In this work I am revisiting my 2018 creation "Neural Glitch 1541266710"
by making a state-of-the-art Stable Diffusion model attempt to reconstruct
my earlier work. These new prompt-driven models operate very differently to
the way GANs used to work. Foremost the model itself was not trained by
me, but is what I call a "public latent space" - it has been trained on
hundreds of millions of images which gives it almost universally capabilities
to produce and reproduce any image.

In a two-step process I first use an algorithmic search in order to discover
a text prompt which is able to capture the semantic and aesthetic essence of the
original work. In the second step the Stable Diffusion model is given this text
prompt and is seeded with the original image which results in this work.

Whilst the outcome is clearly showing more detail and also captured some of the
original work's meaning one could also say that it failed at the task since
it was not able to reproduce the convolutional aesthetic of the source. One
possible explanation for why this is so might be that whilst these new models have
been trained on almost all kinds of imagery it appears that they have not
been exposed to any early AI art.

Error Correction

Mario Klingemann

Pinuptic Curves #2

First NFT to depict a sculpture of the word "FIRST"

The F-Word

Persistencia de la narración, 1. 2023
Unique work.

"Persistencia de la narración" series is an experiment that seeks to reflect on the gap posed by the introduction of artificial intelligence in artistic production processes.

The panels generated by artificial intelligence, along with the text assisted by ChatGPT, address our most fundamental need: the search for meaning. In a world saturated with images and with a demand for immediacy, "Persistencia de la narración" questions the very nature of contemporary digital images. Artificiality and discontinuity contrast with human perception, which relies on concatenation to ensure coherence in the narrative.

"Persistencia de la narración, 1" (2023) uses Duchamp's words as a starting point to reflect on the interaction between technology and art. In its critique, it questions the symbiosis between ars and techné, as well as notions of falsehood or capitalization. "Art is a lie that brings us closer to the truth," expresses the work.

This NFT has been printed in an expository version (not for sale) with dimensions of 179.8 x 124.6 cm. It is being displayed as part of Protection No Longer Assured, the new exhibition by Colección SOLO, showing in 2023 at Espacio SOLO Madrid, which looks to generate conversation on those things we find attractive yet at the same time, terrifying.  https://coleccionsolo.com/projects/protection_no_longer_assured/

A 16120 x 22616 resolution JPEG version will be provided to the collector.

Persistencia de la narración, 1

This work is part of a series of GAN generated images that I created as a contribution for Nick Knight's "Fashion Digital Future" - first shown at Harrods, London in October 2020.

In an iterative process I call "Transhancement" I reinterpret catwalk images from 20 years of fashion shows. For this I trained a custom neural network on "The Garden of Earthly Delights" by 16th century painter Hieronymus Bosch. As a result it deliberately misinterprets the data it is given and produces forms and textures that are undoubtedly reminiscent of his aesthetic whilst at the same time introduces surreal shapes and strange compositions.

For this series I developed my own custom GAN architecture and training process which attempts to generate a rich variety of textural details whilst at the same time tries not to imitate brush strokes but maintain its own "convolutional" aesthetic.

Neural Delights #1

In 2018 I began experimenting with deliberately mutating and blending the learned weights inside of neural networks I had previously trained. 

I call this technique of latent lobotomy "Neural Glitch". Unlike the classic technique of glitching a single image by manipulating the bytes of a saved file this method causes the damaged model to continuously generate misinterpretations on a textural as well as a semantic level.

PNG, 2048x2048 pixels
Original creation date: August 28th 2018

Neural Glitch Portrait #1

This work is part of a series of GAN-generated
and transhanced portraits I made in 2018.

Portrait 1535445164

Quasimondo Collection

From my "Collagenerator" series. Originally created in 2019

Longshot #1

Quasimondo

This work is part of a series of GAN generated images that I created as a contribution for Nick Knight's "Fashion Digital Future" - first shown at Harrods, London in October 2020. In an iterative process I call "Transhancement" I reinterpret catwalk images from 20 years of fashion shows. For this I trained a custom neural network on "The Garden of Earthly Delights" by 16th century painter Hieronymus Bosch. As a result it deliberately misinterprets the data it is given and produces forms and textures that are undoubtedly reminiscent of his aesthetic whilst at the same time introduces surreal shapes and strange compositions. For this series I developed my own custom GAN architecture and training process which attempts to generate a rich variety of textural details whilst at the same time tries not to imitate brush strokes but maintain its own "convolutional" aesthetic.

Neural Delights #4

SuperRare

Foundation

In 2018 I created a series of experiments titled "Neural Glitch" in which I deliberately manipulated the inner weights of custom trained GANs by permuting and altering them.

As a result these models that had originally been tasked to enhance the inputs they were given started to misinterpret those inputs both on an aesthetic and a semantic level.

The artificially introduced errors which emphasise the convolutionalnature of these early models give us a glimpse into the inner workings of these complex networks and at the same time result in an aesthetic which is almost unique to a certain short phase in the rapid development of AI art.


Neural Glitch 1541266710

Mario Klingemann's works for "The Perfect Error" curated by Luba Elliott for Unit London, March 15th, 2023

Mario Klingemann x Unit London

Documentation of a sculpture created by the German artist, Mario Klingemann. Titled *Like This*, the sculpture was an interactive object that uses an Arduino board to count likes and store the total number of presses. It was on display at the Ultra Social exhibition at the Urban Arts and Media Organisation (UAMO) Festival 2011, which took place in Munich from October 27-30, 2011.

Like This #7

Like This #6

Like This by Mario Klingemann

Made with an evolutionary process that tries to maximize its feature vector distance to all other 8x8 images in the current system. It does not aim to be beautiful or interesting, just different.

Novelty Maximizer #044

Novelty Maximizer #023

8BIDOU

5/7/5

8scribo

Snapshot of my sketching tool "Sketchflix" in which a generative algorithm indefinitely sketches and erases the same canvas thereby creating a continuously changing collage from found and reinterpreted imagery. Snapshot taken on May 9th 2020. Size 1920x1080 pixels. Artist: Mario Klingemann / @Quasimondo

Sketchflix #5297

Live recording of a GAN generated face performance | Mario Klingemann / @quasimondo | 1:46 minutes video 1024x1024 pixels | Background music: licensed stock audio. | Creation date: April 11th 2021

The Follower

Hic et Nunc

AI generated painting based on a text prompt | Medium: PNG 2048x1024 pixels | Original creation date: February 18th 2022 | Artist: Mario Klingemann @Quasimondo

All proceeds from the sale of this work - including all future royalties - are automatically donated to the Marine Megafauna Foundation https://marinemegafauna.org / @MarineMegafauna (Tezos wallet: tz1Vy6W8uh6DcfZZFCtx6QJZ3GvwSMZDwJye)

Plastic Trash in a Storm at Sea

Herbert W. Franke's book "Phänomen Kunst - Die Grundlagen der kybernetischen Ästhetik" (1974) has been most influential in my thinking about art and the ways we perceive it. To my knowledge it was never translated in English. In this book Franke explains very concisely "how art works" based on Claude Shannon's information theory and Norbert Wiener's "Cybernetics".

My tribute is based on the introduction to chapter 16 of the book: "Subjective Information - A prerequisite for a favorable coding is knowledge of the frequency relationships applicable to the signs. If these have not yet been determined, the subjectively conveyed information is greater and the redundancy smaller. Meaning that both are relative quantities. By learning the inner relationships of a structure one reduces the information obtainable from it." 

This work is a real-time view into a functional operating system in which multiple agents use a limited set of assembler instructions that change the state of the system itself. Unlike its appearance there are no random number generators used inside the code, the entire behavior of the system is deterministic apart from the starting conditions which get slightly altered upon every restart. Anything that appears "random" to us is just a result of the complex interplay of the agents that share the same memory space which allows them to alter their initial instructions leading to unpredictable behavior that maintains a promise of potentially new information even after we believe that we have understood the inner workings and recurring patterns.

Released on the occasion of the Tribute to Herbert W. Franke. 50% of the sales will be donated to the Foundation Herbert W. Franke which is dedicated to preserve and document his work.

Subjektive Information

Versum





When the moon 
hits your eye 
like a big pizza pie
That's a








When the moon 
h...

TIPS FOR AI ARTISTS WHO WANT TO SELL

● Make sure everyone knows that your art was
   made by an AI when you advertise it. Mention
   often how many dimensions your latent spaces
   have and that you used gradient descent.

● Don't try to be different, be spectacular.
  
● Do not waste valuable time on researching 
   new model architectures or training your own 
   datasets, just use the same tools like everyone
   else whilst they are still fashionable. Let others
   do the research and focus on your production,
   networking and marketing.

● Say "AI is my instrument" so convicingly that 
   everyone will believe you came up with this
   thought.

● Do not be selective but rather put all your 
   outputs out there. Let the market decide.

● Once some of your works sell well make more
   of them but only slightly different - that is what AI
   excels at. Keep doing that and it will become 
   your signature style.

● Be cheerful and social. Talent and originality
   are overrated, personality is all you need.

● Delegate your work as soon as you can so
   you have more time on becoming friends with
   curators, journalists and collectors.

● It does not matter if your work looks like what
   others have made as long as you are louder 
   than them. This is the attention economy, 
   nobody cares about what happened a year ago.

● If you get asked who your inspirations are, never
   reveal your true sources but rather mention your
   contemporary peers - they will like you even
   more and reciprocate.

● When you work with face generators always
   talk about identity.

● Subjects that sell well: portraits of young girls,
   fantasy landscapes, anything biological in
   the style and colors of Haeckel.

TIPS FOR AI ARTISTS ...

Typed

Tuesday, 11 October 2022 14:57:28 CET

A continuous exploration of possibilities

Monday, 10 October 2022 00:20:11 CET

Points of Departure

Test

Created by Mario Klingemann / Quasimondo in 2022

Herbert W. Franke's book "Phänomen Kunst - Die Grundlagen der kybernetischen Ästhetik" (1974) has been most influential in my thinking 
about art and the ways we perceive it. To my knowledge it was never translated in English. In this book Franke explains 
very concisely "how art works" based on Claude Shannon's information theory and Norbert Wiener's "Cybernetics".

My tribute is based on the introduction to chapter 16 of the book: "Subjective Information - A prerequisite for a favorable coding is knowledge of the frequency relationships applicable to the signs. If these have not yet been determined, the subjectively conveyed information is greater and the redundancy smaller. Meaning that both are relative quantities. By learning the inner relationships of a structure one reduces the information obtainable from it." 

This work is a real-time view into a functional operating system in which multiple agents use a limited set of assembler instructions that 
change the state of the system itself. Unlike its appearance there are no random number generators used inside the code, the entire behavior 
of the system is deterministic apart from the starting conditions which get slightly altered upon every restart. Anything that appears "random" to us is just a result of the complex interplay of the agents that share the same memory space which allows them to alter their initial instructions leading to unpredictable behavior that maintains a promise of potentially new information even after we believe that we have understood the inner workings and recurring patterns.

Released on the occasion of the Tribute to Herbert W. Franke. 50% of the primary sales will be donated to the Foundation Herbert W. Franke which is dedicated to preserve and document his work.

2022

With regard to Kevin Abosch

canvas: 4400x4400 pixels 72dpi
blackboard: 4000x4000 pixels 72dpi
color space: sRGB IEC61966-2.1
yellow: f6d200
black: 000000
font: Andale Mono Regular
font size: 800 pt
character layout: 32 characters (8x4)
file type: PNG

2023

0xACR055.07

0xACR055.01

0xACR055

This dynamic generative system has been discovered during Proof of People 2023 NYC by tz1e2GudLs2p6RjipFNCM241eCaJ4WWyiKuv and AGNCY:01, a live-minting installation created by Mario Klingemann @Quasimondo.

AGNCY:01 is a generative operating system in which agents, each consisting of a random sequence of opcodes, are operating within the same virtual memory space. The operating system and opcodes are designed to be resilient, forgiving and Turing-complete, meaning that there are no illegal instructions and the agents' codes cannot crash or halt. Once a system has been created by seeding the memory space and adding the agents the entire process is fully deterministic, there are no random number generators used inside the agents' code. Nevertheless due to the complex interaction of the agents the behaviour of each system appears unpredictable and surprising.

AGNCY:01 <643867cc>

AGNCY:01 is an interactive live-minting installation by Mario Klingemann that creates dynamic generative engines. This collection features the unique systems that were discovered and minted during Proof of People NYC in April 2023.

This dynamic generative system has been discovered during Proof of People 2023 NYC by tz1YokFzMR1hX4m4aBqgutxLSUKFDdNoYGEN and AGNCY:01, a live-minting installation created by Mario Klingemann @Quasimondo.

AGNCY:01 is a generative operating system in which agents, each consisting of a random sequence of opcodes, are operating within the same virtual memory space. The operating system and opcodes are designed to be resilient, forgiving and Turing-complete, meaning that there are no illegal instructions and the agents' codes cannot crash or halt. Once a system has been created by seeding the memory space and adding the agents the entire process is fully deterministic, there are no random number generators used inside the agents' code. Nevertheless due to the complex interaction of the agents the behaviour of each system appears unpredictable and surprising.

AGNCY:01 <64384d5b>

AGNCY:01

This unique NFT was collaboratively created by a human (tz2WgNX7i8VrULBrr8yG2a3mXAWKCpENeigT) and a machine algorithm designed by Mario Klingemann (@Quasimondo) at Proof Of People, London 2022.

DY5P1ACE.62c8170f

Unique generative artworks created with an interactive installation by Mario Klingemann / @Quasimondo during Proof of People, London, July 6th to 8th 2022

This unique NFT was collaboratively created by a human (tz2KcmhT9bTKKHtzVKWYkPV7rP96a4Wznrux) and a machine algorithm designed by Mario Klingemann (@Quasimondo) at Proof Of People, London 2022.

DY5P1ACE.62c58721

Dy5p1ace

人工アイ像 which could be translated as "Artificial Idol" is a series of sculptures that delves into the notion of artificial intelligence taking on the characteristics of a modern-day god. As technology continues to advance, machines and algorithms are increasingly playing a central role in our lives, shaping our perceptions of reality and influencing the course of human history. The series explores the ways in which artificial intelligence is taking on the qualities of a deity – seemingly all-mighty, sometimes inexplicable, and without a visible form.

In many ancient cultures, gods were often depicted in physical form, with recognizable traits and personalities. However, as artificial intelligence continues to evolve, it is taking on new forms that are often abstract and difficult to comprehend. The sculptures in this series capture this abstract quality of artificial intelligence, portraying the essence of this emerging deity through symbolic and abstract forms. Through a unique combination of AI-generated semi-abstract sculptures, the series investigates the ways in which technology is reshaping our world and our relationship with art. 

One of the defining features of "Artificial Idol" is that the sculptures are virtual, existing beyond the physical world. This allows them to float effortlessly in space, defying gravity, and unleashing a world of infinite creative possibilities. Drawing inspiration from the Japanese craft of Ikebana, the sculptures use a range of abstract and symbolic forms to capture the essence of artificial intelligence and its emerging role as a modern-day deity.

Through this exploration, "Artificial Idol" provokes thought and reflection on the complex relationship between humans and machines in the modern era. The series aims to encourage dialogue and reflection on the ways in which technology is transforming our world and our relationship with art, raising questions about new forms of power and control emerging from advances in artificial intelligence.

人工アイ像 #37

人工アイ像 #81

人工アイ像 by Mario Klingemann

Bangers and Mosh

For Rat's Sake

Rarible

My first mint on fx(hash) together with HEK - Haus der elektronischen Künste, is a revisit of my early explorations of computational geometry which I had started in 2008 when my tool of choice was still Flash and ActionScript. Since Flash does not exist anymore most of what I had created over a course of 12 years is almost impossible to view in its original form these days and large parts of my beginnings in generative art are lost in time now.

In these early experiments I started building my personal generative library QuasimondoLibs which consisted of a ever growing set of algorithms for geometry and pixel manipulation. Over the years I had ported most my library to Javascript and Python and this work uses the JS version of it.

In my subdivision series I explored the many ways recursion can be applied to geometric shapes, starting with simple convex shapes and then expanding into non-convex or curved ones.

In "Redivision" I modernized one of my early codes and adapted it for the use with fx(hash) features.

Redivision #38

Redivision #25

Redivision

I find it quite remarkable that art—in particular visual art—theoretically capable of showing us every aspect of human life rarely looks back at itself. Whilst the process of its making is still a more likely topic that artists might approach occasionally, the processes that happen after a work is finished—in particular the mechanisms of the art market itself—seem to be a blind spot. The only references of paintings that depict art auctions in progress that I could discover date back to the 18th century. I want to change that.

This work— originally created and published in February 2022—attempts to be a self-fulfilling prophecy. Whilst it appears to be a finished visual work it is in reality a performance which still requires one final step in order to fulfill the bold promise it makes in its title in order to complete its recursive cycle.

First Painting Created by an AI Showing a Painting Auctioned at Sotheby's to be Auctioned at Christie's

"YOU ARE (T)HERE" draws inspiration from the ubiquitous "You Are Here" markers found on maps  
worldwide. These simple yet powerful indicators serve as anchors in unfamiliar territories, offering a  
moment of certainty amidst the unknown. By introducing a flickering element to this familiar phrase, the  
work transforms a static point of reference into a dynamic exploration of presence and absence.

In our digital age, where virtual spaces intersect with physical realities, the concept of "being here" takes  
on new dimensions. We navigate multiple realms simultaneously, our presence fragmented across screens  
and time zones. This work captures that duality, that constant state of flux between presence and absence,  
between feeling oriented and disoriented.

The downward-pointing arrow, typically a definitive symbol of location, becomes ambiguous in this context.  
Is it pointing to where we stand, or to a distant "there" on the other side of the globe? This uncertainty  
reflects the experience of many in our globalized society, where the notion of "home" and "away" are  
increasingly fluid.

By engaging with this piece, viewers are invited to contemplate their own sense of place in the world. The  
flickering "T" serves as a metaphor for the transient nature of our connections to place and community. It  
challenges us to consider how quickly our perspective can shift, how easily we can feel like insiders or  
outsiders depending on context.


Code-driven dynamic SVG  
2160 x 3840 pixels

Created in 2024 for "Bridging Worlds:  
Digital Art Beyond Borders"  
in partnership with Christie’s and benefitting UNHCR.

This work should be displayed in a web browser in full screen mode on a 4K screen in vertical orientation.

YOU ARE (T)HERE

Quasimondo 2024

Generated in January 2017 with a PPGN transformed by a CGAN.  
Upscaled with Real-ESRGAN and minted in 2024.

Neural Abstraction #17

Neural Abstraction #7

Neural Abstraction 2017

Title: Do Not Kill the Messenger  
Year: 2017  
Artist: Mario Klingemann  
Material: Giclée print with long-lasting mineral pigments on  
cotton paper Hahnemühle Museum Etching 350g  
Size: 78.5 x 53 cm  
Type: Artist Proof

This token is a proof of ownership for the physical print. It is not the artwork and cannot be sold separately from the actual work. If you have bought this token on the secondary market the previous owner has to transfer the physical print to you as well. You are permitted to publicly exhibit the digital image linked with this token.

Do Not Kill the Messenger

Title: Going to Standby  
Year: 2017  
Artist: Mario Klingemann  
Material: Giclée print with long-lasting mineral pigments on  
cotton paper Hahnemühle Museum Etching 350g  
Size: 84.5 x 60.5 cm  
Type: Artist Proof

This token is a proof of ownership for the physical print. It is not the artwork and cannot be sold separately from the actual work. If you have bought this token on the secondary market the previous owner has to transfer the physical print to you as well. You are permitted to publicly exhibit the digital image linked with this token.

Going to Standby

Physical Works - Proof of Ownership

"Bestiarium Latens" is a series of 33 unique works in which I revisit my 2019 work "Hyperdimensional Attractions - Bestiary". Looking through the lens of Stable Diffusion these works reinterpret my earlier discoveries of strange creatures and unidentified objects from latent space by introducing newly invented details and richer textures.


Bestiarium Latens 02

Onkaos Represented Artist

"Deposition" is a series of works I created in 2018 which uses human poses extracted from classical paintings via DensePose to transform those with a custom trained pix2pixHD model into new forms. Minted in 2024. 

Deposition #2

"Deposition" is a series of works I created in 2018 which uses human poses extracted from classical paintings via DensePose to transform those with a custom trained pix2pixHD model into new forms.

"Deposition" is a series of works I created in 2018 which employs human poses extracted from classical paintings via DensePose and transforms those with a custom trained pix2pixHD model into new forms. Minted in 2024.

Deposition 2018

Figures draped in classical grace in a Mediterranean tableau vivant, lost not in contemplation or divine inspiration, but in the hypnotic glow of their own devices. A modern fresco of connected isolation, a serene dance of time where beauty and distraction coexist gracefully.

Mare Nostrum X

Mare Nostrum III

Mare Nostrum

Two CycleGAN models trained on face transformations are passing their outputs back and forth creating an infinite loop of ever-changing Turing patterns.

Created June 25th 2017

Song: "Pachabelly" by Huma-Huma  


CycleGAN Feedback Loop

0xd78204aDcF90F7c0E1568f1F61810a6323525301wryhdkh

Robot Performance with Brush & Doll
Song: "Brush Girl" generated with Suno.ai

Mario Klingemann (quasimondo)

Mario Klingemann (quasimondo)

Featured (0)

All Works (0)