Artificial intelligence opens up new space for art creation
Author：Eve Cole
Update Time：2024-11-22 11:18:01
In recent years, with the emergence of deep neural network technology, especially AI (artificial intelligence) algorithm models, the development of AI painting has accelerated. AI painting follows CG painting (computer-generated painting, that is, painting with the assistance of graphic design software), DSE painting (digital special effects painting, that is, painting using a set of digital special effects technology with computers as the core) and VR painting (virtual reality Painting, that is, painting using virtual reality technology, the fourth generation of digital painting after artists can use VR equipment to paint in a simulated three-dimensional space). Compared with the first three generations of digital painting, AI painting has undergone major changes in three aspects: creation mechanism, artistic ontology and reception aesthetics.
"Bamboo forest" themed landscape painting created by artificial intelligence. Image provided by the author. Human-computer interaction and collaborative creation of AI paintings construct a new creative mechanism in which humans and AI platforms interact and collaborate. Under this mechanism, after logging into the AI painting platform, users only need to simply provide creative requirements, style and interest, and theme implications in text or language, and the AI will do the rest. For ordinary people, this creative method is unparalleled in convenience and ease of learning. Compared with the first three generations of digital painting, AI painting has lower requirements on users’ computer software and hardware operating skills. The first two generations of digital technology, marked by computers and the Internet, have expanded the public’s rights to artistic creation to a certain extent, opened up the road to inclusive technology, and incubated the fields of geometric aesthetics, spectacle aesthetics, and interactive aesthetics respectively. The first three generations of digital painting, but these three painting methods require the painter to master certain computer software and hardware operating skills, and the scope of painting subjects is limited. With the advent of the AI era, the technical threshold for digital painting has been greatly reduced. Theoretically, with the help of the AI painting creation model, anyone can freely create art - they can paint Chinese paintings or Western oil paintings; they can paint classical oil paintings in the style of Leonardo da Vinci, or they can paint modern paintings in the style of Picasso. Oil painting; you can paint flower and bird paintings, landscape paintings or figure paintings... If you are not satisfied with the work, you can modify the keywords at any time, adjust your appeal and theme, until you finally see the work you are satisfied with. With the empowerment of the AI painting platform, ordinary people can easily create all types of paintings and paintings they can think of - technically speaking, there is nothing that AI cannot draw except what you can't think of. In AI painting, users are more like planners, proposers, conceivers and decision-makers of artistic creation, while the AI platform is the specific executor of these artistic conceptions, artistic planning, artistic imagination and artistic decisions. This new type of human-machine collaborative creation mechanism not only expands the scope of creators, but also greatly improves creative efficiency. Aesthetic styles are diverse and the aesthetic characteristics of popular AI paintings have also undergone significant changes compared with the previous three generations of digital paintings. First, the artistic styles of AI paintings are more diverse. On the mainstream AI painting platform, we can see a large number of surrealist-style works, a large number of highly subjective works, and a large number of naive, naive, and natural-style works. This is completely different from the first-generation digital painting dominated by geometric aesthetic style, the second-generation digital painting dominated by spectacle aesthetic style, and the third-generation digital painting dominated by interactive aesthetic style. Second, the mainstream aesthetic taste of AI paintings tends to be popular and popular. Although there are some works with unique styles among AI paintings, they are not mainstream in the entire AI painting ecosystem. The aesthetic style of most works is obviously popular and popular. This is related to the fact that the vast majority of users of the AI painting platform are non-professional painters. Most of them do not have professional art knowledge and draw on the AI platform based on their hobbies. The works they create reflect popular aesthetics. AI painting does not emphasize professionalism like traditional painting and the first three generations of digital painting. It is more popular, reflects the aesthetics of most people, can better represent the true emotions of the public, is more likely to resonate with the public, and is more likely to generate hits. work. Third, there are very few fake and harmful AI paintings. This feature is related to AI’s unique algorithm mechanism. The AI algorithm has set up three firewalls to prevent the generation of counterfeit and harmful works. First of all, when cultivating the algorithm, the works of art used must be carefully selected. This mechanism excludes all works that are harmful to human physical and mental health, ensuring that the works of art for AI learning have high artistic quality; secondly, the works When outputting, any work that is harmful to human physical and mental health will be naturally eliminated by the AI system and will not flow out from the output end; finally, the creation end also has a protection mechanism. When the AI receives instructions from the user that are contrary to morality and law, it will reject them. Accept instructions. These three AI built-in creative protection mechanisms block the output of counterfeit and harmful works to the greatest extent. AI painting also has an aesthetic feature that is very different from traditional painting and the first three generations of digital painting - acceptance and creation have the same origin .
Based on the AI painting creation mechanism explained above, we can easily find that users as the subject of AI creation mainly exist as appreciators of AI works. In addition to issuing specific and clear creation instructions, they do not undertake specific creative tasks. A large part of their attention is spent on appreciating and judging the works instantly output by AI. In practice, given that it is difficult to satisfy the user's aesthetic requirements with one output, AI is likely to continuously receive modification instructions from the user and continuously output a series of works, and the user must continue to appreciate the work before he or she can provide modification opinions. This aesthetic feature of creation, that is, appreciation, is very different from the previous three generations of digital painting and traditional painting. In the first three generations of digital painting and traditional painting, if artists want to create an ideal "bamboo in the painting", they must first obtain the "bamboo in the eye" through purposeful life experience, and on this basis, after repeated deliberation and conception , further generate the "bamboo in the heart" with a distinct subjective aesthetic consciousness, and finally, under the appropriate opportunity, use artistic language, creative skills and auxiliary tools (such as computers) to draw the "bamboo in the painting". This means that in traditional painting and the first three generations of digital painting creation, the creation and reception of works are largely separated. The recipients of these works are mainly the audience, not the painter himself. The fundamental reason why AI painting has such unique aesthetic characteristics is that AI has super computing power, understanding ability, independent learning ability and imitation ability. It is this ability that gives the neural network algorithm model generativity, which is a significant advantage that traditional AI algorithms such as early decision trees do not have. In order to be different from the geometric aesthetics, spectacle aesthetics and interactive aesthetics corresponding to the first three generations of digital painting, we define the aesthetic characteristics of AI painting as computational aesthetics or intelligent aesthetics. Judging from the current AI painting creation ecology, there is another phenomenon that cannot be ignored, that is, although there are very few fake and harmful works generated by AI, there are also works with high aesthetic value, artistic value and ideological value. not much. This phenomenon has a lot to do with the limitations of the current AI painting algorithm model design. The large language model corpus used in AI painting training is all set by humans. The information that AI comes into contact with during training has been manually selected and subjected to multiple symbol conversions. In other words, what the AI algorithm learns and imitates is only human painting. The pattern and style of the work. The creation of human artists is also inseparable from learning and imitation. However, in addition to learning and imitating the patterns and styles of similar works, human artists must also learn and imitate the experience and logic in life, society, reality and history, and follow these logics to engage in work. Art creation. In addition, all inspirations and motivations for human artistic creation originate from the needs of one's own life and life; human artistic creation is all intended to express the understanding of life, society, reality, history, and human nature. Therefore, all works of art created by humans are deeply imprinted with human thoughts and emotions. When relevant literary and artistic works show the essence of life and human nature, they can arouse people's resonance. In contrast, the current AI painting algorithm model has not yet learned to experience human life, and has not yet understood the complex relationship between human nature and art. This is the foundation that all AI art, including AI painting, must overcome in order to achieve new development in the future. problem.
