[ExI] Colours
Adrian Tymes
atymes at gmail.com
Sat May 4 18:57:36 UTC 2013
On Sat, May 4, 2013 at 9:38 AM, spike <spike at rainier66.com> wrote:
>
> HEY that gives me an idea. Perhaps we CAN mathematize fashion, at least
> to some limited extent. Simple proof: take any pile of fashion photos,
> digitize same, this converts every conceivable fashion into a series of
> pixels, each of which consists of some color, which is a frequency, which
> is a number. These digitized images can be made into any standard size,
> say a 1 MB file, 1k by 1k, so this can be a matrix with each cell a
> frequency on some standard background. Anything wrong with any step in my
> reasoning so far?
>
>
>
> If every conceivable fashion can be reduced to a 1 MB 2D matrix, then we can start to work with our mathematical bag of tricks. We bring in our fashion gurus, such as that guy from Queer Eye and Yves St. Lauren for instance, to separate the matrices into hip and non-hip fashions. Once we have that, then we can do all kinds of cool tricks: finding mathematical correlations, perhaps use all our advanced digital filtering tools in Matlab and so forth. Then we should be able to extrapolate a hip fashion from the known existing ones. Then totally square L7 tiny microscopic SPIKE can become the WORLD FASHION DICTATOR! I will have discovered the mysterious KEY to something which has puzzled me endlessly. I will RULE the FASHION WORLD, with an IRON FIST, Muuwaaaaahahahahahhahaaaaaaa…
>
> You assume there are such correlations to be found. To start with, to
be pictures that can be compared, the fashions must be pictured on
the same model from the same angle under the same lighting
conditions - which is a tougher job that it seems. It's theoretically
possible, but it will take a lot of time.
Also, you'll need negative data - stuff that is not fashionable. And
you'll need this over time. Such data is imprecise to say the
least: error rates over 10%.
My hunch is, you'd find the noise drowns out any signal from this
approach.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.extropy.org/pipermail/extropy-chat/attachments/20130504/7c04594a/attachment.html>
More information about the extropy-chat
mailing list