Luma and Chroma Mapping for HDR-to-SDR Color Accuracy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image representation technologies struggle to efficiently transition between different dynamic range images while maintaining chromatic accuracy, particularly in transitioning from High Dynamic Range (HDR) to Standard Dynamic Range (SDR) without causing color distortions or clipping important details.
Innovation Solution
A method involving luminance mapping functions, such as FL_enc and FL_dec, is employed to convert HDR images to a lower dynamic range format, ensuring chromatic accuracy by using invertible functions and metadata to guide display adaptation for various display capabilities.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If HDR images are converted to SDR using conventional luminance mapping, then the dynamic range is reduced for SDR display, but color accuracy deteriorates due to chroma clipping and distortion
Solution Approach 1:
The patent segments the luminance mapping process into two distinct functions: a forward mapping function (FL_enc) that encodes HDR luminance to SDR luminance, and an inverse mapping function (FL_dec) that decodes SDR luminance back to HDR luminance. This segmentation allows independent optimization of each mapping direction to preserve chromatic accuracy while achieving display compatibility.
Solution Approach 2:
The patent introduces an intermediary offset value (D) in the luminance mapping equations that acts as a mediator between HDR and SDR luminance spaces. This offset compensates for luminance differences and prevents chroma clipping during conversion, thereby maintaining color accuracy across different dynamic ranges.
2Device complexity
If simple luminance scaling is used for HDR to SDR conversion, then processing complexity is reduced, but color distortion increases significantly
Solution Approach 1:
The patent changes the parameters of the luminance mapping by introducing an offset value (D) that can be adjusted based on the specific HDR and SDR luminance ranges. This parameter modification transforms the simple scaling relationship into a more accurate mapping that preserves chromatic information while adapting to different display capabilities.
3Loss of substance
If aggressive chroma compression is applied during HDR to SDR conversion, then file size is reduced, but chromatic detail is lost
Solution Approach 1:
The patent applies preliminary luminance mapping with offset compensation before chroma subsampling or compression. By pre-adjusting the luminance values using the forward mapping function (FL_enc), the chromatic information is preserved in a more favorable luminance space, allowing subsequent compression to retain more chroma detail.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
To get an even more accurate luminance mapping usable for dynamic range conversion of images, the circuit and method of pixel color processing comprises: nd luma mapping using a luma mapping function (F_Ymap) an input luma into a non-linear output luminance (L'_out), or output luma; establishing a multiplier (m) for the input luma which depends on the input luma (Y_in), and multiplying an input blue and red chroma by the multiplier to obtain normalized chroma components (Cb_m, Cr_m), and clipping the normalized chroma components to a range between e.g. -0.5 and 0.5; transforming the clipped chroma components and a luma component which is equal to 1.0 into normalized red, green and blue components (R'_n, G'_n, B'_n) by using a color conversion matrix; multiplying the non-linear output luminance (L' _out), or output luma, by the normalized red, green and blue components (R'_n, G'_n, B'_n) to obtain scaled red, green and blue components (R' _a, G'_a, B'_a); wherein the luma mapping applies to the input luma a secondary luma mapping function (F_YmapC) which is based on the luma mapping function by applying a first calculation which equals: F_YmapCY_in=F_YmapY_in+D/1+D, wherein D is an offset value which is constant for the processing of at least one image; wherein the multiplier is established as a multiplication of a first multiplier component which depends on the input luma by a second multiplier component which equals a reciprocal of a value which equals the sum of an output of the luma mapping function when having the input luma as input and the offset value (D); performing a transformation of the scaled red, green and blue components (R'_a, G' _a, B'_a) into final red, green and blue components (R'_f, G'_f, B'_f) which depends on a second calculation which equals: calculating a first result by subtracting from a respective one of the scaled red, green and blue color components a value which equals: the offset value divided by one plus the offset value; calculating a second result being a maximum of zero and the first result; and multiplying this second result by one plus the offset value.