Is the best thing you can draw with a pen or brush something that looks like a 3-year-old’s kindergarten homework?
Enter the age of text-to-image art generators. With a few keywords, ta-da! An image generated by artificial intelligence (AI) is created.
The lightness of it all unsettles 2D designer Lily Liu a little. She works for a game studio where her daily routine consists of designing user interfaces and creating artwork for games on mobile devices. Recently, her boss directed her to study artificial intelligence art generators.
A week later, she came to the grim conclusion that her current job might one day become obsolete since it would be largely subordinate to a computer program.
“I have to admit that such AI art generators are very, very powerful,” said Liu. “They make some of the drawing work so much easier that I don’t think any company will be able to resist them. You save time and manpower. With them it is possible for one artist to do the work of three or four.”
Liu tried all the popular artificial intelligence art generators available, including Midjourney, Stable Diffusion, and Dall-E 2. She tried different styles of works, including impressionistic, realistic, and Japanese animation.
Courtesy of Lily Liu
Using the same keywords, three different AI art generators returned wildly different results.
“In my experience, these art generators are good at creating background images that focus on overall atmosphere and light and shadow effects, such as: Such as wild landscapes and cityscapes,” said Liu. “Sometimes such images don’t even need to be modified. I assume they could be widely used in fields like graphic design and novel illustration.”
An image drawn by an artificial intelligence generator. The key words of the picture are: city at night, top view and blue lights.
But the machine artists are not perfect. They are still not “intelligent” enough to generate a specific Japanese anime-style object, logo or character ready to use.
“Because generated images have only one layer, it is difficult to make changes to them,” Liu said. “They aren’t helpful when it comes to high quality requirements, like drawing animation characters.”
AI generated animation style characters are not normally allowed to be used in professional works.
Nevertheless, artificial intelligence is making faster progress than many expected.
“I believe that one day AI art generators will overcome all disadvantages and artists, whether they like it or not, have to accept this fact,” she said.
Machine-made art drew a lot of attention in 2018 when the painting Portrait of Edmond de Belamy, created by a French team, fetched US$432,000 in the UK. The image was generated from data from 15,000 portraits taken from the 14th to the 20th century.
“Portrait of Edmond de Belamy” was the first AI artwork to be sold at auction.
Last month, Jason Allen won a Digitally Assisted Art Award at the Colorado State Fair Art Competition in the US with his entry “Théâtre D’Opéra Spatial.” But it turns out the work was more than just “assisted” by technology. It was created by him. After the award ceremony, it was revealed that Allen had created the artwork using the Midjourney AI art generator.
Artists and others reacted with outrage, calling his work a “fraud”.
“Théâtre D’Opéra Spatial”, a painting by Midjourney that won a digital painting competition in the United States
It’s an outcry heard around the world from those who believe machine art is nothing more than high-tech plagiarism. Developers feed millions of existing artworks into a machine so it can “learn” the relationship between text and images. The original artists are usually not paid to use their work.
Earlier this month, South Korean illustrator Kim Jung-Gi, known for his large, highly detailed illustrations and ability to draw from memory, died of a heart attack in France.
Just a day after the news broke, a Twitter user, screennamed 5You, claimed he had “trained” an artificial intelligence model using Kim’s work and shared it online as a “tribute” to Kim.
“I’m pretty happy with the results considering how intricate his style is,” he wrote. “I would prefer, yes, you would give me some credit for using it, because creating a file like this is hard work, and I give it to you.”
He soon faced angry backlash from netizens, who denounced his efforts as an inappropriate way to pay tribute to Kim.
“Art generated by artificial intelligence is not your work and falls under the public domain,” wrote a netizen using the alias Kindrick. “Not only do people not have to and shouldn’t credit you for this artwork that you didn’t create, but using artificial intelligence to replicate Kim Jung-Gi’s artwork is an insult to his work.”
Above: an authentic artwork by the late South Korean illustrator Kim Jung-Gi. Below: an image drawn by a computer trained from Kim’s works.
where do you draw the line
When a South Korean artist livestreamed a painting online this week, a viewer took a screenshot of the unfinished work and fed it an AI art generator. He then posted the resulting image on Twitter before the artist could finish his work.
The incident sparked another wave of condemnation.
As a South Korean artist was livestreaming painting (right), a viewer took a screenshot of the unfinished work and created an AI art based on it (left). The incident drew a lot of criticism on Twitter.
So are these machine artists art thieves?
The jury is still out on this question.
Lawyer Wang Renxian said on his official WeChat account that if an artificial intelligence-generated image looks very similar to an existing work of art, it could be defined as copyright infringement.
Although no existing cases on such art disputes are available for reference, cases involving AI texts have been settled in court.
In 2019, tech giant Tencent filed a lawsuit against Shanghai Yingxun Technology Co, accusing the company of stealing an original item created by Tencent.
The article was written by “Dreamwriter”, an artificial intelligence writing program developed by Tencent and published on a Tencent official website. Yingxun then pasted the same article on his own website.
The Nanshan District People’s Court in Shenzhen ruled that the original Tencent article was classified as a “literary achievement” and thus protected by copyright.
“From this case, we can see that artificial intelligence works appear to still be protected by law,” Wang wrote. “It should be noted that AI art generators are ultimately a means of production and are therefore still regulated by current legislation. We advise caution when using images generated by artificial intelligence.”
