Back to Search
Document details
CE13: HEC with Pre-rotation based on test 1.1a and 1.1b (Test 4.2)
Abstract
This contribution reports simulation result of CE13 test 4.2 Hybrid Equiangular Cubemap (HEC) implementation over 360Lib8.0 merged with VTM3.0. This test is implemented with both test 1.1a (HEC with in-loop filters disabled) and 1.1b (PHEC with in loop filter disabled). Test 4.2a shows the gain of {-0.96%, -1.54%, -1.33%} over CTC anchor and {-0.95%, -1.25%, -1.20%} over CE anchor (test 1.1b), while Test 4.2b shows the gain of {-0.40%, -1.18%, -1.32%} over CTC anchor and {-0.39%, -0.89%, -1.19%} over CE anchor (test 1.1b) under 360 Common Test Condition (CTC) for E2EWSPSNR. Gains are observed majorly in 8k sequences while 6k sequences did not show any loss. We have used a coarse motion estimation algorithm that decides if rotation need to be applied or not. Maximum gain is observed for Gaslamp sequence, which is around -2.71% for test 4.2b.
JVET-M0235 CE13: HEC with Pre-rotation based on test 1.1a and 1.1b (Test 4.2) [C. Pujara, A. Konda, A. Singh, R. Gadde, W. Choi, K. Choi, K. P. Choi (Samsung)] [late]
References:
Cited:
Decisions
JVET-M0235 CE13: HEC with Pre-rotation based on test 1.1a and 1.1b (Test 4.2) [C. Pujara, A. Konda, A. Singh, R. Gadde, W. Choi, K. Choi, K. P. Choi (Samsung)] [late]
Citation