<s>
In	O
video	B-Algorithm
coding	I-Algorithm
,	O
a	O
group	B-General_Concept
of	I-General_Concept
pictures	I-General_Concept
,	O
or	O
GOP	O
structure	O
,	O
specifies	O
the	O
order	O
in	O
which	O
intra	B-General_Concept
-	I-General_Concept
and	O
inter-frames	B-Algorithm
are	O
arranged	O
.	O
</s>
<s>
I	O
picture	O
or	O
I	B-General_Concept
frame	I-General_Concept
(	O
intra	O
coded	O
picture	O
,	O
also	O
called	O
keyframe	O
or	O
i-frame	B-General_Concept
)	O
–	O
a	O
picture	O
that	O
is	O
coded	O
independently	O
of	O
all	O
other	O
pictures	O
.	O
</s>
<s>
P	O
picture	O
or	O
P	B-General_Concept
frame	I-General_Concept
(	O
predictive	O
coded	O
picture	O
)	O
–	O
contains	O
motion-compensated	B-Algorithm
difference	O
information	O
relative	O
to	O
previously	O
decoded	O
pictures	O
.	O
</s>
<s>
In	O
older	O
designs	O
such	O
as	O
MPEG-1	B-Algorithm
,	O
H.262/MPEG	O
-2	O
and	O
H.263	B-Algorithm
,	O
each	O
P	O
picture	O
can	O
only	O
reference	O
one	O
picture	O
,	O
and	O
that	O
picture	O
must	O
precede	O
the	O
P	O
picture	O
in	O
display	O
order	O
as	O
well	O
as	O
in	O
decoding	O
order	O
and	O
must	O
be	O
an	O
I	O
or	O
P	O
picture	O
.	O
</s>
<s>
These	O
constraints	O
do	O
not	O
apply	O
in	O
the	O
newer	O
standards	O
H.264/MPEG	B-Application
-4	I-Application
AVC	I-Application
and	O
HEVC	B-Algorithm
.	O
</s>
<s>
B	O
picture	O
or	O
B	B-General_Concept
frame	I-General_Concept
(	O
bipredictive	O
coded	O
picture	O
)	O
–	O
contains	O
motion-compensated	B-Algorithm
difference	O
information	O
relative	O
to	O
previously	O
decoded	O
pictures	O
.	O
</s>
<s>
In	O
older	O
designs	O
such	O
as	O
MPEG-1	B-Algorithm
and	O
H.262/MPEG	O
-2	O
,	O
each	O
B	O
picture	O
can	O
only	O
reference	O
two	O
pictures	O
,	O
the	O
one	O
which	O
precedes	O
the	O
B	O
picture	O
in	O
display	O
order	O
and	O
the	O
one	O
which	O
follows	O
,	O
and	O
all	O
referenced	O
pictures	O
must	O
be	O
I	O
or	O
P	O
pictures	O
.	O
</s>
<s>
These	O
constraints	O
do	O
not	O
apply	O
in	O
newer	O
standards	O
H.264/MPEG	B-Application
-4	I-Application
AVC	I-Application
and	O
HEVC	B-Algorithm
.	O
</s>
<s>
D	O
pictures	O
are	O
only	O
used	O
in	O
MPEG-1	B-Algorithm
video	O
.	O
</s>
<s>
An	O
I	B-General_Concept
frame	I-General_Concept
indicates	O
the	O
beginning	O
of	O
a	O
GOP	O
.	O
</s>
<s>
Afterwards	O
several	O
P	O
and	O
B	B-General_Concept
frames	I-General_Concept
follow	O
.	O
</s>
<s>
Typically	O
,	O
encoders	O
use	O
GOP	O
structures	O
that	O
cause	O
each	O
I	B-General_Concept
frame	I-General_Concept
to	O
be	O
a	O
"	O
clean	O
random	O
access	O
point	O
,	O
"	O
such	O
that	O
decoding	O
can	O
start	O
cleanly	O
on	O
an	O
I	B-General_Concept
frame	I-General_Concept
and	O
any	O
errors	O
within	O
the	O
GOP	O
structure	O
are	O
corrected	O
after	O
processing	O
a	O
correct	O
I	B-General_Concept
frame	I-General_Concept
.	O
</s>
<s>
In	O
the	O
newer	O
designs	O
found	O
in	O
H.264/MPEG	B-Application
-4	I-Application
AVC	I-Application
and	O
HEVC	B-Algorithm
,	O
encoders	O
have	O
much	O
more	O
flexibility	O
about	O
referencing	O
structures	O
.	O
</s>
<s>
The	O
second	O
one	O
tells	O
the	O
distance	O
between	O
two	O
full	O
images	O
(	O
I-frames	B-General_Concept
)	O
:	O
it	O
is	O
the	O
GOP	O
size	O
.	O
</s>
<s>
Instead	O
of	O
the	O
M	O
parameter	O
the	O
maximal	O
count	O
of	O
B-frames	B-General_Concept
between	O
two	O
consecutive	O
anchor	O
frames	O
can	O
be	O
used	O
.	O
</s>
