When using Google Earth, people can take for
granted the ability to zoom in, rotate, pan and tilt
on an image
The trick of Google Earth is not in compiling and storing all these
images, but in getting them to your computer quickly and
Do you know?
The Earth is approximately 40,000 km around the waist. If you
stored only one pixel of color data for every square kilometer of
surface, a whole-earth image would be about 40,000 pixels wide and
half as tall.
=> With a 56k dial-up modem, it would take 12,400 years to
download a one-meter resolution image of the Earth.
How Google earth deal with
Decrease the transfer time:
One way to cut down on transfer time is by using your computer's
disk cache, If your computer can temporarily store viewed images of
the Earth, it won't need to retransfer the images from the source.
how your computer processes and displays
Universal Texture is used.
Universal Texture uses two methods for
getting large amounts of information to your
computer Mip-mapping & clip stacking
Mipmaps are collections of bitmap images that work within a texture
to create the illusion of depth. They work in an inverted pyramid
structure, stacked on each other with each level having twice the
resolution of the one under it.
These stacks ultimately create a three-dimensional representation
of a two-dimensional image.
So, for example, if you provided the hardware with a nice 512×512
source image, it would compute and store 8 extra mip-levels for you
(256, 128, 64, 32, 16, 8, 4, 2, and 1 pixel square)
Clip stacks are portions of giant mip-maps that are clipped to a
specified maximum size. The benefit of clipping these enormous
files is simple - Google Earth relies on the fact that users will only
need to see one piece of the mip-map at a time.
When the user tells the program where they want to fly, the
program uses an algorithm to figure out which sections of the larger
virtual texture are needed and uses only those sections to create the
what about the earth case where last level
resolution is about 66 million pixels ?
Google earth clip (divide) the texture level into many smaller distinct
tiles, 256 pixels square, where each gets mip-mapped individually
and instead of loading and drawing the whole-earth texture all at
once — which is impossible on most current hardware- The angle
and height of this virtual column changes depending on our current
Level of Detail (LOD)
This level consist of group of regions.
Google earth use a kml language to represent data of these regions
Region has a <LatLonAltBox> that defines a bounding box for
data, the <LatLonAltBox> in a Region has North, South, East, and
West boundaries. If the data contained by the Region is 3D, or is 2D
at altitude, then the Region's <LatLonAltBox> also needs to include
a minimum altitude, <minAltitude>, and a maximum
Ref. Google earth technology researches
For more information:
Phd researches, Software Developer