Home       Overview       CV

Georg Eckmayr


OVERVIEW ︎︎︎
ARTWORK ++
    digitally signed ︎︎︎    SURFACES ︎︎︎
      Networked Sculptures 
  

    Machine Languages︎︎︎

    Digital experimental & algorithm based
 
    Other Work ︎︎︎
        Installations, films, sculpture, …
 


THEORY ++
COLLABORATION ++

Info / CV ++
  1. The Viennese Georg Eckmayr is an artist, researcher and lecturer.

    He studied digital art at the class of Peter Weibel under whose supervision he also completed and defended his doctoral thesis on digital images.

  2. Please contact me via one of these platforms below or the email on the CV page.

Mark

Research
Projects


Fields of Research


My research on digital culture is based on mediatheory approaches combined with artistic practices. Besides system theory, new materialsim and other theories of digital culture I use theories of visual studies, epistemology and philosophy of technology.

︎ RAW MATERIAL & DATA GESTURES
(Artistic & Habilitation Project)

non-conscious extraction of data, resource, labor via affective machines
 
WORK IN PROGRESS 

Do we run the machines or do the machines run us?

The artistic part of this project is currently adapted to large scale white-cube presentation. In my in progress habilitation project I focus on the distributed subject within digital power structures. This serves as the theoretical background for this project.

References: Haraway’s Cyborg Manifesto and Zuboff Surveillance Captialism help to reflect on the reading of The Structural Transformation of the Public Sphere by Habermas. 

︎ Untitled Me (Artistic Experiment)

Schema, Patterns, Meaning Creation, Generative Machines, Visual Analogy

Me and the statistical approximation of me 

Results of generative AI (Dalle2) recreating Portraits. This machine offers the recreation of individual portraits by statistical means. Doing this it focuses* on general distinctive features (like hair and clothes) while lacking individual features (like shape of the face, distances between mouth, nose and eyes). It recreates rather the look, than the face. 

(*This is true for portraits which are not highly present in the learning data set)


This experiment took place 2023 at the medienfrische festival in the alps, Bschlabs, Austria.

Leading questions: How is iconic meaning created? Further, what differences can be found between meaning creation based in statistics, human experience and communicative systems. 

Method: Reverse-engineering experiment aiming to find an approach to describe the gradual quality of visual analogy of results by generative AI systems.

References: From Boehm’s image anthropolgy to the posthuman patterns as described by Hayles.   

Hypothesis: Results show, that the generative machine (Dalle2) uses increasingly general distinctive visual features (hair, clothes, glasses, ...) to produce the effect of visual analogy while lacking individual features (shape of face, mouth, eyes, etc.). This seems especially applicable for portraits which are not highly represented in the learning data set. Portraits which are more likely to be represented in the learning data set (e.g. famous people) are recreated with more individual features. But how does the machine produce similarity without matching patterns? The experiment shows, that machine and human alike focus on general distinctive features when lacking detail.

Findings: 1) The people who took part in the experiment knew each other pesonally and had no problems at all identifying each other via their statistical abstractions. 2) Interestingly even the resemblance between siblings got reproduced while lacking their individual features. Both persons do not have round faces like that neither in the original pictures nor in real life. This finding should be researched in more detail.
The resemblance between siblings was reproduced in the results, even if the original features got lost. Both do not have round faces like that neither in the original pictures nor in real life.  

In a further experiment we focus on the structure at the core of the process of recognition. On a theoretical level we compare Boehm’s image-anthropological concept of Schema to the the term pattern at the core of machine-made sensing.

As Arcimboldo and many others already showed, it is the structural difference and not the representation of material, which triggers the humans process of recognising something. This aligns with Boehms concept of the Schema. We can conclude, that the machines patterns function in a similar way. In the following image “Is it me, or an misunderstanding” we see my portrait recreated via the representational regime of a landscape generator (Gaugan).

Is it me or an misunderstanding?

︎ HOMEOSTASIS – The Digital Ability to perceive Walls (Artistic Research Project)

Perception, Individual Behaviour, Systems, Coupling, Nonconscious, Machine Learning

Artificial Intelligence, Experiment, Animationfilm, Ink-Jet Prints / 2020
>> Creating a perceptive System & an attempt to simulate individuality


Background
The concept of identity is widely system agnostic. It is found in biological systems, various media systems and digital systems alike. The claim here is that human identity is an effect which uses all the before mentioned systems as medium. Neither the abstraction of human identity to data of behaviour fully grasps all relevant aspects of identity (e.g. self-perception, non-conscious imprinting, are  hardly represented in data etc.) nor do sociological, neurological and psychological approaches. To analyse and describe the phenomena identity we need a system-theoretic approach which focuses on identity as an effect between biological, social, and digital systems. 

References
We follow the approach Dirk Baecker proposed to define Intelligence as a result of the coupling of different systems in Intelligenz, kuenstlich und komplex. Further, Marie-Luise Angerer’s term of the nonconsious paves the way for a trans humanistic conception of identity, Nonconscious: On the Affective Synching of Mind and Machine]

Method
Creation and studying of an custom algorithm, which is able to perceive (or at least register) its surroundings and react according to it. Since identity is a highly complex phenomena, we focus on individual behaviour. An agency showing individual behaviour differs from other agencies and thus shows the effect of identity at least compared to its peers.   

Research questions
1) Can the appearance of individual behaviour be an effect of a simple set of rules for a system interacting with its surrounding? To tackle this question we developed a very basic artificial intelligence and studied its behaviour in different settings. Conclusio 1: The appearance of individuality can be created by means of interaction. Once a system with basic behavioural properties (perception, reaction) is defined, it appears to show individual behaviour if a) the relation to its surrounding is complex and the rules intransparent. And b) there is a body we attribute the behaviour to. We see that for the appearance of individual behaviour a very simple system of rules can be sufficient.  

2) What are the differences between human and digital sensing machines?
The human understanding of the world often uses projection of human characteristics (like intention, systemic autonomy) on non human entities. This renders the distributed characteristics of algorithms opaque and thus leads to miss interpretation of certain digital phenomena. In our example it is not the body of the shark which perceives the world, but a collision object is „sensing“ the walls and is controlling the sharks body. Further, the script which is at the core of this behaviour is stored neither in the sharks body nor the collision object. Hence the system is distributed in a sense, that the body, we attribute the behaviour to, is only a part of minor importance for its functioning. This is very different from human perception as the phenomenological concept of first person experience explains. Further to prove: digital intelligences and sensing machines are to a much higher degree distributed than the human equivalents.

3) What are the aesthetic differences between simulating individual behaviour and hard coding “individual” behaviour? Finally we compared the attempt to simulate individual behaviour via a perceptive algorithm to the method of key-framing. Conclusio 2: The former proves to be far more adaptive situation-wise, because we simulate social behaviour, while simulation via the imitation of body movement allows more detail in movement, but needs to be manually adapted in any new situation.

Presentation
This project was the foundation of the following course at the University Vienna: The Ability to Perceive Walls - Künstlerische Forschung und maschinelle Wahrnehmung

All the works from this project were released and minted to zeroone: https://zeroone.art/profile/georg_eckmayr

Part of this project lead to the Film Open Water which was presented at:

▤ 2024 Patterns of Disillusion / Galery 5020 / Salzburg, Austria
▤ 2023 Vienna Artweek exhibition / Vienna, Austria
▤ 2023 Short Waves Festival / Poznań, Poland
▤ 2022 Crossing Europe Film Festival / Linz, Austria
▤ 2022 Short Waves Film Festival / Poznań, Poland

VIEW OPEN WATER HERE

Screenvideo showing the artificial intelligence at work. It grants the shark in the pool the ability to perceive borders and react according to it. The shark is programmed in a way, that it recognises obstacles and turns if they appear ahead.
These are tracings of the behavior of the shark. Different individual patterns emerge from various ways to engage with the shark via purposely placed obstacles.

The Simulation of Individuality through Interactionvia Perceptive Algorithm.

The Simulation of Individuality through Imitationvia Key-Frame Animation.

Further material:

Escape the System via Perceptive Fails. Since, the body of the shark has no density, it passes through walls, if its perceptive system faills to react properly. This happens very rarely.


︎ Omnipresence & multiperspective (Archive Project)

Capturing the unforseeable.

We created an archive of audio visual fragments representing certain events* to find out about the relation of digital audiovisual representation and the event itself. The hyothesis is, that  the omnipresence of digital imaging devices like smart phones leads to a genre of press photography which captures even unforseeable events.

The archive contains material from various media sources which often offer contradicting or at least not aligning perspectives on the represented events. This material is widely used to illustrate my lectures. 

In the followng video we aligned the video-fragments of the explosions (first and sceond) of the bombing at the Boston Marathon 2013. This research was also the foundation of the application Die Multiperspektivitäts-Blockchain, netidee 2020,  Antrag ID: 5163, together with FH-Prof. Dipl.-Ing. Dr. Franz Fidler

( * The event was selcted on the availability of professional press photography and video as well as a large corpus of amatuer images.)

Recreation of the chronology of an terroristic act via audio visual fragments from the web.

FBI/Handout/Corbis//Getty Images
Terrorists Tamerlan Tsarnaev and Dzhokhar Tsarnaev shortly before they detonate the two bombs.


︎ Culture Signage (Artistic Research Project)

Comperative approach to the genesis of photographic composition in art history.

Artistic research on what is considered a “strong photographic image”: This project was partly used in my dissertation which had an rich appendix of a visual-discourse after p. 153. The findigs of this project were critical for the conception of the term “Contemplative Space“.

PDF: Link to my Dissertation