Terrain Rendering Project for Defense Intelligence Agency
I used to post frequently on several computer graphics, animation,
and video oriented mailing lists. A contractor working on
research for the Defense Intelligence Agency saw my posts and
asked me questions about how to render terrain and atmospheric
phenomenon. During the conversations he convinced me to
join the effort.
Officially this work was described as "in support of
American troops in the Middle East" due to the politically
sensitive nature of the geography involved. For all practical
purposes this was really to assist in the national defense
This is all considered unfinished/prototype work. A few images
were produced that were displayed to people as examples of the
product this effort could create. The goal was to be able to
produce realistic animations of views of "atmospheric phenomenon"
over the "area of interest". This required effort to model the
ground of the area of interest and then add an accurate rendering
to represent the atmospheric phenomenon.
Some working proofs of early animation concepts were produced
in Lightwave using lower resolution, faux-terrain data. The real
animations of terrain were expected to be done in World Construction
Set using satellite-collected, DTED data sets for the surface
topography. The atmospheric phenomenon data was to be generated
by another group of people on a supercomputer. I was to write
software to translate the supercomputer-produced data sets into
3D graphic objects representing the shape and density (read:
toxicity) of the atmospheric phenomenon and Lightwave would
render the atmospheric objects over the terrain images rendered
The contract was cancelled for basically unimportant reasons
unrelated to the actual work. All the work that was completed
was done on the Amiga. Our bid for the contract price was only
about 20% (if I recall correctly) of the next lowest contract
First Color Map Rendered Terrain Image
This is the first presentable rendering of terrain in the area of
interest done using Lightwave. The data used for the ground is
VERY inaccurrate. What you see is simply a satellite photo taken
of the area of interest laid down on a flat plane.
The color picture was converted to a black and white photo
(Art Department Professional) which I then
touched up by hand (Deluxe Paint IV) to simulate the land
topography with high points as white and low points as
black and intermediate levels as shades of grey. This image was
then laid down on the plane to describe the altitude of the
ground. It it very inaccurate, but it looks convincing.
First Animation Frame of Color Map Rendered Terrain
This is the first frame of an animation (150 frames, I think)
done entirely in Lightwave to demonstrate the concept of doing flybys over
terrain. The camera view moves closer to the terrain and flies a
couple circles around the land. Halfway through the animation the
camera spins around to look behind itself, so that the viewer is
traveling backward. One very clever feature of the animation is that
the words labelling points of interest in the view continually turn
to face the camera, so that they are always readable.
Latitude and Longitude lines Overlaid on Terrain
This is a test frame of my experiments to use features of
Lightwave to draw graphics lines on terrain.
The purpose was to be able to draw illustrations and
map points of interest on the land.
The ground is grey, because the color image used is actually
the greyscale picture used for the altitude mapping.
One ugly problem is caused by the map supplying the green lines
being too low-res. So, Lightwave "smooothed" the lines
by blurring them which makes the lines look fuzzy. I suppose
this could be useful in the right circumstances, but not here.
Last Animation Frame of Atmospheric Rendering Test
This is the last frame of an animation done in Lightwave to
demonstrate how atmospheric phenomenon might be rendered.
In the animation a small cloud expands and drifts with the
wind. The various colors represent density of the atmospheric
The purpose is just to simulate the visual representation, not
show real data. The ground is the same ground used for the
inital flyby animation, used here only to show some kind of
terrain in the view. Accuracy was not important. For
instance, if this view occurred in the actual
area of interest, the cloud expansion from east to west
would instead progress from west to east.
One ugly problem with the rendering is that the scale for the
altitude mapping was increased dramatically. This is what gives
the land the horribly jagged and vertical appearance. I don't
know why I wasn't paying more attention and fixed it before
rendering. Probably, because I was busy making the clouds look
Early Test of Rendering DTED Terrain Data
This is a rendering done in World Construction Set of a large set of
the satellite-collected digital terrain data (DTED) of the area of
Note that the DTED data only describes altitude. Ordinarily
without any other kind of information the program will render this
generic altitude data in grey. Great for doing lunar simulations.
Bad for earth terrain.
A very simple option supported by the World Construction Set
program is to assign
types of terrain (called ecosystems) to varying altitutes. In this
picture the bulk of the land is assigned a "grassy" ecosystem,
while the highest elevations are assigned "snow". The program is
also smart enough to transition the green grass to bare rock before
rendering snow caps on hills and it knows that snow collects in
gullies at slightly lower altitudes than the actual snow line.
This generic ecosystem simulation makes the rendering look much better,
but it still is very unrealistic. In the distance is a
large flat area that should be rendered as a large lake. The
problem with this area of terrain is that the lake is at a higher
altitude than much of the land in the foreground. In fact, some
of the foreground land is even below sea level. So, if the
altitude method was used to choose to render water at sea level
or at the lake level, then water would cover the land in the
The solution to the problem of rendering ecosystems on terrain is
to have data that describes the ecosystems on the land in addition
to the simple altitude data. While World Construction Set supports
some official formats for ecosystem data it turned out to be
next to impossible to aquire the appropriate Ecosystem data for the
area of interest. A manual solution was devised for the
Rendering DTED Terrain Data with Ecosystem Color Maps
This is a rendering done in World Construction Set of a the
satellite-collected digital terrain data (DTED) of the area of
This view is close to the lake that was rendered in the previous
picture as a flat, grassy plain. This improved rendering was
achieved by exploiting World Construction Set alternate method
for specifying Ecosystem information.
I used World Construction Set to save the DTED data in sections as
graphics pictures. I loaded the pictures into a paint program
and every place I could identify as the lake I painted the same
color (blue, in this case, for easy reference) and black for
everything that is not part of the lake. Then I instructed World Construction Set
to use the blue color in the pictures as a water ecosystem.
Finally I directed World Construction Set to load the
modified graphics pictures and use
them as ecosystem information for the corresponding terrain.
The end result is the lake renders where it is supposed to be.
To produce a truly convincing rendering using real ecosystems
(in lieu of having real ecosystem data) a lot of time would
have to be spent analyzing satellite photos and hand tuning
pictures used as ecosystem information.