Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!sun-barr!olivea!samsung!usc!elroy.jpl.nasa.gov!decwrl!stanford.edu!leland.Stanford.EDU!pangea.Stanford.EDU!rick
From: rick@pangea.Stanford.EDU (Rick Ottolini)
Newsgroups: comp.graphics
Subject: Re: Digital Holography
Message-ID: <1991May10.233224.13587@leland.Stanford.EDU>
Date: 10 May 91 23:32:24 GMT
References: <1019.282B28FD@nwark.fidonet.org>
Sender: news@leland.Stanford.EDU (Mr News)
Organization: Stanford Univ. Earth Sciences
Lines: 43

In article <1019.282B28FD@nwark.fidonet.org> Samuel.P..Uselton@p0.f13.n391.z1.fidonet.org (Samuel P. Uselton) writes:
>Newsgroups: comp.graphics
>
>In article <1991May9.153446.21742@leland.Stanford.EDU> rick@pangea.Stanford.EDU (Rick Ottolini) writes:
>>They are similar to other imaging mathematics such as my field
>>of seismic imaging.  
>	I've also consulted with "Big Oil".  Seismic imaging is more similar
>	to image processing and scene analysis than to image generation
>	techniques.  You HAVE the image, and are trying to guess the most
>	likely scene which could have created it.   

I represent Big Oil.  Seismics is my business and graphics an avocation.
The two disciplines use approximately the same universe of algorithms,
but in different porportions.  The cross-fertilization of ideas is fruitful.

>>Even with all kinds of shortcuts thrown in, it will take
>>billions to trillions of calculations per second to display interesting
>>holographic images.  
>	Current realistic image synthesis techniques can take from 100 million
>	 to 1 billion operations per image.  Laser holography is AT LEAST
>	a couple of orders of magnitude more.  And you still WANT the animation
>	so add another one or two orders of magnitude.  I see trillions of
>	operations per second as a LOWER bound on what it might take.
>	The NAS project at NASA Ames regards pushing industry into producing
>	a teraflops computer by the year 2000 as a "Grand Challenge" problem.
>	It'll be quite a while longer before that capacity finds its way
>	into workstations for the broad market.

A typical seismic imaging algorithm that took 100,000 seconds in the early 1970s
takes about a second these days.  Two orders of magnitude are due to smarter
algorithms and three orders of magnitude are due to that my desktop RS/6000 is
a thousand times faster than my old PDP-11/34.  These improvements will continue
for both seismics and graphics.

>>With the computing speeds increasing an order of magnitude
>>every five years and no end in sight, 
>	          ^^^^^^^^^^^^^^^^^^^
>	There is a growing number of "experts" pointing out limits to current
>	hardware techniques that we ARE rapidly approaching.  

I've heard this doom and gloom for the past 15 years and remain unconvinced.
"Breakthroughs" aren't always obvious when they start.  There is enough stirring 
in the pot now to keep us occupied for a long time.