Making Video Quality Assessment Models Robust to Bit Depth

We introduce a novel feature set, which we call HDRMAX features, that when included into Video Quality Assessment (VQA) algorithms designed for Standard Dynamic Range (SDR) videos, sensitizes them to distortions of High (HDR) videos are inadequately accounted by these algorithms. While features not specific HDR, and also augment the equality prediction performances VQA models on SDR content, they especially effective HDR. modify powerful priors drawn from Natural Statistics (NVS) enhancing their measurability where visually impact brightest darkest local portions thereby capturing often poorly existing models. As demonstration efficacy our approach, show that, while current state-of-the-art perform 10-bit HDR databases, greatly improved inclusion tested distorted videos.