Will AI kill “Visual China”?

Welcome to the WeChat subscription number of “Sina Technology”: techsina

Text/Wang Yi

Source: Entrepreneurship State

There is no doubt that Visual China has a bad reputation.

China’s largest online copyright trading platform for visual content, representing GettyImages, Agence France-Presse, Reuters, and BBC’s photo copyrights in China, acquired the world’s third largest photo library Corbis Images founded by Bill Gates, and owned 13 million images. 500px, a photography online community for registered photographers, is ridiculed that its core business model is litigation.

According to data from Tianyancha, there are over 10,000 disputes involving the main company of Vision China and its two subsidiaries, ranging from Tencent Ali to beauty salons in 18th-tier cities that have never heard of the name, all of which have received lawyers from Vision China. letter.

In July 2018, Zhang Ying, a partner of Jingwei China, posted on Weibo, “Visual China, it is said that it has developed a system since the year before last, and began to systematically search for various companies that use their pictures without authorization and negligent use. Then they ask for huge amounts of compensation. Usually, a small negligence does not accept deletion of a picture, and directly asks for hundreds of thousands of yuan in compensation, and coerces the company to sign an annual contract. Infringement really shouldn’t be, and it has become the core business form of this company now, which is also funny. I don’t believe that such a business model of blackmail can continue and be maintained. Just wait, one day…”

4 years later, that day may be coming soon.

With the public beta of DALL·E2, an artificial intelligence online drawing application under Open AI, and the open source release of Stable Diffusion at the end of August, the technology and art circles have set off a wave of AI painting, and various netizens use these tools to generate various styles. Different works.

Will the AI ​​painting tool platform replace the gallery website represented by Visual China?

It’s entirely possible.

The business model is to sell “insurance”

Of course, the business model with litigation as the core is just a joke. The gallery website represented by Visual China is more like selling “insurance” to users.

Users pay the gallery website and get the copyrighted pictures, so they can focus more on the business.

But such “insurance”. Not always safe.

In April 2019, the first black hole photo in human history was released. But then some netizens found that the black hole photo was included in Visual China’s “edited pictures”, and Visual China stated that if it is used for commercial use, its customer representative should be contacted.

But in fact, the black hole picture belongs to the “Attribution 4.0” license (Creative Commons Attribution 4.0 International License, CC 4.0 for short), and Visual China has no right to license it to a third party.

Afterwards, the Central Committee of the Communist Youth League questioned Visual China on Weibo, is the copyright of the national flag and national emblem also owned by your company?

Major companies have also expressed that their company’s LOGO has been included in Visual China.

The real problem with the “selling insurance” business model is not that insurance is sometimes unavailable, but that it is too expensive for SMEs.

In an interview with Hedgehog Commune, Chai Jijun, co-founder and editor-in-chief of Vision China, said that in the past, Vision China’s customers were all so-called “professional users”, professional media agencies, creative design companies, etc. There are thousands of users, and now there are a large number of WeChat public accounts of nearly 30 million, which may be hundreds of thousands or millions of users. The user base has greatly increased, and the picture market has grown from a professional niche market to more than ten billion. market. Our company speaks of “the pond becomes the sea”.

For large organizations such as media and advertising companies, they use a large number of pictures every year, and spending money on “insurance” is a necessary means of hedging. “Insurance” is undoubtedly too expensive.

Currently, the pricing of gallery sites is very high. The price of a picture on Visual China ranges from several hundred to several thousand yuan, and a picture on TuCong.com costs hundreds of yuan. As for the annual frame agreement, it is often tens of thousands to hundreds of thousands of yuan.

Such a price is a necessary cost for large institutions; for small and medium-sized companies, it is a burden that cannot be afforded at all.

AI pictures are copyright-free and risk-free

There are legal risks in using copyright-free images, and using copyrighted images is too expensive. How can we use copyright-free images without legal risks?

Images generated with AI.

Regarding the copyright issue of AI, the team of lawyers Xiao Sa of Beijing Dacheng Law Firm told Chuangyebang that there are only two kinds of authors recognized by my country’s Copyright Law: natural persons, legal persons or unincorporated organizations. AI is not recognized by my country’s Copyright Law. author”.

In January 2021, the U.S. Copyright Office issued a document entitled “CopyrightableAuthorship: What Can Be Registered?” ” guidance document, in the document, they make the following provisions on whether a work has copyright:

Among them, a very important one is “Was the work created by a human author?”

The U.S. Copyright Office believes that “human authorship” is essential to identifying a work, and if the work is purely the product of AI and algorithms, it is difficult to identify as a “work” recognized under U.S. law.

“That is to say, the US Copyright Office believes that it is meaningless to claim copyright for works generated by AI.” The team of lawyers Xiao Sa explained, “AI is a creative tool, and the works created by using it are to some extent a kind of Things that have entered the public domain since they were created, provided that there is no separate contract with the AI ​​technology company or algorithm designer (for example, there are special clauses in the user agreement of an AI-generated original, indicating that the user uses his AI to create works) The copyright belongs to the company), and anyone can make fair use, including commercial use.”

As for how to determine whether an AI painting work constitutes infringement, Xiao Sa’s lawyer team believes that due to the extremely high randomness and uncertainty of AI-generated works, the possibility of infringement of their works is unlikely. Even if it constitutes infringement, in judicial practice It is also difficult for the owner of the Zhongyuan work to fulfill the burden of proof.

That is to say, the pictures produced by AI have no copyright, and anyone can use them freely. AI will not issue a lawyer’s letter, and there is no legal risk.

At present, platforms such as Midjourney and Open AI have clearly stated on their websites that users have the ownership of the works they generate, including the rights to reprint and sell.

The future has come

Although there is a real demand, there are still many problems facing the large-scale popularization of AI painting works, such as the stability and usability of the image quality.

Ran Nuochen, creative content director of “Zanku Hailuo”, a genuine visual content trading platform under Zoku.com, told Chuangyebang, “From the current stage, there are still many flaws in AI painting works. The abstract and freehand works generated by AI It may be better, such as the styles of Monet and Van Gogh, but when it comes to styles such as photos that require extremely high precision, AI creation is prone to problems, such as the wrong structure of the facial features of the human face, or the lower right corner. There will be icons with unknown shapes, these are because AI does not have such a strong resolution when learning, and will recognize the watermark on some training material pictures as the content of the picture itself.”

Will AI paintings have an impact on gallery websites that rely on copyright?

In this regard, Ran Nuochen believes that the impact will inevitably exist. Once our country’s law clarifies the copyright ownership of AI paintings, and the works generated by AI become more and more exquisite, there will inevitably be many people who choose to generate the works they want instead of Buying copyrighted images from a gallery site will cause a heavy blow to the gallery site’s business.

But at the same time, Ran Nuochen believes that AI painting will not completely replace designers and illustrators, but will become an “auxiliary tool” for these art creators to assist them in their artistic creation.

“Many creators have embedded AI drawing tools into their workflows, and in the process of making sketches, they use AI to assist themselves in realizing their creative ideas,” said Ran Nuochen, “AI can very well convey the creator’s ideas. Visualization, secondary creation based on the images given by AI, can actually produce better works faster. Therefore, AI is likely not our enemy, but an auxiliary tool for us, not a competition with us , but a cooperative relationship.”

At present, there are already platforms that “snap the wind” and launch their own AI-aided design tools. The online collaborative design platform “Instant Design” is a typical case.

On September 15th, Instant Design launched its own AI painting plug-in “Instant AI”. On the basis of the Stable Diffusion model, it has made more in-depth optimizations to the function of “Mat Map (the AI ​​is generated based on the original image after the user uploads the picture)” . Since its launch for half a month, nearly 30,000 users have used its products, and the favorable rate is as high as 80%.

In order to experience the effect of the “Instant AI” plug-in, we uploaded a template image of the tunnel, entered the word “time tunnel” in the document description area, and just waited for a few seconds, the AI ​​generated the following image for us:

It can be seen that real-time AI can already generate works with perfect clarity and imagination. Although the details still need to be optimized, it can greatly improve the work efficiency and productivity of designers.

When it comes to “whether AI drawing tools can replace gallery websites”, Zhou Ning, co-founder of Instant Design with many years of technical background, believes that this is something that will happen with high probability.

At present, AI painting tools represented by Midjourney, Stable Diffusion, etc. have crossed the threshold of “face deformity” to generate highly simulated portrait pictures of people. For landscape photos that require high picture realism, AI painting tools can also be easily generated.

With the continuous iteration of AI painting tools and underlying models, we have reason to believe that there may be such a situation in the future: we no longer need to buy pictures from gallery websites, but only need to pay 10 pieces of AI painting tools per month. Money, enter a piece of text, and you can get the picture you want in seconds.

At that time, all public account illustrations, website illustrations, packaging design drawings, and children’s picture book drawings could all be done with an AI tool, and there was no need to go to Visual China to buy pictures.

Of course, no one will send you a lawyer’s letter.

(Disclaimer: This article only represents the author’s point of view and does not represent the position of Sina.com.)

This article is reproduced from: http://finance.sina.com.cn/tech/csj/2022-10-12/doc-imqmmthc0608358.shtml
This site is for inclusion only, and the copyright belongs to the original author.