What I am sharing today is an update on the each 3 project I am pursuing.
Project 1 Physical data:
Brief:
Interpreting asynchronous signal extracted from raw EEG data into a physical computing project.
Process:
My instinct is telling me that there's something extremely interesting and poetic to extract electrical signal from the brain and .reinterpreting it as electrical current to activate a device here the options I explored:
As electricity:
Thanking it into the literal term and going from one brain area to the other, as seen in the slow motion simulation I made in Openframework. the data is interpreted as electrical arc where Tesla coils would be the brain areas and the connection between two coils as strong functional connectivity. This option offer the possibility to see the brain activity closer to its actual activity speed, but is very unsafe.
As motions:
These structure would reproduce the connection level for one area and between all others. Thus there would be 10 of those structure total to account for all the areas an EEG system is picking up. Activated by a stepper motor each ball position from the center is the data at a certain time. Very cool to see the motion of the full data set, also if we use led's for the ball we can then make long exposure photo thus presenting the full data set in a still image, but it is something that can be done easily in the virtual world and is space consuming and $$$.
As heat :
Using rapid thermocolour sheet activated by a nichrome wire. The nichrome is well known for its ability to heat up and is used in numerous household product like toaster, when combine , at low voltage with the thermocolour sheet, that reacts to heat, we can convert electricity into heat and then colour. Nichrome and thermocolour sheet a both super interesting materials to work with , but I can reproduce the same thing with a shader, and it can be a bit unsafe.
As fluid:
Ferrofluid was originally created by NASA to move fuel into space, as it is easy to control with magnets (as show with the small sample I have). In this case the goal would be to translate the data into magnetic attraction. Using electromagnets the data can now activate the fluid to move in unexpected way to the rhythm of brain wave. I am currently investing more into this kind of option due to its creative potential as I am following this tutorial to experiment a bit with electromagnets. Visually stunning, cost effective (since I can make my own electromagnet), easy to activate (just need to pass electricity into the electromagnet to create attraction) can interpret the data as a whole by moving the shape, but ferrofluid is expensive and messy.
As magnetic sculptures:
Still using electromagnet but this time with metal powder. As seen in Roman De Giuli video once under the influence of a magnet the powder will create those intricate and complex structure. Furthermore, if you activate it in two different position they will follow the magnetic field created by the electromagnets thus offering an opportunity to, perhaps, see connection between brain areas and electrodes. And lastly the result may vary depending on the metal used, in Roman case he is using soft iron, a metal that does not maintain magnetism but if one where to use
steel powder can keep some of that magnetism, thus we could also account for the over time effect of the data. Lastly, there might be a way to solidify the structure, maybe with resin, to make a still artefact. Luckily, for me there's endless steel powder in the Goldsmiths metal shop.
Conclusion:
Using the knowledge I will learn from following the ferrofluid project I would like to expand the project to steel powder with the hope of finding a way to turn it into a permanent sculpture. By December I would like to present both the ferrofluid Instructable project as well as my rework of the project with steel powder. For the exhibition I could present the device working accompanied by some sculptures.
Project 2 EEG, Slime Mould and FMRI:
Brief:
Following on last year projects brain mapping 1 and brain mapping 2 I would like to push forward the work into a proper 3D visualisation of the slime mould functional connectivity algorithm inside an fMRI scan of the brain.
Process:
In a way I am starting from scratch to account for one of the biggest problem I had when working with this simulation, and it is the limit of particle I can animate. Thus, I want to start on the right foot by utilising the GPU.
Thus I started reading "The Book Of Shaders" and I am currently experimenting with exponential functions, to have a grasp on GPU coding.
The next step will be to review my 3D slime mould hopefully with Andy Lomas expertise and help.
Conclusion:
By December I would like to have a good understanding of shaders and able to offer an early data visualisation piece on a screen. And later it could be presented as a G05 performance inspired by the work of Ryoji Ikeda. Overall this project is crucial for me to level up into computing project if I want my project to shine with complexity.
Project 3 Data visualisation for Company's Transparency:
Brief:
I am developing visual tool that allows consumers to assess the environmental, social and economic credentials of a company by placing the company's logo over your phone or by using the web extension for online purchasing. With the help of machine learning the consumer will be able to check, not only who owns the company, but also support them to make a decision about purchasing from the company based on how the company responds to environmental, social, and economic issues.
Process:
I am currently enrolled, as an optional module, in Jamie Forth awesome class on data visualisation. This class and Jamie experience are a huge help for this project. As I have the opportunity to design and collect data on my first survey that aim to know how relevant the project is and if the visual convey the informations properly.
The survey close on Friday the 5, leaving me the reading week to analyse the data and produce a report. The next step will be to improve on the data. As it is, I am using raw unclassified text from the web to train the model, thus, I need to improve the training data first, perhaps by training it on well known journals like the Guardians, the Sun ext... Next the visual needs to be updated to better convey the data gathered. During a talk with Jamie he pointed out that the articles sentiment analysis are biased to promote the positive. We cam up with the idea of separating each section in the middle with a line that separate positive and negative articles. Then further down the line I would like the user to be able to review the texts that made the visuals. Thus a zoom-in option will be added, as you can get the texts that made the arc you are zooming-in. Then it is just a question to fix my logo recognition AI. As a final piece it would be present as a web page (then web app) that you can use to scan some object, clothing, phone, food and more where the visualisation appear to inform you of the product you are about to buy. For December I could present printed pictures that express the result of my rapport and the accuracy of the AI with perhaps a couple of design to look at where I could survey the class.