From: bungi@milton.u.washington.edu (Timothy Wood)
Subject: Re: Virtual Mars
Date: 16 Jan 91 01:24:01 GMT
Organization: University of Washington, Seattle



In article <14441@milton.u.washington.edu> scarlson@csa1.lbl.gov (Shawn Carlson)
 writes:
>
  ...
>     Suppose we begin our Martian adventure by deploying a few 
>satellites to take high resolution pictures of the entire surface 
>of Mars.
  ...
>You are flying a simulator and exploring a 
>computer generated "virtual world" that blends those high 
>resolution satellite photos into moving 3-D images and is 
>therefore identical in every detail to the real Martian surface.  

        It seems to me that in order to be even slightly interesting to the
virtual explorers, one would have to be able to image the surface of Mars
with realatively fine resolution no matter how far one was from the virtual
surface.  Mars has a mean radius of something close to 3,400km.  This give
a surface area of about 145.3E6 square kilometers (assuming Mars is
spherical, the actual figure could be much higher, depending upon how
'bumpy' Mars is...)  To obtain an even slightly usefull picture, let's say
that we use 8 bits per square centimeter.  The appoximate size of the file
would be something like 116E15 bits.  Too be _fun_ we would need to be
working on a much smaller scale, say square millimeters.  At this small of
a scale, the slight fractal nature of the surface would cause the area
mapped be be far greater than the spherical approximation, and the
spherical approximation _ain't_ small ( 'bout 11.6E18 bits ).
        If this were a mere pleasure simulation, one could easily use a
rougher scale and make whatever random bumps we wanted.

        My question is this, is it possible to access and image this much
data at a rate that would be necessary for a good graphical simulation?


---
Timothy Wood
tjwood@wolf.cs.washington.edu
---