#DEMONSTRATIVE EVIDENCE
03-Feb-94 13:54:52
Sb: #80968-#DEMONSTRATIVE EVIDENCE
Fm: John Serth 73553,3314
To: Rick Miller 76170,3214
Rick,
For my monitor / scanner a 8.5" X 6.3" photograph fills the background
on a 3ds sceen. It may be the same for all but it may change for
different scanners, I don't know. The important thing is to center
the photo.
If you stood in the center of a straight flat road and photographed
the road aiming your real camera at the point where the road vanishes
into the horizon, this point would be in the center of the photo and
the paint lines would be ballanced on both sides of the centerline.
Next, if you scanned this photo into 3ds and it tiled, the center of
the photo would not be in the center of the monitor.
Now assume you place the 3ds camera on the centerline of the model and
adjust the camera so that the 3ds camera lines up with the offcenter
vanishing point on the photo. The edges of the road will not be
ballanced in the 3ds model as they are in the photo. You might be
able to move the camera to the side of the road and fudge the error
but you are not reprodusing the origional conditions. If the expert
on the other side explained this to the jury he would probably feel
better about his day in court than you would. Very ungood!
The photo in step 2 is the photo or photos you use for the backgrounds.
What ever you see in a photo is the light coming into the camera from
a unique angle. Think spherical coordnates. You don't have to but, if
you placed a long cylinder in the 3ds model, you could move and
rotate that cylinder so that in the camera viewpoint you would see a
circle, the end of the cylinder, on the point of interest in the
photo. The cylinder is now covering all of the possible points where
the light could have come from. Where the cylinder intersects an
object, flat or hill or any 3d object is where the light came from.
If you have two different viewpoints think of two cylinders
intersecting and measure the x,y,z of objects that you have no
measurements to. As with any triangulation, views at 90 degrees from
each other are going to be most accurate.
After using this method for awile I did find a paper on it S.A.E.
paper930904 It explains the theory but l feel it has shortcommings on
the practical application. Use his theory and my method.