Game Development Reference
In-Depth Information
(
(
M
1
)
−
1
i
+
1
)
×
(
r
[
(
M
1
)
+
1
+
i
]
ih
=
(4.27c)
−
r
[
(
M
1
)
−
1
−
i
]
i
=
0
(
(
N
1
)
−
1
+
)
×
(
)
+
+
i
1
c
[
(
N
1
1
i
]
v
=
i
(4.27d)
−
(
)
−
−
c
[
N
1
1
i
]
j
=
0
ib
mult
[
5
]={
13
;
17
;
5
;
11
;
23
}
(4.27e)
ib
shift
[
5
]={
7
;
10
;
11
;
15
;
19
}
(4.27f)
For bilinear prediction, prediction sample
P
[
x
,
y
]
is calculated as:
x
val
×
N
+
y
val
×
M
+
((
M
+
N
)
1
)
M
+
N
,
if
(
x
val
+
y
val
)
≥
0
P
[
x
,
y
]
=
,
(4.28)
x
val
×
N
+
y
val
×
M
−
((
M
+
N
)
1
)
,
otherwise
M
+
N
where
x
val and
y
val are computed as follows,
c
[
y
+
1]
+
((
kx
×
(
x
+
1
)
+
(
M
1
))
Log
(
M
)) ,
if
kx
≥
0
x
val
=
c
[
y
+
1]
+
((
kx
×
(
x
+
1
)
−
(
M
1
))
Log
(
M
)) ,
otherwise
(4.29a)
r
[
x
+
1]
+
((
ky
×
(
y
+
1
)
+
(
N
1
))
Log
(
N
)) ,
if
ky
≥
0
y
val
=
r
[
x
+
1]
+
((
ky
×
(
y
+
1
)
−
(
N
1
))
Log
(
N
)) ,
otherwise
(4.29b)
and
kx
,
ky
are computed as follows,
c
[
N
]
×
N
−
r
[
M
]
×
N
+
((
M
+
N
)
1
)
M
⊨
⊩
r
[
M
]
+
×
(
y
+
1
)
+
(
N
1
)
Log
(
N
)
+
N
−
c
[
y
+
1]
,
if
(
c
[
N
]
×
N
−
r
[
M
]
×
N
+
((
M
+
N
)
1
)
≥
0
kx
=
c
[
N
]
×
N
−
r
[
M
]
×
N
+
((
M
+
N
)
1
)
M
r
[
M
]
+
×
(
y
+
1
)
−
(
N
1
)
Log
(
N
)
+
N
−
c
[
y
+
1]
,
otherwise
(4.30a)
r
[
M
]
×
M
−
c
[
N
]
×
M
+
((
N
+
M
)
1
)
M
⊨
⊩
c
[
N
]
+
×
(
x
+
1
)
+
(
M
1
)
Log
(
N
)
+
N
−
r
[
y
+
1]
,
if
(
c
[
N
]
×
N
−
r
[
M
]
×
N
+
((
M
+
N
)
1
)
≥
0
ky
=
r
[
M
]
×
M
−
c
[
N
]
×
M
+
((
M
+
N
)
1
)
M
c
[
N
]
+
×
(
x
+
1
)
−
(
M
1
)
Log
(
M
)
+
N
−
r
[
y
+
1]
,
otherwise
(4.30b)
