Graphics Reference
In-Depth Information
Fig. 6.9 QP predictor
calculation using QP values
from the left, above and
previous QGs [ 21 ]
START
NO
QP ABOVE is available?
QP ABOVE = QP PREV
YES
QP LEFT is available?
NO
QP LEFT = QP PREV
YES
QP PRED = (QP ABOVE + QP LEFT + 1) >> 1
END
The delta QP is transmitted only in coding units with non-zero transform
coefficients. If the CTU is split into coding units that are greater than the QG size,
then delta QP is signaled at a coding unit (with non-zero transform coefficients) that
is greater than the QG size. If the CTU is split into coding units that are smaller
than the QG size, then the delta QP is signaled in the first coding unit with non-zero
transform coefficients in the QG. If a QG has coding units with all zero transform
coefficients (e.g. if the merge mode is used in all the coding units of the QG), then
delta QP will not be signaled.
The QP predictor used for calculating the delta QP uses a combination of QP
values from the left, above and the previous QG in decoding order as shown in
Fig. 6.9 [ 21 ]. The QP predictor uses a combination of two predictive techniques:
spatial QP prediction (from left and above QGs) and previous QP prediction. It uses
spatial prediction from left and above within a CTU and uses the previous QP as
predictor at the CTU boundary. This is shown in Fig. 6.9 . The spatially adjacent
QP values, QP LEFT and QP ABOVE are considered to be not available when they are
in a different CTU or if the current QG is at a slice/tile/picture boundary. When
a spatially adjacent QP value is not available, it is replaced with the previous QP
value, QP PREV , in decoding order. The previous QP, QP PREV , is initialized to the
slice QP value at the beginning of the slice, tile or wavefront.
Search WWH ::




Custom Search