Artist Projects

Database Divination: Classifying The Machinic Gaze Part 1


I was pretty pleased with the video and image results acquired from yesterdays webcam experiments to continue proceeding down this path.

 

However said path was marred by my creaking Lattitude laptop with Ubuntu Interpid Ibex installed. It proved incredibly hard to get software installed on it, however I did manage to improve the contrast of the images snapped yesterday. By increasing the contrast the pattern is given greater relief, something that will be easy to automate via gimp or image magick.

 

Adjusted Images which will help computer detect patterns more readily. Final picture remains unadjusted to illustrate the importance of calibrating webcam for consistent white balancing

The ever generous Derek Shaw of SoSlug offered to oversee the upgrading of the Ubuntu software, and in the meantime I cracked ahead with using Graham Harwoods laptop to do the various image software investigations. For a while I got quite involved in finding ready made ways of analysing the motion of the tea leaves as I had detected in video recorded last night. This process has been documented here and it did offer up some nice stills of tea reading from a different perspective. I’ve collated a series of frame grabs of the video from last night (a selection of which I have added here), which have their own bewitching patterns contained within and perhaps offer a totally parallel site for exploring the different patterns seen by computer and human.

However I found that the myriad of options available to catalogue patterns within moving video (and the associated complexity attending each method, including this which at first glance appeared to be relatively pain free in terms of implementation) was taking me too far in one direction. I wanted to return to conceptualising the process of distilling the overall practice of tea leaf reading into a database. Ergo a second bout of tea leaf reading was required.

 

 

This time around I was more concerned with engaging with the practice itself rather than proving that the webcam apparatus would hold up. To that end I dove into “Telling Fortunes By Tea Leaves”, a canonical text in tasseomancy dating from 1922. This provided some useful background information, some of which buttressed areas which I was keen to interrogate and some of which make me think again about the suitability of tasseomancy as a scrying practice.

The rhetoric contained within the book follows an established trope in occult & magickal practices, the idea that the subconscious must assert itself (‘self’ is hardly ideal to use here but it will have to suffice) at some point within the ritual. Within tea leaf reading this opportunity for subconsciousness to bubble to the surface occurs when the tea cup, with a small amount of tea remaining in it, is swilled by the left hand and quickly inverted. During this moment the mind must be either completely absent or intensely focused, again this correlates with what previous research into Chaos Magicks ‘Sigil’ projection. Interestingly Cecily Kent (the author) notes that:

“time has no meaning for the subconscious self”

This is interesting for me as I am interested in using the video capture length as data for the computer to add to its database table record of the tea leaf reading experience.

The person gazing at the tea leaves is invited to discern shapes and letters. However it is at this point that tasseomancy differs from the sorts of scrying practices I was initially attracted to: there is a rigid symbolism associated with the symbols detected, in a manner not too dissimilar from other means of fortune telling. This is problematic in that the divination practice suddenly become tied to a fixed symbolic register external to the experience of the practitioner. However it is interesting as the idea of the patterns being codified per some external authority is very pertinent to the broader conception of code, (code as embedded intelligence such as postcodes, code as ordering structures and protocols external to computers) that we have developed within this lab.

 

Note: Sigil as used herein is not to be confused with the use of Sigil in Perl. I only learnt about this crossover in the context of this lab, but it represents another interesting transfer between the occult and code

1 Comment

Databases & Divination

My initial project concept of exploring glitch between bodies of code (i.e. vocal recognition software) has been postponed to a day when the software gods are smiling upon me.

However the other area which I wished to explore was the concept of juxtaposing practices of divination with the algorithmic action of data mining.

I feel that pattern detection is a massive part of digital culture, especially ubiquitous computing. I am really interested in interrogating the idea of prediction that underpins the faith in data mining as a worthwhile endeavour. I am interested in this movement at all scales (for instance I consider it appropriate to deploy within science but nevertheless find the refiguring of science (it is now figured as a fourth paradigm of scientific enquiry) occurring therein politically interesting (from a science studies point of view). The area which interests me most is the application of data mining to social contexts, the notion that a form of knowledge alchemy can occur through weight of data aggregation and data mining algorithm execution.

 

So several areas therein interest me. One is the respective distillations of experience that occurs in divination and data mining. Non excitatory (and non substance – induced) divination tends to have an object of focus and there is a communion between diviner and the object. Per an Object Oriented philosophical approach I would be sympathetic to conceiving of a relational agency occurring in divination practice. In divination the distillation of experience is difficult to locate exactly, it is fluid. Data mining is reliant upon fast computers and database structures. But it is at the data entry stage of a data mining process that the distillation of experience occurs. In this regard I am very interested in the processes of classification highlighted by Leigh and Bowker Starr. I am also interested in what categorising an experience in this manner does relative to the intuitive process that characterises other divinatory practices.

I aim to explore this through various means of scrying: initially I drifted towards forms of lecanomancy (water gazing, using oil or ink drops as a focal point). However I have drifted towards focussing on tea reading, or tasseography (or tasseomancy), as the preferred practice.

The concept behind the project works as follows. I want a user to engage in the practice of tea leaf reading. They will follow the procedures of this process, and upon completion will be encouraged to enter their experience into a command line prompt. All the data entered by the user will be stored in a database.

The tasseomancy process will also be observed by the computer via webcam. Upon completion of the task the computer will take a picture of patterns of tea leaves in the tea cup. It will log the data as relevant to it’s processing of the pattern in the same database.

I hope to gradually accrue records in the database wherein the process of human intuition, machinic vision pattern detection and pattern detecting database algorithms will all interact in the one assemblage.

Leave a comment

Swan Dreams

1 Comment

Digitally Detecting Tea Leaves

I have done some experiments to explore means by which I can explore tasseomancy as a site for critiquing the pattern seeking prerogative of knowledge discovery in database (KDD) algorithms by juxtaposing divination alongside them.

The eventual aim is to explore the distillation of experience that occurs exclusive to each system, as I mentioned in my earlier post. This means some experiments into how a database and appropriate tables can be constructed to capture every aspect of the tea seers process and practice. In some ways I am setting this concept up to fail as I believe that something of this process will escape the intense categorisation which is a prerequisite of relational database architectures and their need for normalisation. In so doing maybe this will prompt reflection on how every database system, irrespective of their predictive prowess, is ultimately a modelling that holds more true to set theory logic than the nuances of our lived experiences (and said nuances being elided is not trivial given the faith and belief placed in the predictive power of data mining).

In order to fully explore KDD algorithms I will need a suitably large amount of data. That is unavailable to me at present so I need another area of interaction between the tea reader and the machine with which they will interface. A visual analysis of the tea leaf pattern left in the cup provides a ready pattern for the computer to interpret.

I have settled upon using a webcam and I’m currently exploring two ways of providing the computer a chance to see the tea leaf pattern. One method comes via a snapshot being taken after the tea pattern has been scrutinised by the tasseomancer.

The above photos are rough webcam snaps, exposure modifications will display pattern but it is important that exposure process can be simply automated by code

 

Another comes via the whole process of tea drinking being recorded via a webcam positioned at the base of the container from which the tea is drank. Doing this necessitated using a glass, which according to most tasseomancy practice is not ideal. However it does provide an interesting perspective on tea leaf reading. (see the video below, which is quite poor quality due to the fact that I am getting to grips with the exposure functions of Linux UVC webcam software)

Webcam Rig used to capture above video

Capturing the duration of the entire tea drinking process is something I would like to do, and also a period of time I would like to analyse. I am struck by the idea that duration is an important aspect of any of these introspective, gnostic (in the Peter J. Carroll sense) practices. This contrasts markedly with the types of temporality at work in the CPU crunching the numbers which determine the patterns discerned by the image analysis algorithms (silicon flip flops and clock times).

Once the software crimps have been ironed out and I get a clearer idea of the registers by which the machine will pattern analyse the tea leaf patterns I hope to engage with the tasseomancy practice in depth. In essence I will be the entity responsible for determining the data architecture which at some future data KDD algorithms will explore and generate new knowledge (defined within KDD discourse as “novel patterns of information – an epistemic shift in what we deem knowledge).

Leave a comment

CCTV in Operation on these Premises

Yesterday evenings theft of jackets from the inside of the doorway* of Chalkwell Hall re-kindled interests I have in CCTV.  See below Chalkwell Hall’s sophisticated Media Management System:

The simplistic rhetoric surrounding widespread usage of CCTV in the UK (coming both from those who would advocate the use of CCTV and those who would oppose it) had already been on my mind, and I think this is because several of the technological presentations over the last few days have echoed dystopian visions of panopticon future:

e.g.
- flowcharts, system and control diagrams for user (meaning person) management
- massive multimedia education/propaganda systems for the adjustment of attitudes
- indoor and outdoor locative systems for the tracking and monitoring of persons

The fact that this house is equipped with such a high-tech system of cameras and monitors and yet does not use any CCTV for the purpose of crime prevention struck me as particularly ironic following the thefts, so I decided I would have a go at using Chalkwell Hall’s much celebrated Media Management System to make a simple closed circuit television monitor system (like those used in pubs and clubs) which would be able to watch the front door:

There are several reasons why I wanted to do this but a simple motivation was to get a feel for how this media management system works (or doesn’t work) and to see exactly how accessible the system is “with the wires out”.

I decided to reposition one of the cameras and, whilst playing with it I found that it had a face recognition system onboard which was, in my experience, fairly sophisticated for an off-the-shelf domestic appliance.  The camera setup could be programmed to recognise individual faces by storing a photograph and then comparing biometric measurements in live footage.**

I tried out the face recognition with a few different people and it was successful at distinguishing and recognising their faces.  I would have liked to have programmed in all of the inhabitants of the house for the week and in this way built a system for distinguishing invited guests from intruders but the camera (Panasonic HDC-TM700) can only store reference images for 3 different faces.  I looked at hacking the camera and expanding for this capability (eg by adapting the firmware or using an API) but after a little research this seemed like a heavy approach to making this quite light sketch.

I decided instead to repurpose this unmodified consumer device for use as a kind of surveillance/screening system which would be able to search for and compare three pre-assigned faces.

The three people I chose to introduce to the system are facilitator Graham Harwood and our two esteemed guests due tomorrow – Jude Kelly and Lewis Biggs.  Portrait images of each of them have been taken from the internet and uploaded to the camera (without their consent) and now each time someone comes through the door their face is compared to a portrait of Graham*** Jude and Lewis.

(Below) Reference files stored in the camera on the three individuals:

Graham Harwood Image:
http://www.flickr.com/photos/24722640@N02/2692374244/

Jude Kelly Image:
http://www.womenspeakers.co.uk/speakerdetail.asp?speakerid=85

Lewis Biggs Image:
http://www.bbc.co.uk/radio3/freethinking/2008/free-thought/lewis-biggs.shtml

* thankfully the jackets were recovered the following day

** I struggled to understand what practical purpose this face recognition system would have for the target demographic of the camera (families etc.) although, Caroline pointed out that the tagging of photos has become a commonplace procedure on Facebook and could be seen within a wider trend of adding different kinds of meta-data to images.

***Later this afternoon the system was unable to identify Graham using the photo held on file (perhaps because he sports glasses and a versatile beard). When Jude and Lewis arrive tomorrow I hope the system will recognise them!

Guidelines for CCTV: http://www.ico.gov.uk/upload/documents/cctv_code_of_practice_html/9_responsibilities.html

Leave a comment

Sun Gazing and GSR

I just went and did a quick experiment. I went out and stared at the sun for half an hour until it set behind the horizon. I was recording my galvanic skin response and drawing a graph with processing.

The measurements are very scattered, picking up and down… There is a quieter pattern in the middle, when I was on the phone, and it goes disperse again towards the end when I go back to the staring. Then came in a took some more readings. Pattern is much more regular.

It is impossible and definitely not my intention to make assumptions regarding the meaning of these. A human body is a very complex system and its oscillations are not direct effects of clear causes but rather aggregations of hundreds of quasi-causes. Yet how cool would it be to see our interactions with other bodies expressing at the level of bodily potentials, leaving traces in the form of microchanges.

Leave a comment

Another Digital Surface

The huge amount of input during the first days of this Media Lab made something very clear. The digital (art) world desperately needs new ways of presenting digital images. Digital screens are outrageously limiting.

Does anybody know of any type of screen that is enjoyable to look at? Even the most expensive designed digital screens on the market are horrible things to look at. This might be a subjective matter, but there is more. (more…)

Leave a comment

Infraverbal communication between bodies of radically different nature

These days at Metal I’m trying to explore the idea of Infraverbal Communication among bodies of radically different nature. I am very interested in how bodies communicate outside language and symbols. I am interested in all those signals that connect bodies that lie beneath consciousness, outside the discursive space, beyond rationality: deep media.

I am interested in the idea of modes of perception that exceed the realm of the five senses. Through exploring other modes of connection we can begin to understand the level of entanglement among bodies (human and nonhuman). At present I am working on a relationship between three very precise bodies:

- The Sun
- A Human Body
- A Machine (a media system)

What I want to do to start with is to build small prototypes that provide the human body with a new sense that allows the Sun to affect it in a different way. In doing so my aim is to see whether infraverbal communication can be expanded, opened up, allowed to flourish. Further than that there can’t be a particular aim since it will be the communication itself that will have to take us to the next step. I don’t want to enter this experiment with an idea of what that communication needs to do, I’ll just work at the level of the architecture of the channels of communication.

Leave a comment

Measuring Light Wavelength

TEST: How can we sense light on a particular range of the visible spectrum?
Certainly one of the most accurate sensors for such a thing is the human eye. Equipped with highly refined photosensitive cells the eye is capable of discerning between frequencies which are just a few nanometers apart. That’s why we say we have colour vision.

One of the things I am interested in though is in being able to detect the presence of blue light in an environment. In the last years a lot of work has been put into finding the photoreceptors in the retina that mediate non-visual responses to light. Regulation of circadian rhythms, mood alteration, concentration capacity, all fall under the label of “non-visual responses to light”. It seems like this could be an interesting entry point for our process of experimentation.

The cells that are held responsible for such a task are Intrinsically Photosensitive Retinal Ganglion Cells and the photoreceptor with which they carry that task is Melanopsin. Apparently Melanopsin is sensitive to a light frequency of approximetely 480 nanometers. 480nm correspond to light in the blue range of the visible spectrum as shown in the diagram below.

Visible Spectrum

If these cells can be responsible for such a broad range of “responses” they are probably a good starting point to look into new modes of communication with the sun.

What I am trying to do is to use an LDR to sense blue light. But LDR have a spectral sensitivity very similar to the human eye (since they are build to respond to “visible light”). This means they are more sensitive to frequencies in the middle range of the visible spectrum and less sensitive towards the fringes of it.

All this requires some calculation and experimentation. So the first thing I did was to set up an LDR inside a box and light up a superbright led in front of it. I did this with a blue, green, red and yellow LEDs. Then I repeated the experiment with a blue filter covering the LDR and wrote down the measurements again. The results are:

BLUE LED WITHOUT FILTER: 400 ———— WITH FILTER: 151/180
GREEN LED WITHOUT FILTER: 720 ———— WITH FILTER: 15/20
YELLOW LED WITHOUT FILTER: 710 ———— WITH FILTER: 5
RED LED WITHOUT FILTER: 720/780 ———— WITH FILTER: 5/6

So, blue light gives us lower values from the start (which makes sense if we take into account the LDR’s spectrum sensitivity. And also blue light is the one that is decreased in a smaller percentage when passed through a blue filter.

Percentage of light intensity reduction with blue filter depending on colour light:
Blue light was reduced in: 62.5% – 55%
Green light: 98%
Red light: 91% – 99%
Yellow light: 99%

So clearly the light that was less reduced through the blue filter is blue light as we could expect. Now the second thing we tried is to set up two LDRs and read their values. They were both giving similar values (although never exact). See values below.

Then we covered one of them with the blue filter and tried moving it around different sources of light. This is the setup:

Circuits of LDRs with and without filters

I plugged it into processing to be able to quickly analyse the result as I was moving around the room with my computer and the arduino plugged in. The red line in the graph corresponds to the LDR without filter and the blue light corresponds to the LDR with the blue filter (and therefore more sensitive to blue light). What the graph shows is that the differences between indoors and outdoors are more extreme for the blue light sensor.

Leave a comment

Electromagnetical Attunement: Initial Project Ideas

For some of my investigations I was really quite eager to explore the above experiment in the telephone game (or Chinese Whispers as it is known on this side of the pond, for reasons I am still not clear about). My twist on the experiment was to be training the computer’s speech recognition to listen to electro-magnetic vibrations via magnetic pick up mics.

However things have stalled at the first, and crucial, software hurdle. I was attempting to install speech recognition software on ubuntu, which does exist in the form of Julius and Vox-forge. However synaptic package manager was somewhat hamstrung and though I have managed to download the packages from here I am wary about getting sucked into the Linux terminal line (i.e. via DPKG) installation vortex (I suspect my grasping knowledge of code installation hasn’t installed everything from the ubuntu archive).

The hardware end of the project would be relatively easy to cobble together, a quick visit to a music store would suffice. However I really would like to build this electrical stethoscope as described here. I think this is interesting for two reasons, one being my recently kindled interest in biodynamic psychotherapy, and two because this method of attuning to computer hardware is intimately linked to earlier audio debugging methods

Leave a comment