Abstract
Converting standard dynamic range (SDR) to high dynamic range (HDR) content relies on computationally intensive ML models and heavy graphics processing unit (GPU) image processing. These requirements present barriers to implementation on mobile devices. This disclosure describes techniques, referred to as enhanced SDR (eSDR), that achieve the perceptual benefits of HDR, such as highlight sparkle and deep shadow detail, without leaving the SDR container. Global statistical features of an SDR video frame are extracted. A base luminance gain curve for the video frame is determined based on the luminance histogram. An artifact risk metric for the video frame is determined based on the one or more global statistical features. The base luminance gain curve is modulated based on the artifact risk metric to generate a modulated one-dimensional (1D) look-up table (LUT). The modulated 1D LUT is written to the hardware DPU. The video frame is output and a non-video user interface (UI) layer is output to the display panel. The output includes elevating a brightness of the display panel by a first factor; applying the modulated 1D LUT to the video frame via the hardware DPU while maintaining the video frame within the SDR color space; and applying a dimming factor to the at least one non-video UI layer, wherein the dimming factor is inversely related to the first factor.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
Huang, Sianyi; Wang, Yu; and Lee, Jongseo, "Perceptual HDR Enhancement within SDR Color Space via Dual-Model Content Analysis", Technical Disclosure Commons, ()
https://www.tdcommons.org/dpubs_series/10844