Back to Search Document details
16th Meeting: Geneva, October 2019 2019-09-30 20:16
Crosscheck of JVET-P0042 (Non-CE7: A cleanup for inter_pred_idc coding)
Authors: K. Panusopone Nokia Krit Panusopone
Abstract
This document reports a crosscheck of contribution JVET-P0042 by Broadcom on a cleanup for inter_pred_idc coding. It is confirmed that the software provided implements the tool as described in the document and the coding results match with those provided in JVET-P0042.
JVET-P0713 Crosscheck of JVET-P0042 (Non-CE7: A cleanup for inter_pred_idc coding) [K. Panusopone (Nokia)]

In VVC, inter_pred_idc syntax which specifies inter prediction mode uses five context models. Especially, the 5th context model is being shared by the 4x16, 8x8, or 16x4 bi-predicted CUs and uni-predicted CUs. This contribution proposes two alternative methods to prevent from the 5th context model being shared between bi-prediction and uni-prediction cases. From experiment results, luma BD-rate changes of method 1 is reported as -0.01% and -0.02% for RA and LDB. luma BD-rate changes of method 2 is reported as -0.02% and -0.03% for RA and LDB.

The solution of JVET-P0042 is simpler and more straightforward to understand for implementers.

References:
JVET-P0042
Decisions
The solution of JVET-P0042 is simpler and more straightforward to understand for implementers.
Citation