olin_piclist@embedinc.com (Olin Lathrop) wrote:
> Ken Pergola (sent by Nabble.com) wrote:
> > 2) Were the images produced by an Apollo workstation?
> 
> Yes!

108 of them, in fact.

> > 3) Quest1 and Quest2: Was 'Quest' a video game?
> 
> No.
> 
> > 4) Was the graphics engine based on a Motorola 68xxx processor?
> 
> These were all rendered in software.

And yes, the workstations were all 68xxx-based.

> > Well, I'm out of guesses...tag you're it! :)
> 
> OK. The images are two frames from the animated movie Apollo produced
> for SIGGRAPH'85 called "A Long Ray's Journey into Light". The
> significance is that they were ray traced, which everyone at the time
> knew means it took hours to render each frame. This was the first time
> anyone had put together a movie of more than a few seconds that was
> entirely ray traced. Think about the enormous compute power required.
> It took several CPU hours to render each frame, and you needed 24 of them
> just to make one second of animation. If I remember right, the movie was
> about 45 seconds long, which was about half the time of the standing
> ovation it received from the SIGGRAPH audience immediately after. It may
> not seem like a big deal today, but it was a very big deal back then.
> Rendering a few ray traced frames was tough enough, but over 1000 of
> them!!?

Close enough. I'll give you partial credit. I was working at Apollo at the
time, and my own desktop machine participated in the effort.

"Quest: A Long Ray's Journey Into Light" was actually the *second* such
movie produced by the network at Apollo. The first was called "Fair Play"
and was about 1:45 of actual ray-traced animation (the opening and closing
credits add to the length of the overall recording). "Quest" is actually
slightly over 2 minutes long, and the credits say that it was accomplished
in 7 weeks of elapsed time, using 52,000 hours of CPU time. In other words,
the 108 machines spent an average of about 10 hours each day on the
project.

> To put this in perspective, Al Barr got major applause at SIGGRAPH'82
> when he showed about a 10 second ray traced animation of an object
> rotating. His object was deliberately chosen to have symmetry, so he only
> had to render 1/4 of the frames and just repeated them 3 more times to
> show the object rotating a full circle.  Still everyone was quite
> impressed, and it took him many nights on a Prime 750 to render the
> images.  The Apollo movie was only 3 years later.

... and the animations of the myriad objects in the Apollo movie are all
out of sync to prove that no such shortcut was taken -- every frame was
fully and independently rendered.

> Apollo made this movie to point out the aggregate computing power of a
> large network of "little" machines.  There were over 1000 machines on the
> Apollo network at the time, and special software was written so that you
> could allow your node to participate in the collective rendering when you
> weren't using it, like at night.  Really only Apollo could have done this
> at the time.  Nobody else had that kind of untapped computing power
> hanging around. Others, like government labs, had substantial computing
> power, but it was expensive and needed to be put to good use.

It was done in the same manner as SETI@Home was done more recently -- work
was farmed out to idle machines by a master server during off hours.
Therefore, the group of engineers who conceived and managed this called
themselves "The Midnight Movie Group".

-- Dave Tweed
-- 
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist