Overview

Pietbot is a robot artist designed to emulate the way an artist selects a subject or scene based on evidence of beauty. He also mimics artistic interpretation by manipulating proportions to fit a more beautiful ‘ideal’.

 

The main ‘robot’ part of Pietbot is a white cylindrical pillar approx.1.5m tall x 0.5m diameter.  The top section rotates and houses a webcam with which Pietbot scans the room, looking for suitable compositions.  The lower section houses a small screen showing Pietbot’s image scanning.

The secondary part of the installation is a wall-mounted screen (e.g. 32” TV) mounted at eye-level in portrait format, with a small wall mounted box alongside.

 

Method

 

Pietbot looks for beauty by scanning every incoming video frame and testing for the presence of a brightness ratio of 1:1.618 (the Golden Ratio).

His method is to cut the picture into two parts either side of a horizontal line, then compare brightness between the upper and lower sections. His ideal of a beautiful composition is an image split at the point of the Golden Section, where the average brightness above the line is exactly 1.618 times greater than below it. By splitting each image many times at every horizontal line, he comes up with an optimum split-point to achieve his ideal brightness ratio in that frame.  Pietbot then weights each result according to its proximity to the ideal split point: the Golden Section.

[If at this stage the test shows that no close match to the Golden Ratio is present, Pietbot rejects the image and moves onto the next frame.]

This part of the process is to emulate the artist who observes scenes around him, attempting to spot a composition which contains naturally or chance-occurring beautiful ratios.

 

 

If the brightness test shows that an adequate approximation of the Golden Ratio is present using a particular split-point in the frame, Pietbot takes the two parts of the image and resizes them so that the split-point now falls exactly on the Golden Section.

This part of the process emulates ‘artistic licence’ taken with proportions of a subject in order to achieve a more aesthetic composition.

Splitting the original image with the Golden Section creates a square and a new Golden Rectangle. The process is repeated on this rectangle and twice again, following the pattern of the Golden Spiral.

[If Pietbot’s ratio tests fail at any stage during these repeat stages, the image is rejected.]

Upon successful testing and manipulation of the fourth rectangle, Pietbot ‘publishes’ his artwork to the gallery wall, signing it and updating an artist’s card containing an electromechanical counter to read “Composition No. XXXXX”.

 

A selection of automated artworks by Pietbot are on show 14 Feb – 13 Mar 2014 at the Hide & Sequenced exhibition in the Brighton Media Centre.

 

Muse Yourself is a composition tool in the form of an interactive digital art installation.  It loops a live camera and projector via a piece of code written in Processing language which allows the user to ‘burn’ their photographic image into the canvas by pressing against it.

An infinite number of photographic layers can be burnt into the canvas although older layers soon deteriorate.  This distortion is not artificial but due to feedback and loss of resolution as the equipment tries to re-sample and reproduce each successive layer. It’s the same kind of thing you see if you point a digital video camera at an LCD screen, or comparable to the way notes deteriorate when played through a digital delay pedal.

Above being just a fun painting tool, Muse Yourself plays with the distinctions between gallery visitor, subject, artist and artwork. By using the tool you do, to an extent, play all of those roles at once.

In the gallery, each finished image lasts only until someone decides to restart the process, beginning with a blank screen. However the program is easily able to store images for posterity; here are a few samples:

What?

The website www.flickratio.com is a data visualisation tool I built in 2012 to analyse the aspect ratios used by people uploading images to the internet.

It displays image data based on a user-defined keyword (just like in a regular image-search engine), but instead of displaying the images themselves in order of relevance, it plots each one as a line according to its aspect ratio.  So, ‘letterbox’ shaped images are represented on the far-left, panning through the Golden Ratio, 16:9 widescreen, landscape 4:3 until we reach square photographs in the middle.  Images to the right of centre are the portrait ratios, become increasingly tall until finally the narrowest vertical formats on the far right.  This tool uses the Flickr API data stream and Processing language to ‘process’ the results.

flickratio homepage, 2012

Why?

Social media allows us to do experiments which were previously near-impossible.  In this case we can take a sample of 500 people from across the English-speaking world (assuming we use an English keyword) and observe how they create media – in particular how they interact with aspect ratios.  We can see their decisions and ask questions about what led to them.  For example:

• Do people tend to pick similar aspect ratios when photographing similar subjects?
• How many or how few of the images uploaded to flickr have been cropped?  And how does that change according to different subjects?
• When people crop images, do they produce ratios close to those adopted outside the world of image-capture?
• Which ratios are popular at the moment?  Are the square images a product of Instagram and Lomography?  Is the iPad helping to preserve the 4:3 format used old televisions?
• Is there a relationship between the different aspect ratios we use in modern life?  Various ratios were made ‘standard’ over the years because it was felt they were aesthetically pleasing.  Given the mathematical root of ‘ratio-aesthetics’, can we see a pattern?

filckratio visualisation (using keyword: “example”), 2012

Can I see the pictures?

In the development version it was possible to ‘preview’ images by pointing your mouse cursor at the corresponding plot-line, like in the screenshot below; however it caused technical issues when live (crashing) so remains disabled for the moment.

filckratio beta visualisation (using keyword: “aeroplane”), 2012

 

These images use location data recorded by police crash investigators from 1999 to 2010. Each dot marks a fatality due to an accident on the roads of Brighton & Hove. Medium is photographic paper exposed directly against a monitor screen (sometimes referred to as ‘laptopography’).

 

‘Blackspots’, 2012 (41 x 51cm)

 

‘Deathstars’, 2012 (41 x 51cm)

 

These photographs were exhibited at the MADMA show, at the Phoenix Gallery in Brighton, on the 24-25 February 2012. You can view some footage of the show here.

 

These mondrianised websites, made with the Mondrian Composer Tool, show how a ‘classic’ html webpage follows Mondrian’s principle of rectangular divisions. Rollover each with your cursor to view the original (may initially take up to a second to load into cache).

 

1

2

3

4

5

6

7

8

9

10

Key: 1) Google; 2) Bing; 3) BBC News; 4) MSN; 5) Facebook; 6) Twitter; 7) Flickr; 8) YouTube; 9) Amazon; 10) Wikipedia

The screenshots on this page are shown as the sites appeared in February 2012.

 

This Mondrian Composer tool is based on the Neo-Plasticist or De Stijl paintings of Dutch abstract artist Piet Mondrian. It was created to explore the geometry that occurs within rectilinear html websites.

As well as black or white paint-fills, this composer offers the colour addition palette of De Stijl (red/ yellow/ blue), plus green to make up the colour subtraction palette (red/ green/ blue). To compose ‘webpage’ layouts, it is also possible to fill spaces with image or text.

Try it for yourself…

Click in the box to draw horizontal and vertical lines. New lines are naturally perpendicular, so clicking nearest to a vertical line will create a horizontal line – and vice versa.

Press C to clear the canvas and use the keys below to switch between lines and fills:

L – draw lines   X – black   W – white   R – red   Y – yellow   B – blue   G – green   T – text   I – image