AI Painting Popularized

One of the hottest topics in the science and technology industry recently is AI painting. This time, in terms of creative design, people have found that machines can do better. It seems inevitable that the day when human beings will be replaced by AI is getting closer and closer.

The rate of evolution of AI painting is quite amazing. In January 2021, OpenAI, a famous US artificial intelligence laboratory, launched the “DALL-E” algorithm, which can “create realistic images and art through natural language descriptions.” In February this year, the Disco Diffusion algorithm was launched. The main principle is to rely on the machine learning of pixel diffusion to simulate the painting process. By April of this year, the DALL-E2 2nd generation has launched, and is shown to be equipped with a very polished text-to image function.

In a painting contest in September, a piece called the “The Space Opera House” stood out among several hundred. When the artist received the award, it was discovered that the work was created by AI instead of the artist himself, which seemed to contradict with the category definition of painting and caused controversy among other artists. At the same time, it also successfully popularized AI painting by showing that AI algorithms are already accomplished to such a great degree in the field of creative art.

The contestants used an open AI painting platform called Midjourney, which painted an image based on the contestant’s descriptions. After some modifications on the AI product, the contestant produced the winning work. In other words, it took him less than two years to cultivate a master painter.

How does creating a well-conceived painting generate revenue?  At present, there is only a simple appreciative value.
In the field of art commercialization, art is generally used in the field of publicity and display, film and television, comics and other media creations. However, in these industries, art needs to convey a lot of specific emotions, whereas aside from painting, there are a lot of things that cannot be expressed using AI.

For example, a comic requires character building and setting the general story direction. Obviously, AI can’t grasp characters nor the story. What it can only do, is offer help in designing characters.

Or, if we want an artistic piece that can equally express content and ideas with pictures and text, such as a poster, obviously AI cannot create a poster with both text and images, since the text-to-image algorithm and text will contradict and meddle with each other. As for conveying complex emotions through images, such as images with subtle implications for example, it is difficult for AI to do based on diffusion algorithms alone.
The advantage of AI is that it takes a very short time to complete a picture that is exquisite in conception, rich in details, and one that contains creativity that humans may not necessarily think of.
In short, mechanical creative labor, such as drawing people, models, patterns, etc., can be appropriately replaced with AI. Besides, as AI is able to pay a high degree of attention to detail, a lot of time spent on drawing these repetitive and tedious details can be saved. That is to say, in the future of painting design, the workflow may change. No longer may artists start with a blank canvas, but they may choose from an archive of AI images an image worth modifying, and then constantly adjust it until they finally get the result they want.
But regarding this application of AI use, there are still many more details that need to be considered. For example, for different industries, AI responsible for smaller portions for art created by different personnel need to be differentiated, and which customer groups should the vendor charge, how much the actual purchasing power of users is, and how to balance the high cost of computing power are keys to whether AI painting can smoothly adjust to all aspects of life.
For some time in the visible future, AI painting will still be used to supplement art to the Internet gallery.

PHP Code Snippets Powered By :