On 10/10/2014 05:40 PM, Norbert Kern wrote:
>
> IMHO, it's understandable seeing at the center.
> In short, rotation + warp repeat y flip y acts as a kaleidoscope, as long as you
> rotate 180 degrees (ncount_max * fA) or a multiple of it.
>
> To show, what I mean, I add an image with code spippets included.
> Distance from origin is needed to get rid of the kaleidoscope effect...
>
> Norbert
>
Hi Norbert,
I agree with your outline for how the method works. However, at a 
fundamental level this trick still confuses me. Attached is a series of 
4 renders of a unit square centered at the origin. On the left is just 



All good thus far in that it makes sense to me. The introduction of the 
warp in the third introduces no change as I'd expect given the repeat 
should not happen until +- 1.0 in Y. Now for the final image on the 

nothing more than what I would see from the two rotations of the 
original pattern.

Instead, there is a discontinuity in the pattern in the upper left and 
lower right quadrants. I believe this a necessary component of the 
method, but I do not really understand what happened to pull the repeat 

quadrants of the square +-0.5 in x&y. In other words I expect the second 

than the original rotated -36*z pattern into the first warp.


and this continues in the full method as we layer the repeats. Is the 
trailing repeat partly folding or collapsing the spatial coordinates of 
the one before it?

We get very cool results I suppose no matter whether I can understand 
things...




and I thought - more interesting result.

Bill P.

Post a reply to this message

Attachments:
Download 'z.jpg' (40 KB) Download 'dualflips.jpg' (21 KB)

Preview of image 'z.jpg'

Preview of image 'dualflips.jpg'