GMU:Dataflow I WS12/Jeremy: Difference between revisions

From Medien Wiki
No edit summary
Line 92: Line 92:


Black cloudbank broken
Black cloudbank broken
==limitations==
The moses patch is at prototype stage and obviously has many limitations. These include:
Limited range of outcomes from images of similar tone. Black and white images will mostly return a similar result unless the text file is changed in between.
Only 48 possible lines of text.
The scale of numbers that can be triggered need to be more finely studied in relation to the most common ranges of values.
Ideally the patch would export a new text file in response to each image. ideally, the patch would somehow combine the text and image into a new jpg file, for example.
It would also be great if the patch would communicate with another patch, which would jumble the lines of poetry for it.
These considerations all require further work

Revision as of 16:30, 29 January 2013

440px-Hugo Ball Cabaret Voltaire.jpg

Hugo Ball at the Cabaret Voltaire, Zurich, 1916

photographer unknown

background

I wanted to use Pure Data to perform some kind of human function, to reveal some form of judgement or thought process which would speak about its own inability to do so. And so I wanted to create some kind of patch that would interest itself in the nature of computer programming, while also displaying some level of self awareness.

I decided to pursue the idea of a dadaist rearrangement of text based on images. I was thinking about the ways in which the Dadaists used the idea of automatic and chance-based occurances in order to remove or at least diffuse their own artistic control and intention over an artwork. This led me to the idea that images could be an interesting starting point to perform and compose and thereby create the illusion of some form of personality or discernmen.

Moses

My patch is called Moses, it composes haiku poetry, or rather, recomposes the lines of poetry that are available to it in response to an image.

The images used are all sourced from the internet and have no known photographer. In selecting these images I am interested in the idea of a deficient human element. Found images can sometimes feel personal to us and the narrative elements reveal themselves quickly. At the same time, an everyday or observed feeling reminds us that we can detach ourselves from them and their subjects quite readily.

The text being used is sourced from Japanese haiku poetry, translated into English. The poets include Buson (1715-1783), Basho (1644-1694), and Buson (1715-1783) and other major poets to make use of the Haiku.

The Haiku often presents a contrast or a 'cutting' aspect, including the juxtaposiition of text and image. This is a key aspect of my use of the two, and allowing a pure data patch to perform a simple and traditional act.

how does it work?

Moses reads the RGB levels in an image by using the 'pix mean color' object. These levels are 'unpacked' and channeled into three different but identical sub-patches. Here, the levels filter through an number of 'moses' objects that result in the incoming value aligning with one of sixteen possible numbers. These numbers are outputed from each of the three identical sub-patches into another sub-patch. This sub-patch uses a 'textfile' object. The three incoming numbers dictate which lines from the textfile are to be used. These lines of text are extratced from the file and printed into the dialogue window.

File:Moses.pd

File:Haiku.SacredTexts.1.txt

documentation

1407 girl024 med.jpg

Comical Dutch script

Flip-flapping across the sky

In this windy nest


Student woodworking class.jpg

On the death of his child

And dry my dreaming but still ...

Insatiable fleas


Woman.jpg

Women planting rice

Drawn up from my frozen well ...

Weird hollow echo


Unknown diver.jpg

Issa stepchild bird

Dew evaporates and all our world is dew ...

Black cloudbank broken


Maori.jpg

But their ancient song

Flip-flapping across the sky

Wild geese write a line


Portraitofachild.jpg

The songs of froglings

Our old scarecrow topples down

In silent midnight


Girl 1943.jpg

White they meet

Scatters in the night ... now see

Black cloudbank broken

limitations

The moses patch is at prototype stage and obviously has many limitations. These include:

Limited range of outcomes from images of similar tone. Black and white images will mostly return a similar result unless the text file is changed in between.

Only 48 possible lines of text.

The scale of numbers that can be triggered need to be more finely studied in relation to the most common ranges of values.

Ideally the patch would export a new text file in response to each image. ideally, the patch would somehow combine the text and image into a new jpg file, for example.

It would also be great if the patch would communicate with another patch, which would jumble the lines of poetry for it.


These considerations all require further work