Performance of ITU-T P.1204.4 on Video Encoded with AV1, AVS2, VVC

The recent standard P.1204.4 contains a reduced-/full-reference video quality assessment model for the prediction of mean opinion scores for videos encoded with codecs H.264, HEVC, VP9. In this paper the performance of P.1204.4 on videos encoded with new codecs AV1, AVS2, VVC outside of the scope of the standard P.1204.4 is analyzed using two open datasets, the Waterloo IVC-Lab dataset and the Bristol VI-Lab dataset. The pixel-based model should be codec agnostic. This is analyzed by comparing subset performance on data encoded with new codecs versus within-scope codecs. The equally good performance on the subset of samples encoded with new codecs relative to the performance of the model for data using the codecs within scope of the standard indicate the possibility to use the model for data encoded with these new codecs.