Re: A Cellular-Automata Challenge

From: Arnim Sauerbier (arnim@gene.med.umn.edu)
Date: Mon May 10 1993 - 19:29:24 UTC


Joerg Richard Weimar (jweimar@ulb.ac.be) wrote:
: 3. If the camera is slightly rotated around he optical axis with respect  
: to the monitor, you get interesting effects, but this introduces a global  
: rotation, which is practically impossible to mimic with CA's (because  
: there is a global reference point)

Yes, there are many interesting ways to adjust the parameters of the feed-
back.   Holding a mirror adjacent to the monitor so that the camera can
see the monitor surface reflected in the mirror creates truly incredible
patterns.  I believe that the mirror combined with feedback creates multiple
copies of ever-smaller screen-images, each copied recursively onto the final
image with a fixed rotational offset and scaling factor.  This is the video
analogy of the famous "photocopier fractal"  that one can generate by 
iteratively photoreducing and skewing an image drawn on paper.   

 
: To conclude: with a CA you could only mimic the experiments where all  
: controls are fixed and the camera doesn't do any automatic adjustments.  
: Also the camera needs to be aligned exactly with the monitor, and the  
: field of view must be exactly filled with the monitor.
: I doubt that the patterns in this regime are as interesting as the ones  
: you get.

Indeed, this is the problem, none of the Cellular-Automata algorithms I've
seen take into account the position of the cell when computing its behavior.

To trivially emulate video-feedback, perhaps one could simply recursively
superimpose the original image onto the 'screen-space' with user-steerable
skew, rotation, and magnification.   

To superimpose 100 copies of a 320x200 image, one would need approx 6,400,000 
'superimposing-operations'.  To do this at 30 times per second.... Wow.

Anyone have a Cray to spare? 

_\\\/ ----------------------------------------------------------------------
= ..' 	Arnim Sauerbier		"The United States constitution isn't per- 
C  \)	arnim@gene.med.umn.edu   fect, but it's better than what we've got"
 \ -  ----------------------------------------------------------------------


This archive was generated by hypermail 2.1.7 : Tue Oct 14 2003 - 21:44:13 UTC