Salvato in:
Dettagli Bibliografici
Autori principali: Jindal, Akshay, Sadaka, Nabil, Thomas, Manu Mathew, Sochenov, Anton, Kaplanyan, Anton
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:https://arxiv.org/abs/2506.11546
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866913891648274432
author Jindal, Akshay
Sadaka, Nabil
Thomas, Manu Mathew
Sochenov, Anton
Kaplanyan, Anton
author_facet Jindal, Akshay
Sadaka, Nabil
Thomas, Manu Mathew
Sochenov, Anton
Kaplanyan, Anton
contents While existing video and image quality datasets have extensively studied natural videos and traditional distortions, the perception of synthetic content and modern rendering artifacts remains underexplored. We present a novel video quality dataset focused on distortions introduced by advanced rendering techniques, including neural supersampling, novel-view synthesis, path tracing, neural denoising, frame interpolation, and variable rate shading. Our evaluations show that existing full-reference quality metrics perform sub-optimally on these distortions, with a maximum Pearson correlation of 0.78. Additionally, we find that the feature space of pre-trained 3D CNNs aligns strongly with human perception of visual quality. We propose CGVQM, a full-reference video quality metric that significantly outperforms existing metrics while generating both per-pixel error maps and global quality scores. Our dataset and metric implementation is available at https://github.com/IntelLabs/CGVQM.
format Preprint
id arxiv_https___arxiv_org_abs_2506_11546
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle CGVQM+D: Computer Graphics Video Quality Metric and Dataset
Jindal, Akshay
Sadaka, Nabil
Thomas, Manu Mathew
Sochenov, Anton
Kaplanyan, Anton
Graphics
Computer Vision and Pattern Recognition
While existing video and image quality datasets have extensively studied natural videos and traditional distortions, the perception of synthetic content and modern rendering artifacts remains underexplored. We present a novel video quality dataset focused on distortions introduced by advanced rendering techniques, including neural supersampling, novel-view synthesis, path tracing, neural denoising, frame interpolation, and variable rate shading. Our evaluations show that existing full-reference quality metrics perform sub-optimally on these distortions, with a maximum Pearson correlation of 0.78. Additionally, we find that the feature space of pre-trained 3D CNNs aligns strongly with human perception of visual quality. We propose CGVQM, a full-reference video quality metric that significantly outperforms existing metrics while generating both per-pixel error maps and global quality scores. Our dataset and metric implementation is available at https://github.com/IntelLabs/CGVQM.
title CGVQM+D: Computer Graphics Video Quality Metric and Dataset
topic Graphics
Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2506.11546