More selected projects

Cannibal Manifesto

Cannibal manifesto is based upon the text of the same name by by Oswald de Andrade (Manifesto Antropófago):

'Against social reality, dressed and oppressive, defined by Freud - in reality we are complex, we are crazy, we are prostitutes and without prisons of the Pindorama matriarchy'

Artist: Daniel Evans


This is a projection mapped piece, in which five different parts of a larger generative image are divided and split up on to the faces of five picture frames. It is inspired by the concept of the cannibal, one vilified and yet also abundantly observed in the natural world, the notion of eating ones own kind is saturated in ideas of self similarity, liminality and taboo. I wanted to bring together many different mediums into a singular piece of work. The source materials span across history - I include images/paintings, words and website entries. These are all brought together and mediated by the digital, as the singular picture seems to travel across the blank space between the picture frames as the images devour themselves.

Concept and background research

This piece was not necessarily inspired by other computational artists, in fact it aims to move away from an overly digital or geometric aesthetic. I wanted to utilise the picture frames not only for their appearance, but also for the semantic weight attached to them. This is why I chose to use paintings as my first scene, in order to emphasis not only the picture frame mapping but also to subvert the painting form with a layer of digital mediation, as each painting continously over and undercuts the other, devouring them in a way. The two paintings are themselves very famous historical depictions of cannibalism. The first is one of Goya's black painting, 'Saturn Devouring His Son' and the second is Peter Paul Rubens' depiction of the same myth called 'Saturn'. I love how the pictures seem at home within the frames, but then they are fragmented and broken up by the digital modulation.  Each scene is cut up by a strobing effect that is not a smooth transition but instead acts to disorientate and produce perceptual artefacts. I also utilised a colonial etching of cannibalism which you can barely see in amongst the red strobes. I included this not to ignore the impact which that era of history has upon perceptions of the 'cannibal other'. This project is on one hand aesthetic, but on another archival.

I also include lots of text in my piece. The first is scraped text from the Wikipedia article on cannibalism. This was, again, to emphasise the archival quality of the piece. It attempts to collate source data from a whole host of different times and mediums. The programme shows you the wikipedia text in a non-linear manner, it slowly moves throughout the text but each frame depicts a jumble of words in no discernable order, making it particularly hard (but not impossible) to read. This adds to the disorienting quality which I aimed for in the work. After another set of strobing the piece finally finishes with the 'Cannibal Manifesto' itself. Behind it we see big droplets of blood red moving in and around the frames. This is, of course, a reference to liquid blood but it also reinforces the way in which this piece avoids hard angles and corners. I wanted a fluid and organic approach to shape and form, not a angular or overly 'geometric' one.

I am proud of the strong thematic groundwork which is present in this piece, and feel like my usage of text and historical paintings allows it to be a bricolage or collage more than it is a generative geometric artwork. However the generative element of the piece, as it does indeed look different every time, acts a mediator that gels together the source material into a singular experience.


This was programmed in OpenFrameworks, a comprehensive C++ library for artists and designers. Although I used a whole host of different mechanisms to manipulate my source data the main thing I utilised was perlin noise. I created a two-dimensional perlin noise field and then established a conditional boundary at around 0.5 (half way between the min and max values perlin noise outputs). In the first sketch the two different images were mapped to either side of this moving perlin noise field as it moved along the graph and also 'zoomed out' in its scope. This meant that the two became broken up and interwoven with one another, different parts of the painting revealed as the programme generated new perlin data. For the first scene I also had the conditional boundary move apart from each other slightly over time, thus producing a kind of glitchy 'no mans land' in between the two paintings as time went on. For the final scene I also utilised this 2D perlin noise field, however I simply split the two halves into two colours - red and black. This field was also very 'zoomed in' and so the colours would really highlight the geometry and shape of the picture frames where colours would come in and out of view, splitting and contorting as it went. Not only did this look great, but it also highlighted the non-space between the frames. I think this worked very well, and was potentially my favourite part of the whole project. I am interested in seeing how simple mechanisms can produce exciting results and I think this is a good example of that. I also used a lot of string manipulation, it was great to learn how I could scrape and split up a whole host of string data into individual words, and then generatively (and non-linearly) present that to the viewer.

Future development

Although I am not necessarily interested in doing something similair to this in future, I learnt a lot from the process. One thing it did make me want to develop was a particular and specific approach to computational arts that I think I enjoy. I liked working with the string datasets, and being able to manipulate and play with these. In my general practise I am not interested in visual works but sound pieces. In this vein I am often looking for ways to control my sound pieces, to pick root frequencies or trigger/modulate events. As such I would quite like to work on having a more simplistic and immersive front end in the future - the thing that the viewer/participant actually percieves. On the back end however I would like to use some heavier, data driven processes (in Python or other similair languages) to control and manipulate the perceptive experience by scraping or taking data from online sources around the world. The exact conceptual meat has yet to come into my head, but just an idea that came from the project !

Self evaluation

I definitely think I should have made the project a little neater. I could have thought through my fonts and text a little better, thinking more about the visual appeal of the typographic elements. I also wish I could have spent more time optimising the perlin noise field. Particularly when grabbing the pixels from the images and mixing them together; it could be a little slow with a compromised framerate. I would like to have tried more alternatives in the code to optimise performance. I did however achieve what I set out to achieve, my first inspiration was the idea of an image fuser, something which chimerized two images into one. I think the results were quite appealing and it also taught me a great deal about how to use perlin noise, something I could use for procedural generation in the future. I am also happy with the way I utilised the picture frames, as I was one of the few people in my group who tried to have images travel between different frames. I also learnt that I am not that interested in projection mapping, and probably will not use it for any of my projects in future. I am looking forward to going out of my comfort zone even more with the next project and exploring more complex algorithmic forms.


Saturn Devouring His Son - Francisco Goya []

Saturn - Peter Paul Rubens []

Wikipedia Article on Cannibalism []

The Manifesto Antropófago (Cannibal Manifesto) - Oswald de Andrade []

Engraving of Cannibalism - Theodor de Bry []