Computational Art
In 2018 I started part-time MA programme in Computational Arts at Goldsmiths, University of London. Creative coding and the intersection where art and technology meet was always my personal interests. Below is a showcase of some of the projects I worked on.
Machine Learning Models experimentation
I wanted to learn techniques of using Machine Learning in visual design. My main interest was in training models on custom datasets. I hoped to learn different pipelines and training methods. My personal style inky marks made a good base for custom model that I used to train styleGANs via RunwayML interface.
I also wanted to experiment with typography. I created custom dataset of 3500 images and annotated it for pix2pix model. I then trained 3 variants of the model (tried and failed on Spell but eventually managed to train locally) and tested the results on linear images in custom written p5js app.
Technology used:
Models - Style transfer, pix2pix, styleGAN,
Training - Paperspace, Spell, RunwayML, local machine,
Testing UI: P5js, ML5
Data Characters
In this project for Theory and Research module I looked at data use as a creative material and its storytelling aspects. In addition to essay I also got dataset from Kaggle around pollution in London and used it to create characters artefacts. By making data more relatable I wanted to encourage questioning its 'origins and intentions' just as we would question newly met individual.
Technology used: openFrameworks (C++)
Composition Generator
I created this composition generator tool to augment my own creative process. I work a lot with accidental compositions, using inks and mark making. Here, I created several objects with random characteristics that I can re-arrange, again randomly, on the canvas and download the resulted compositions as vectors for further work in analogous media.
Technology used: Processing (JavaScript)
I used outputs of the tool to further create datasets and train StyleGAN on it.
As a final exploration in this project I took the generated graphic and used it as a starting point for a traditional painting. Closing the loop of this experiment and including technology in my existing process instead of replacing it all together.
Projection mapping
The light projection was an exploration of coding in openFrameworks. It was one of the earliest projects I made and the goal was to learn manipulating graphics with code. I created animated scenes and used a projection buffer to map it to Olga kit, triangle based form.
Technology used: openFrameworks (C++)
