More selected projects

A Word Says One Thousand Six Hundred Pictures

A Word Says One Thousand Six Hundred Pictures Is a novel google image search

produced by: William Parry

Introduction

A Word Says One Thousand Six Hundred Images Is a novel google image search that plays on the understanding of visual perception. Using this search it is difficult to understand what the overall search image symbolises, despite one thousand six hundred other representations of it on screen.

Concept and background research

When brainstorming for ideas to begin the project I came across a memory of when I was at an arts festival in Winchester. The festival had a specific theme of capital control, and with that there were areas themed to Sin .In the ‘Sin Area’ there was a large LED motorway sign style matrix board that was playing back pornographic scenes. As the matrix was of a low resolution and in one color it took a while for people to notice it. This reminded me of an Image I’d seen "Studies in Perception", “a computer generated nude... Using small electrical symbols” by Kenneth Knowlton and Leon Harmon. Both pieces have a link to image perception and sexual gaze. I aimed to play on the warping of visual perception in my project..

Technical

The overall design works by creating a low resolution reproduction of one google image search result, through the tinting and copying of four hundred google image results. The main struggle was finding a way of getting the image live form a google search into processing.

For this I used a piece of code by a Jeff Thompson contributed to a github page. The code retrieved 20 URLs in one execution from the google image search API using a search term and an offset to gather more images.

To contain the search function neatly I edited the piece of code to function as a class, the class worked well providing the search term returned some URLs. The URLs where then the source for the ‘loadImage’ object within processing, this would {draw 20 images twice across the page (40 per row), the 40 image row was then drawn starting from the middle of the display downwards.} This was looped 20 times creating a 40 x 40 matrix of images loaded into processing.

The searches are made through an interfacing ‘search bar’ at the top of the page.

If the search word were a ridiculous word that returned no results this loop would break and the processing environment would crash due to a ‘null pointer exception’. To combat this I added a small piece of code to the search class that flagged up and ‘null searches’. This would then stop the loop and ask the user to search for ‘something less ridiculous’.

Future development

I would like to develop this project further by introducing an image file save system on all the google image searches. This way I could scan through the images implementing a colour averaging function to sort the images and position them relative to the macro image, eliminating the need for a ‘tint’ on the overlaying micro images.

Using a file save method would also allow me to render different macro images without the need to download all of the separate micro Images. I would like to explore this method and perhaps run results through a piece of machine vision/ learning algorithm.

Evaluation

The interfacing could definitely be improved by adding a better form of successful/unsuccessful search feedback, and text editing functions.

Loading the images row by row can be quite a cumbersome task to view. Although I feel as though it adds a sense of James Bridles "New Aesthetic", through its simple use of simple digital image editing, and this old style ‘top to bottom’ rendering.

I also surprised myself with consistencies in perception results when using the novel search, when testing the sketch (searching for images then asking friends to see if they could recognised what I’d searched). I found that obviously people find it easy to recognise faces/ images of people. But I also found that people found it very easy to recognise nudes/erotic images. This could be mainly due to the heavy use of skin color, but like the LED matrix example, I think it could be down to human perception of certain shapes and curves

References

Lieser, W., 2010. The World of Digital Art. 1st ed. Brandenburg, Germany: Ullmann Publishing.

Studies in Perspection #1. Computer-produced mural. Bell Labs. Leon Harmon & Ken Knowlton. 1966

James Bridle / The New Aesthetic. [ONLINE] Available at:http://jamesbridle.com/works/the-new-aesthetic. [Accessed 2018].

Google Image Search. Jeff Thompson.[Processing Code]
https://github.com/jeffThompson/ProcessingTeachingSketches/blob/master/AdvancedTopics/GetGoogleImageSearchURLs/GetGoogleImageSearchURLs.pde [Acessed 2018].