<s>
Geomipmapping	B-Algorithm
or	O
geometrical	O
mipmapping	O
is	O
a	O
real-time	O
block-based	O
terrain	O
rendering	O
algorithm	O
developed	O
by	O
W.H.	O
</s>
<s>
de	O
Boer	O
in	O
2000	O
that	O
aims	O
to	O
reduce	O
CPU	B-Device
processing	O
time	O
which	O
is	O
a	O
common	O
bottleneck	O
in	O
level	O
of	O
detail	O
approaches	O
to	O
terrain	O
rendering	O
.	O
</s>
<s>
Prior	O
to	O
geomipmapping	B-Algorithm
,	O
techniques	O
such	O
as	O
quadtree	O
rendering	O
were	O
used	O
to	O
divide	O
the	O
terrain	O
into	O
square	O
tiles	O
created	O
by	O
binary	O
division	O
with	O
quadratically	O
diminishing	O
size	O
.	O
</s>
<s>
The	O
subdivision	O
step	O
is	O
typically	O
performed	O
on	O
the	O
CPU	B-Device
which	O
creates	O
a	O
bottleneck	O
as	O
geometry	O
commands	O
are	O
buffered	O
to	O
the	O
GPU	B-Architecture
.	O
</s>
<s>
Unlike	O
quadtrees	O
which	O
send	O
1x1	O
polygon	O
units	O
to	O
the	O
GPU	B-Architecture
,	O
to	O
reduce	O
the	O
CPU	B-Device
processing	O
time	O
geomipmapping	B-Algorithm
divides	O
the	O
terrain	O
into	O
grid-based	O
tiles	O
which	O
are	O
themselves	O
regularly	O
subdivided	O
.	O
</s>
<s>
Typically	O
,	O
a	O
fixed	O
number	O
of	O
vertex	B-Library
buffer	I-Library
objects	I-Library
(	O
VBOs	O
)	O
are	O
stored	O
on	O
the	O
GPU	B-Architecture
at	O
different	O
grid	O
resolutions	O
,	O
such	O
as	O
10x10	O
and	O
20x20	O
,	O
and	O
then	O
placed	O
at	O
major	O
terrain	O
regions	O
selectively	O
chosen	O
by	O
the	O
CPU	B-Device
.	O
</s>
<s>
A	O
vertex	O
shader	O
is	O
then	O
used	O
to	O
reposition	O
the	O
vertices	O
for	O
a	O
given	O
VBO	O
,	O
all	O
on	O
the	O
GPU	B-Architecture
.	O
</s>
<s>
Overall	O
,	O
this	O
results	O
in	O
a	O
major	O
reduction	O
in	O
CPU	B-Device
processing	O
,	O
and	O
reduced	O
CPU-to-GPU	O
bandwidth	O
as	O
the	O
GPU	B-Architecture
then	O
performs	O
most	O
of	O
the	O
work	O
.	O
</s>
<s>
Geoclipmaps	O
and	O
GPU	B-Architecture
raycasting	O
are	O
two	O
other	O
modern	O
alternatives	O
to	O
geomipmapping	B-Algorithm
for	O
interactive	O
rendering	O
of	O
terrain	O
.	O
</s>
