[Explanation] Now, enter a few keywords and wait for dozens of seconds, and the AI painting platform can generate paintings that are comparable to human painters.
Even if you are a painter, you can use AI to generate your own fine paintings.
How does AI paint?
How to turn inspiration into beautiful pictures?
[Concurrent] Xiao Xinyan, Chief Architect of Baidu Wenxin Yige
How does AI come to paint?
First of all, AI needs to learn to draw from the data corresponding to the massive images and texts. For example, a mountain peak can be a wall in a Chinese painting, or it can be a majestic and solemn oil painting.
When we input a description of a language, the AI goes through hundreds of rounds of constant revisions to the painting, and finally forms a finished product that is aesthetically highly consistent with human experience and knowledge.
[Concurrent] Cao Yi, Deputy Secretary-General of China News and Comics Research Association
I think the biggest difference between AI painting and traditional painting is whether there is a requirement for the artist's modeling ability.
A painter needs a long accumulation and training to acquire modeling ability, but AI drawing software omits modeling ability, and users only need to enter keywords to get the corresponding picture.
[Explanation] In 2022, after the introduction of the diffusion model, AI painting has achieved a technological breakthrough. It can generate beautiful and clear and delicate images, and begin to enter the film and television, advertising, architecture, fashion, new media and other industries.
[Concurrent] Cao Yi, Deputy Secretary-General of China News and Comics Research Association
It does present a huge challenge for the artist.
First of all, AI drawing is very efficient. News cartoon designers like me negotiate with the editor after receiving the manuscript. The total time required for sketching and drawing the finished news manuscript is three to five hours, while the paid AI drawing software may not take a minute. If necessary, many pictures are generated for the user to choose.
Second, AI painting is very delicate.
Third, AI painting is very cheap, which is very important. It requires a very high cost for painters to draw a picture. Although AI drawing software requires a lot of cost at the beginning of design, its copying cost and use cost are basically affordable. can be ignored.
[Concurrent] Xiao Xinyan, Chief Architect of Baidu Wenxin Yige
AI painting can improve efficiency and assist in creation. It can inspire professional visual creators such as painters, designers, and artists. It can provide high-quality and efficient illustrations for text content creators such as media and authors, and easily achieve both pictures and texts. .
[Explanation] With the continuous maturity of AI painting technology, it has brought challenges to the related art industry. Will AI take the "job" of art practitioners?
Cao Yi said that, in fact, the crisis of grabbing a "job" is not the first time.
Artists should expand their horizons and deal with technological changes calmly.
[Concurrent] Cao Yi, Deputy Secretary-General of China News and Comics Research Association
Around the 20th century, the invention and popularization of photography had a far greater impact on the people at that time than today.
There is fear in people's hearts.
At that time, the realistic style of painting gradually declined.
At the same time, we have also seen new ideas, new genres, and new styles spring up.
AI drawing is a tool after all, we have to use a good tool, and never turn ourselves into a tool.
Artists should expand their horizons, improve their painting skills and aesthetic level, so as to calmly cope with the development of the times and technological changes.
[Explanation] At the same time, according to the observation of industry insiders, there are also many problems that need to be solved urgently in the AI painting industry, such as copyright disputes and related industry standards to be improved.
[Concurrent] Xiao Xinyan, Chief Architect of Baidu Wenxin Yige
First, it is necessary to strengthen intellectual property protection for AIGC (Artificial Intelligence Generated Content), develop AI innovation businesses such as AI painting, and create a secure regulatory and policy environment.
Second, support original paintings to enjoy copyright according to law.
Third, based on different subdivision application scenarios and product model characteristics, classify and grade, and formulate rules that can be implemented.
Dong Yufei reports from Beijing Expo
Responsible editor: [Luo Pan]