There are several video demonstrations for video feedback leading chaos such as [1], [2], [3], [3].

My question is how to model it?

If I have such a color matrix of

struct Color
    unsigned char r,g,b;

Color screen[1280][720];

at time $t$, then how can I calculate screen at time $t+\Delta t$?

I am looking for a model with zooming the camera enough so chaos happens.

  • 3
    $\begingroup$ You would need to know how the pixels are arranged on the screen. Replace each pixel by an image of the color mask with intensities given by the pixel values, rotate and scale down. Probably apply during the down-scaling a slight spherical distortion for non-linear lens effects. $\endgroup$ Nov 6, 2017 at 8:32
  • $\begingroup$ @LutzL, Many thanks. So you suggest a Rotation+Scale+Distortion+Noise? Any recommendation for applying a distortion? $\endgroup$
    – ar2015
    Nov 7, 2017 at 2:05
  • $\begingroup$ The feedback itself is independent of the transformation rules. You make a temporary buffer, and make a copy of the screen at $t$ and "add" it to the source at $t+1$. E.g. loop: copy(temp, pixelbuffer); do_something_to(pixelbuffer); add(pixelbuffer, temp); repeat. how you add the pixels is up to you, if you want a fast routine you can use mask like this: pixel[i]&0xfefefe. $\endgroup$
    – user366820
    Jan 8, 2021 at 5:10


You must log in to answer this question.