JVET-AL0054 AHG17: adjusted QPs for non-CTC sequences [E. François (InterDigital), E. Alshina (Huawei), P. Nikitin (Qualcomm)]

This contribution comprises an updated spreadsheet from JVET-AL0053 with proposed adjusted QPs based on results of viewing performed during interim Ahg17 meeting (March 10-12, 2025). The target of QPs adjustment was to provide recommendations for target rates per sequence, covering a quality range from very bad to very good, with four well distinguishable (during viewing) intermediate steps.

In v1, the QPs adjustment was made for the sequences initially considered for CfE before the interim meeting.

During this interim meeting, new sequences have been considered and the set of CfE candidate sequences has been updated. In v2 of the current document, the QPs adjustment has been completed for all the candidate sequences selected for CfE during the interim meeting.

Based on the adjustment of QPs, recommendations for target rates for each CfE candidate sequence are provided.

It was suggested to make the “quantization” of rates more coarse for highest bit rate points.

It was further suggested to perform the combination with the MOS results (DCR, or ACR when DCR not available) from JVET-AL0053 and check if the proposed rates are also useful in terms of quality ranges of ECM.

It was commented that correction seems necessary for Gregory Cactus (highest rate may be too high, lowest too low).

It was requested to generate plots where

• When DCR results are available, mark the position of the suggested rates compared to the available rates
• When ACR results are available (only for lowest and highest rate points, and only for ECM), to generate plots where the suggested rates are marked in comparison with the rates used in the test

The rates should be adjusted such that likely ECM quality at lowest rate is around MOS 3, but not approaching MOS 2, and at highest rate not higher than MOS 8. Side activity (M. Wien, E. François, K. Andersson, A. Wieckowski, E. Alshina, P. Nikitin)

It is generally agreed that for most sequences the suggested rates are already more or less appropriate for a CfE, which is also meant as a “dry run” for a later CfP (if successful).