Re: CA model of TV Chaos?

From: Arnim Sauerbier (arnim@gene.med.umn.edu)
Date: Mon May 10 1993 - 20:02:56 UTC


Rudy Rucker (rucker@sjsumcs.SJSU.EDU) wrote:

: arnim@htlv.med.umn.edu asks if there is might be a CA which would simulate

: To simplify the discussion a bit, one might think in terms of alteranting
: two CA rules, a transform rule and a Laplace rule.  That is, do a flip
: or a rotation or something, then do a Laplacian smoothing.  Quite
: generally, the transform rule would have the form NewVal(x,y) = OldVal(
: T(x,y)) for T a linear transformation like the ones used in Barnsley
: fractals.

I wish i'd read this before I posted my extremely stupid idea for simulating 
the feedback.  Your 'algorithm' makes considerable intuitive sense, although 
I wonder whether NTSC's interlaced signal, and the seperation of chrominance
and luminance introduce non-linearities into the system that significantly
affect the forms that the colors take.  

I think the algorithm would need to include... 

	-An Iterative process where the current state of a pixel is modified
	by the previous state of itself and adjacent pixels in a location-
	specific way (i.e. not* a CA algorithm).

	-"blending" of adjacent pixels to simulate the bleeding that occurs
	in the monitor

	-Translation, scaling, and rotation of the feedback-pixels (one 
	could flip the image via 180deg. rotation)


A respondent to my post kindly sent me the following references.. for the
curious.

-----------------------
@ARTICLE (Crut84a,
    TITLE = "Space-Time Dynamics in Video Feedback",
    AUTHOR = "J. P. Crutchfield",
    JOURNAL = "Physica",
    VOLUME = "10D",
    YEAR = 1984,
    PAGES = 229,
    KEY = "ThePaper")

@ARTICLE (Crut88c,
    TITLE = "Spatio-Temporal Complexity in Nonlinear Image Processing",
    AUTHOR = "J. P. Crutchfield",
    JOURNAL = "IEEE Trans. Circ. Sys.",
    YEAR = 1988,
    PAGES = 770,
    VOLUME = 37)



-Arnim Sauerbier (arnim@gene.med.umn.edu)	
-----------------------------------------------------------------
define s(n){;auto e,r;for(0;n;n/=b)r+=n%b*(b+1)^s(e++);return(r)}
define f(n){;for(b=2;n;b++)n=s(n)-1;return(b)} for(n=0;1;n++)f(n)
f() is total -- R.L. Goodstein


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