Critical Review Sara Voaides 521838
Why artificial intelligence will not kill illustration
Historical precedent for the illustration industry changing due to technology
Introduction
Artificial intelligence (AI) and machine learning are the final frontiers of technology. The link between technology and the artist is as strong as the link between humans and tools, and artists have historically taken advantage of new tools. However, tools have also replaced entire crafts. The concept of artificial intelligence has been a controversial one, especially when machines substitute people. The discussion around the ethics of neural networks created with stolen images is the most common, but while it is a vital aspect of the AI topic, it isn’t the focus this critical review. The focus is the historical precedent of technology replacing artisans and illustrators, and how illustrators have in turn persevered and evolved by using said technology to their advantage.
Not too long ago most would have said AI cannot produce artwork, not well enough to fool a human, but this is changing. Can computers truly make beautiful images? Does this cheapen art, or make art and creativity more accessible? If so, what does easy access to making imagery mean for illustrators, whose work is by definition commercial and not created for its own sake? These questions can be answered by exploring the relationship between illustration and technology throughout history and by demystifying what artificial intelligence is, how it works, and how people and artists use it.
Technology and Illustration
Artists use technology to aid efficiency and enhance creativity. From wooden dowel to digital tablet, tools for art have existed since the beginning of human history. The tools we use, our minds and our bodies all come together to form an art-creating “unit”(Wamberg and Paldam, 2015). We have been improving on each of these elements as science has progressed. Out of the three, tools have changed the most. The illustrator has historically been pragmatic about technological improvement to make image-making effective and time efficient. Changing technologies have been embraced and used even if it meant the death of a previous medium or tool .
Printing
Figure 1 Johannes Gutenberg (1400-1468), German printer, printing the first sheet of the Bible. after hombres y Mujeres celebres. Barcelona 1877.
Printing technology is perhaps the most significant ally of illustration throughout history. The European introduction of the movable-type printing press by Johannes Gutenberg in 1450 took accessibility to images and the artists’ ability to self-promote to a new level. The printing press removed the need for manuscript copying and later for illuminations. More books meant more images, but even more important than books were cheap and accessible broadsheets and broadsides (pamphlets). A regular person could purchase them, or be exposed to them via other people who did. The printing press also allowed artists to have fledgling independence from patrons.
The industrial revolution brought about advancements in the makeup of printing presses, bringing about the so-called “Golden Age of Illustration”. Printing provided the bulk of entertainment for the average Victorian home, with a plethora of magazines, periodicals and picture books being printed every year and entertaining people from all walks of life. This was a prime time to be an illustrator, as photography was in its first steps and images were more in demand than ever for everything from product advertising to fashion illustrations, to technical drawings for scientific purposes (Doyle, Grove, and Sherman, 2018).
Woodcutting
Figure 2 A pitched Battle, woodcut by Urs Graf, 1521
Woodcutting (or xylography) was the first chapter of the illustration industry. By creating a woodcut, a single image could be printed in multiple copies without the need to start from scratch and could be shared with many people simultaneously. An unlearned person holding and even owning an image made by a master of their craft was unprecedented. Steven Heller famously called illustration “the people’s art”, and the technology of woodcutting was the first to make this possible.
Intaglio (Etching, Engraving, etc.)
Figure 3 Self Portrait, 1630 (etching), Rembrandt Harmensz. van Rijn (1606-69)
Figure 4 St. Jerome in his Study, 1514 (engraving), Dürer or Duerer, Albrecht (1471-1528)
During the Renaissance, engraving was favored over woodcutting for depicting artwork. Engraving allows for a finer line and more detail. Etching (a chemical process) became popular after advancements by Jaques Callot (1592-1635) who helped improve the technique by inventing new tools and varnishes. Woodcutting, engraving, and etching remained as part of a toolset for a long time, one not replacing the other, but each having its own use. This would change during the 19th and 20th centuries with the invention of lithography and offset printing (Doyle, Grove and Sherman, 2018).
Lithography
Figure 5 Mitterndorf in Austria, after an original watercolour (colour litho), Jahn, Gustav (1879-1919) (after)
Lithography allowed for more freedom in texture and color than engraving. It was created as a cheap alternative to traditional printing and was adopted in illustration because of its ability to create line-less images (Bryans, 2000.). Lithography gradually replaced engraving and etching, until it transformed into the modern offset printing process, substituting letterpress printing during the latter 20th century because of its much sharper image quality and production speed.
Photography
Figure 6 Valentino Spring Summer Collection, Rome, 1958 (b/w photo)
The advent of photography was either the biggest boon or the worst blow to the illustration industry, and its effect on illustration has some similarities with current AI advancements. Many would give the photographic camera full credit for replacing the illustrator as image-maker for the masses. Others would say without the need to depict realism, the illustrator can flourish as a communicator (De Font-Réaulx, 2012).
The attitude around using photography as a tool for making art was oftentimes derided in the fine arts because it allowed anyone to capture the world around them without the need for extensive education. Illustrators rarely had such hang-ups. Photography makes illustration much more efficient (Mevlut, U.N.A.L. and OZTURK, M.S., 2019.). It also gives more people access to image-making and capturing the world around them easily. Where printing brought about the democratization of owning imagery, photography allows for a form of creativity accessible to those with little artistic skill. AI art could become an equivalent to amateur photography or product photography.
Digital Technology
Figure 7 The Escape, 2001 (digital collage), Skogrand, Trygve / Private Collection / © Trygve Skogrand. All rights reserved 2022
Figure 8 Cactus Love, 2015 (digital)
Figure 9 Family Race, 2014 (digital illustration)
The 20th and 21st centuries brought many changes to the way illustrators approach their craft, and digital technologies are the ideal medium for the needs of illustrators today (Ozcan, Cidik, and Kandirmaz, 2021.). Computers, phones, drawing tablets, etc., all contribute to the process of making, publishing, selling, and consuming illustrations of all kinds. We have exchanged the physical and chemical processes for vectors and pixels. Most of the images we consume in our daily lives do not require ink, paper, or physical existence at all.
Tools have never defined what it means to make images. Humans make images and the tools, no matter how sophisticated, are the interchangeable medium for our creativity. As technology has evolved, so has accessibility to image making. More people than the few rich can now afford to own images and advanced tools. People can not only use technology to capture reality, but do so realistically and artistically. Yet, this does not seem to eliminate the need for the illustrator, but to reinforce it. Though illustration will never be quite as ubiquitous as it was during the nineteenth century, before the invention of photography, the demand for storytellers is as high as it ever was, if not higher.
Artificial Intelligence and Illustration
Can technology replace the illustrator entirely? Commercial illustration is not the same as fine art, made for its own sake. If the layman is now capable of generating imagery at a quality good enough people cannot tell it’s machine-made, will illustrators be needed? Before answering, let us explore the history of artificial intelligence and how it works.
Early History
Figure 10 Substrate, Processing, June, 2003, J.Tarbell
Figure 11 AARON image created at the Computer Museum, Boston, MA, 1995.
The first images made with computers were created in the 1960s. By inputting fractal, cellular automata or genetic algorithms, one could program a computer to create geometrical forms modeled with absolute mathematical precision (Fig. 10). These programs are still used in architecture today. The algorithmic approach was one of many to come (Manovich, 2019).
In 1973, a program called AARON was created by artist/programmer Harold Cohen. AARON functioned on a series of rules inputted by Cohen himself (Fig. 11). It would make “decisions” while drawing either digitally or with a physical plotter. While AARON could create original pieces each time, it could not learn without the artist’s direct input (Ai Magazine, pp.63-66).
Today’s Artificial Intelligence and Neural Networks
Today’s AI is not strictly algorithmic, nor does it rely on the programmer introducing rules into its system. Using a Neural Network, an enormous repository of data (labeled images) the AI creates associations and builds connections, learning to “recognize” new images. However, it can be lead wrong if the images provided are biased or have not been labeled properly.
Neural Style Transfer (NST)
Figures 12, 13, 14 Neural Style Transfer (NST) Tutorial
AI’s ability to find commonalities between artifacts successfully created the Neural Style Transfer (NST) AI art method, proposed by Gatys et al. This method creates the fastest“artistic looking” results. After building the Neural Network, AI is able to apply “content” and “style”. Whatever image the user chooses as “content” (a banana) will be used as the base of the piece, while the image the user chooses for “style” (Van Gogh’s Starry Night) becomes the “filter” over the “content” image (a banana in the style of Van Gogh’s Starry Night). This process is straightforward, using images the user chooses to create new “artwork” (Cetinic and She, 2022). The process can be observed in the figures above.
Generative Adversarial Network (GAN)
Figure 15
“Text 2 Dream” image created by inputting key words into the program
Figure 16
“Deep Style” image created by style transfer method
Figure 17
“Text 2 Dream” image created by inputting key words into the program
Figure 18
“Text 2 Dream” image created by inputting key words into the program
A more advanced form of AI is the Generative Adversarial Network (GAN), created by Goodfellow et al. Generating a new image requires two different neural networks: a “training” network and a “discriminating” network. The purpose of the first is to “teach” the second what the most relevant images are, the second network can then make a more accurate decision. This way, the AI is able to fine-tune its results and “learn” how to differentiate between images and the hierarchy of their relevance on its own, as long as it is provided with raw content (Cetinic and She, 2022).
Artificial Intelligence Creative Adversarial Network (AICAN)
Figure 19
Figure 20
Figure 21
Figure 22