ForkLog has investigated which AI models are used to work with images and whether such systems could replace artists.<\/p>\n
- \n
- Researchers began applying image-creation algorithms in the 1950s\u20131960s.<\/li>\n
- Neural networks allow copying artists\u2019 styles, turning sketches into photorealistic illustrations, \u201cbringing to life\u201d portraits, and creating new images.<\/li>\n
- The cost of developing and training an algorithm ranges from zero to hundreds of millions of dollars.<\/li>\n
- AI art can inspire, but its accessibility may raise a number of issues.<\/li>\n<\/ul>\n<\/div>\n
A Brief History of AI Art<\/h2>\n
The history of generated AI art can be traced back to the advent of machine graphics and the invention of the computer. Early researchers used basic algorithms to create simple patterns and forms.<\/p>\n
In 1967 the German mathematician and scientist Frieder Nake developed a portfolio titled Matrix Multiplications<\/a>, consisting of 12 images. He created a square matrix and filled it with numbers that were multiplied by themselves in sequence.<\/p>\n
The researcher translated the results into images of given intervals, assigning a visual sign in a particular form and colour for each value. He then arranged the shapes on a grid according to the matrix values.<\/p>\n
In his works Nake frequently used a random-number generator and probably partially automated the multiplication process.<\/p>\n
![\"End](\"https:\/\/forklog.com\/wp-content\/uploads\/1-140.jpg\")
Image from the Matrix Multiplications portfolio, created by Frieder Nake. Data: Tate<\/a>.<\/figcaption><\/figure>\n In 1973 the artist Harold Cohen developed a set of algorithms AARON<\/a> capable of drawing \u201cby hand\u201d specific objects. He found the system began to create forms previously unknown.<\/p>\n
Initially the program generated abstract paintings, then learned to draw more complex shapes, including stones, plants and people.<\/p>\n
![\"End](\"https:\/\/forklog.com\/wp-content\/uploads\/2-603-810x1024.png\")
Artwork generated by AARON. Data: New Atlas<\/a>.<\/figcaption><\/figure>\n Since 1990 researchers and artists began using AI models in robotics, teaching machines to create paintings and sculptures.<\/p>\n
In 2015 the Google engineer Alexander Mordvintsev launched the computer-vision program DeepDream<\/a>, which used a convolutional neural network to search for and amplify patterns in images using algorithmic pareidolia<\/span>.<\/p>\n
The system works by distorting the original image according to which fragments resemble familiar objects.<\/p>\n
When Google published the approach and released the source code of the algorithm, the market saw many tools and services that allowed anyone to transform their photos into \u201cpsychedelic\u201d images.<\/p>\n
![\"End](\"https:\/\/forklog.com\/wp-content\/uploads\/3-361-1024x384.png\")
Source image (left) and processed with DeepDream (right). Data: MartinThoma<\/a>.<\/figcaption><\/figure>\n In 2022 AI art appeared in various domains, including marketing, fashion and entertainment.<\/p>\n
![\"End](\"https:\/\/forklog.com\/wp-content\/uploads\/4-228-819x1024.png\")
Cover of Cosmopolitan magazine created with the DALL-E 2 algorithm. Data: Cosmopolitan<\/a>.<\/figcaption><\/figure>\n Also, these models help create paintings.<\/p>\n
![\"End](\"https:\/\/forklog.com\/wp-content\/uploads\/5-18-1024x576.jpg\")
Painting Th\u00e9\u00e2tre D\u2019op\u00e9ra Spatial, created by Jason Allen with the Midjourney algorithm. Data: Motherboard<\/a>.<\/figcaption><\/figure>\n Neural Networks for Image Work<\/h2>\n
2022 may go down in history as the moment AI art went mainstream. A boom in high-quality tools built on various algorithms makes neural creativity accessible to anyone with a smartphone connected to the internet.<\/p>\n
AI models let users copy painters\u2019 styles, convert sketches into photorealistic illustrations, \u201cbring portraits to life\u201d and create new images. Different tasks employ different or similar approaches and tools.<\/p>\n
Neural style transfer (NST) is a method based on convolutional neural networks that can create a painting imitating the style of another image. A user can transform a photo of a running dog into an engraving by Katsushika Hokusai<\/span> or generate the Mona Lisa in the brush of Jan Vermeer<\/span>.<\/p>\n
![\"End](\"https:\/\/forklog.com\/wp-content\/uploads\/6-12-1024x569.jpg\")
Source photo (left), image in the desired style (center), and result (right). Data: Instapainting<\/a>.<\/figcaption><\/figure>\n Generative adversarial networks (GAN) power the creation of new artworks or paintings in the style of other images. GANs consist of two models: a generator that produces content and a discriminator that evaluates it.<\/p>\n
GAN-based systems can produce images that resemble pictures in the training set, including faces, cats\u2019 mugs, furniture, and other objects.<\/p>\n
![\"End](\"https:\/\/forklog.com\/wp-content\/uploads\/7-55-1024x1014.png\")
Faces of non-existent people, generated by a generative adversarial network. Data: This Person Does Not Exist<\/a>.<\/figcaption><\/figure>\n GANs can also help \u201cbring to life\u201d a sketch of a landscape.<\/p>\n