminary inverse tone mapping on the composite LDR image to generate an HDR image, which then serves as the basis for determining HDR lighting indications. These lighting indications are used in the second pass to re-render visual effects assets at HDR precision, preventing quality degradation before it occurs.
2Device complexity
If visual effects are rendered at LDR precision and then inverse tone mapped, then processing complexity is reduced, but clipping, luminance compression and lack of details occur
Solution Approach 1:
The patent applies local quality by treating visual effects assets differently from natural image content. While the natural content undergoes standard inverse tone mapping, the VFX assets are selectively identified and re-rendered at HDR precision using HDR lighting indications, ensuring high quality where it matters most without unnecessarily complicating the entire processing pipeline.
Solution Approach 2:
The patent changes the precision parameter for visual effects assets from LDR to HDR in the second pass. By determining HDR lighting indications from the first pass results and using these to re-render VFX assets at HDR precision, the patent maintains processing efficiency while dramatically improving VFX quality metrics.
3Illumination intensity
If HDR image is generated from LDR image with standard iTMO, then luminance range is extended, but saturation modification and banding artifacts appear
Solution Approach 1:
The patent performs preliminary inverse tone mapping to generate HDR image and extract lighting indications, then uses these indications to re-render visual effects assets at HDR precision. This preliminary action preserves color information that would otherwise be lost in standard iTMO, preventing saturation modification and banding artifacts in the final HDR image.
Data Source
AI summary
A method and apparatus for inverse tone mapping of an LDR image with computer generated imagery components, such components being also called visual effects, are disclosed. The disclosed method includes accessing a first LDR image, the first LDR image being obtained from at least one computer generated imagery component rendered and composited into an LDR reference image or background plate; determining an HDR lighting indication responsive to information extracted from a part of the first LDR image and a respective part of a first intermediate HDR image obtained by applying a first pass inverse tone mapping to the first LDR image; obtaining a second intermediate HDR image by applying a second pass inverse tone mapping to the LDR reference image; obtaining the first HDR image from the computer generated imagery components rendered using HDR lighting indication and composited into the second intermediate HDR image.