2024-01-19 05:29:00
original title:
Artificial Intelligence: Bringing more innovation and possibilities to art
Young students draw two-dimensional comics, girls are keen to generate beautiful photos of themselves, and children get a Van Gogh-style painting created by themselves… More and more ordinary people are using artificial intelligence (AI) in a relaxed and casual way. ) painting tools.
American computer scientist and psychologist Herbert Simon is a pioneer in the fields of artificial intelligence and cognitive science. His research has had a great influence on the fields of art and design. Simon believes that creativity is not exclusive to artists and writers, nor is it some mysterious and elusive talent, but a basic aspect of human cognition, an ability that can be realized and enhanced through cognitive processes and computational methods, and can be achieved through specific Heuristics and procedures to cultivate.
In the field of art and design, humans often create through so-called inspiration, and AI painting tools can process and analyze large amounts of data, identifying patterns or connections that humans may overlook or are simply unaware of. French-American artist Marcel Duchamp is an important figure in modern art. He is best known for his belief that art should challenge tradition and convention. He believes that the essence of art lies in the creation of new concepts and perspectives, not just the display of skills or aesthetics. In the era of artificial intelligence, using AI for artistic creation is about the creative expression of data, algorithms and programming, so AI will inevitably bring more possibilities to art.
Moreover, the intervention of AI has made the human role and creativity in the art creation process more complex. The artist may no longer be the sole creator of the work, but a guide or collaborator working with AI. This change redefined the role of the artist and the nature of the work of art. This provides new ways of thinking about the definition, function and social significance of art.
The Central Academy of Fine Arts, where I work, took the lead in using AI to carry out a large number of experimental explorations in art and design. I would like to briefly introduce some typical examples, from which we can see the innovation brought by AI to art.
We systematically sorted out the digital data of surviving Song Dynasty paintings, and marked the data from aspects such as layout and composition, brush and ink characteristics, and object classification of Song paintings, forming a systematic and huge data set. How to label manual data belongs to human subject consciousness and cannot be replaced by artificial intelligence. Aesthetically, the role of humans is to tell AI what the definition of “beauty” is. We use the aesthetic characteristics of Song Dynasty paintings to train artificial intelligence, so that it can learn the Chinese people’s unique abstract awareness of natural objects, and cooperate with AI painting tools to “paint” personalized pen and ink language, which is very interesting.
We also materialized these interesting AI Song Dynasty painting images on furniture and utensils, which is the “artificial natural furniture” we designed. Artificial natural furniture is new Chinese-style furniture created using artificial intelligence generation technology and precision digital manufacturing technology. These furniture embody the most advanced technology and design ideas at the moment.
We have also tagged data on Wang Xizhi’s surviving inkwriting, allowing artificial intelligence to learn the rhythm and density of his brushwork. AI’s understanding of Wang Xizhi’s calligraphy is inherently different from humans. It can write non-Chinese character “calligraphy” with the charm of Wang Xizhi, and the results are amazing and interesting. This creative research path can also be introduced into the font design of practical arts. Chinese characters are different from the font design of alphabetic characters. The font design requires character-by-character design, which is very complicated. Artificial intelligence has great potential to design fonts for Chinese calligraphy. Create a new situation.
The murals in Yongle Palace began in the Yuan Dynasty and were repaired and added many times in the Ming and Qing dynasties. They are one of the largest and richest Taoist murals in China and a treasure of Chinese civilization. Historically, these murals have lacked maintenance for a long time, and there have been many damages, losses, and repairs that do not conform to the original intention of the painting. We use artificial intelligence technology to establish a set of workflows from data collection and analysis to image reconstruction, deep learning style, to expert verification and AI model parameter adjustment and optimization. This workflow is more efficient and reversible than manual restoration technology. It is very promising to apply image restoration technology based on AI deep learning to reproduce the style of ancient murals.
As an important part of Chinese traditional culture, Chinese folk art is deeply rooted in the daily lives of ordinary people. It reflects the living habits, customs, beliefs, cultural genes and aesthetic tastes of ordinary people. Many Chinese folk arts, such as paper-cutting, clay sculpture, embroidery, tie-dyeing, folk painting, weaving, etc., have hundreds or even thousands of years of history and contain strong national and local characteristics. With the development of society, some traditional folk art forms have lost their original social foundation and living environment, and many folk art skills that need to be passed down from masters to apprentices are at risk of being lost. AI can solve the problem of continuation and activation of folk art to a certain extent. We have systematically sorted out the genealogy of Miao patterns, trained AI to learn the visual characteristics of Miao patterns by category, and created several small artificial intelligence models that can generate specific categories of Miao patterns, such as cow patterns, frog patterns, butterfly patterns, There are also various Miao patterns expressing scenes of farming, hunting, sacrifice, dance, and festivals. Artificial intelligence can become a good inheritor of folk art.
Data labeling plays a vital role in the machine learning process. It is not only the basis for AI model training, but also a key factor in ensuring the quality and applicability of model generation. The earliest fashion show appeared in the 1860s, also known as the “T-stage show.” To this day, it is still a established format for the industry to release new clothing products, and it continues to flourish. The massive T-stage catwalk photos in cyberspace are a “natural” data set for training fashion artificial intelligence, because these images have been clearly “labeled” by category – they reflect the design concepts and style creativity of different brands and designers. As well as details, cuts, materials and accessories. Reasonable mining and use of these fashion data can allow AI to play an important role in trend prediction, style deduction and design generation of fashion design. In the class of the School of Design of the Central Academy of Fine Arts, we guided students to classify and organize fashion images based on different forms of clothing, and trained an AI fashion generator. Its way of designing is completely different from previous generations of designers and is innovative. .
In the era of artificial intelligence, artistic creation and language text are closely related, and AI also has great potential in this regard. Our teaching team has done a small experiment – using AI technology to judge whether the first 80 chapters and the last 40 chapters of “Dream of Red Mansions” were written by one person. We first used algorithms to extract the features of the two parts of text, including word frequency, sentence structure, sentence complexity, grammatical features and emotional expression, and then trained an AI model to distinguish the writing styles of different authors based on text features. The experimental result is that the first 80 chapters and the last 40 chapters were not written by the same person. This shows that AI’s understanding of human language has reached a whole new level.
After AI lowers the threshold for professional design, more ordinary people can enjoy high-quality design services. For example, in rural China, the total amount of homestead land is huge, and hundreds of millions of villagers need to build houses on the homestead land. Few people in the countryside are willing to spend high design fees to hire architects to build houses for their homes. Is it possible to use artificial intelligence to design houses for villagers with professional structures, excellent performance, decent appearance, and appropriate prices? In the age of artificial intelligence, this is clearly possible. Professional designers can “nurture” the professionalism of design to artificial intelligence through machine learning, so that high-quality design services can benefit more people.
Our country has huge data resources, high-quality information and Internet infrastructure, the largest number of Internet users, and massive artificial intelligence data sets that can be collected and trained. Of course, it also has rich application scenarios for artificial intelligence and digital technology. The new generation of designers explores the design application of artificial intelligence in various life scenes, production scenes, social scenes, knowledge scenes, and consumption scenes. The opportunities are endless, and perhaps what they challenge is the limits and possibilities of human imagination.
(Author: Li Shiqi, associate professor at the Central Academy of Fine Arts)
#Artificial #Intelligence #Bringing #Innovation #Possibilities #Art #Theory #China #Engineering #Network