's dream of "throwing in a novel and coming out as a blockbuster" has never been closer to reality.
Recently, openai, an American artificial intelligence research company, released its first Vincent video model, sora, which can use text commands to generate a 1-minute high-definition video, which immediately attracted widespread attention and shock from many parties - the video generated through sora has a smooth and stable picture, without any Shake and distort, object characteristics are coherent and will not change abruptly. It can maintain the consistency of multiple lenses, simulate physical changes in the natural world, and has a 3D mirror function.
Although the "60 Seconds" that shocked the world also had many mistakes, the effect has made Musk lament that "human beings admit defeat", and many people even called "the director and the later stage will be unemployed." After all, the advent of this model, known as a "filmmaking artifact," means that perhaps every ordinary person who uses it can realize the "director's dream." Many people are already thinking about using chatgpt to generate creative copy and scripts, and then using sora to generate cool few-minute advertising blockbusters from the text. How much impact will the advent of
sora have on the video industry? Will it have a "disruptive impact"? Will the jobs of film and television production practitioners gradually be replaced by them? Recently, Red Star News talked with a number of industry professionals and scholars. When
sora comes out, will the special effects company die?
Practitioners: It is a bit difficult to "liberate" film and television post-production. As soon as
sora came out, many people believed that the film and television industry would be the first to be affected, especially the film and television post-production.
According to China Business News, after learning about the release of Sora’s video, Yu Gang, co-founder of Time Matrix, said that his mentality was “ice and fire”: On the one hand, his first reaction was “Is the special effects company going to die?” He really wanted to "bury" the special effects tools he had learned in the past; but from another perspective, he was very happy. He felt that the most expensive step of dynamic pre-rehearsing, which included shooting videos, making special effects, and putting them on the screen, was completely possible. Realize cost savings through AI.
However, regarding Yu Gang’s second point of view, Mr. Zhou, a film and television post-production practitioner, is conservatively optimistic. He told the Red Star News reporter that it is difficult to "liberate" the post-production period of film and television with Sora: "Because people are social beings, they have a lot of creativity and ideas, and they have to face thousands of demands."
Mr. Zhou tends to think , sora will become an auxiliary tool in the future. At the same time, he also questioned that the generated video is not like the text or image-generated works, and it is not convenient to manually complete the "last mile" reprocessing.
Mr. Zhou further explained that the film and television post-production industry is not a "mechanized operation" like many people's stereotypes. It is not enough to only fulfill Party A's needs when designing, but also to guide Party A's thinking and add his own ideas. " Whether design software or Sora, they are just tools; advancements in hardware and software only improve efficiency. "
" Post-production is a large area, involving shooting, editing, packaging, color grading, and 3D. These tasks are also It is difficult for a program to understand such a huge amount of knowledge clearly if it can be broken down into subdivisions,” Mr. Zhou said. Can
sora have a "subversive" impact?
expert: It is a gradual breakthrough, and many links are difficult to replace.
In fact, Vincent Video’s large model is not a new track, and openai is not the only pioneer in the track. In June 2023, Runway, an artificial intelligence startup supported by Google, released the large model gen-2, which can generate movie-level videos of a few seconds long; in November of the same year, the animation video generation company Pika launched a model that can generate minute-level high-quality animation videos. product.
Despite this, sora can quickly produce high-quality videos up to one minute long, making "subversive technology" and "subversive industry" become hotly discussed keywords in the industry. A research report recently released by Guosheng Securities believes that compared with other previous Wensheng video models, sora has leapfrogged into a practical productivity tool. The 1-minute length is expected to be widely used in the field of short videos, and the ability to expand videos is also expected to produce long videos.
Du Xiaomeng, assistant researcher at the National Development Research Institute of Peking University and deputy dean of Bimba Business School, told the Red Star News reporter that compared with the previous Vincent video large models Pika and Runway, sora has indeed greatly improved the effect, "This breakthrough Mainly reflected in the senses, it is much more realistic than before, similar to chatgpt's leap compared to some previous artificial intelligence voice assistants."
But she also pointed out that, like chatgpt's breakthrough in text understanding and generation, sora is technically It is a gradual breakthrough, not particularly revolutionary, and does not change the nature of productivity and production relations.
Du Xiaomeng believes that this is a tool and efficiency improvement. It does not need to be done frame by frame, but it does not mean that people are not needed. "That is, people who can use this tool may be more efficient than people who cannot use this tool, (but in essence) it is still a replacement of people for people, not a replacement of tools for people."
And substitution is not necessarily is the only result. Du Xiaomeng said that the improvement of efficiency will bring about two options: the compression of personnel, or the expansion of workload. "If you choose to reduce costs, then the work that used to be done by 10 people can now be done by 2 people. But what I see more often is that because of using this tool, I can take on more work. So. I think companies that use this tool will kill those that don’t.”
In addition, Du Xiaomeng also pointed out that there are still many links in the video industry that cannot be replaced, such as creativity.
Well-known screenwriter predicts:
In the future, AI may replace "70-80% creation"
"This is what we want. Especially for text workers like me, videos can be generated directly from text." Wang Hailin, a well-known screenwriter, said in In an interview with a reporter from Red Star News, he said that in his opinion, sora does not involve challenges to screenwriters for the time being, but it does have a greater impact on directors and actors responsible for producing pictures.
Regarding the sudden emergence of sora, Wang Hailin boldly predicts that in the future, artificial intelligence will be able to replace 90% of directors and filming teams, as well as 70-80% of their creative and less creative work. "What can be replaced are those creations that can score 70 or 80 points. Those that are particularly high cannot be replaced, and there is no need to replace those that are particularly low."
Despite this, in Wang Hailin's view, artificial photography has its humanistic value, which may be better reflected The superior spiritual attributes are just like after machine products have taken over the market, there are still handmade products that are highly respected. He guessed that in the future, film and television works produced entirely by humans will become a niche market with a narrow audience, just like small theater plays that are only watched by a few dozen people at a time.
Wang Hailin prefers to use the word "auxiliary" rather than "replacement". He believes that the future is more likely to be a combination of artificial intelligence and manual filming. It is just a matter of proportion and the quality of the combination. CG technology has already been widely used.
Wang Hailin believes that the use of artificial intelligence can greatly improve efficiency and even have some unexpected effects. The industry should not reject it, but it should not be completely dependent either, giving up its own initiative in thinking and creating, and expecting machines to do things that humans cannot.
Wang Hailin is cautious about whether the voice of screenwriters will be relatively improved in the film and television industry. Wang Hailin is cautious and believes that it needs to be observed because the change in the power structure involves more than just technical issues.
And the right to speak naturally involves Internet platform companies. Some analysts believe that artificial intelligence technology largely comes from platform companies, and platform companies also like to use this technology. But as an industry practitioner, Wang Hailin is wary that platforms are more inclined to use technology to replace practitioners rather than assist practitioners, because the Internet has used the "B replaces A" model in previously disrupted industries. In the film and television industry, I’ve also been using this model through trial and error for many years. In Wang Hailin's view, replacement thinking may not necessarily succeed, and integration is the right way.
According to Guan Yadi, a senior filmmaker and host of the video podcast "Open Dialogue", advanced digital tools in the past have been applied to budgeting, scene construction, and cross-department collaborative rehearsals, which have greatly improved productivity, but "this time it seems that the situation has changed again." It’s not the same.” At present, Sora’s image quality can be applied to the generation of cutscenes in theater-level movies, replacing shooting and production.
Guan Yadi told the Red Star News reporter that he believes that front-line practitioners in the film industry will not panic because the core charm of movies is actually the exchange of culture and emotions, which cannot be replaced by humans. At least the current artificial intelligence does not have this ability. But before AI develops the ability to drive the world, humans and AI can coexist together.
Who will be the first batch of "professional players"
Industry prediction: Part-time self-media people may be the first to use
sora is currently characterized as an initial research result and is not for public use. OpenAI said that due to concerns about the misuse of deepfake videos, only a limited number of visual artists, designers and filmmakers now have internal trial opportunities.
So, if sora is "fully opened" one day, who will be the first batch of "professional players"?
According to Du Xiaomeng, some people in the video industry still tend to use the previous tools and previous methodologies, just like some programmers who still write code line by line by themselves. Considering this inertia of use, Du Xiaomeng predicts that some people or amateurs who have not been exposed to video editing before may first use tools such as sora.
At the same time, Du Xiaomeng pointed out that computing power is still very expensive at present, and Vincent Video will not be cheap for the time being. Professional companies will weigh the cost-effectiveness, so small companies and amateurs who do not use that much may apply it first, such as part-time self-media people. ——It can improve efficiency, and the requirements for video are not as high as those of large professional companies. "We-media people may think that Sora makes their videos better, but professional animation companies or film and television companies may still think that it is unsatisfactory and just a toy."
A research report released by Zheshang Securities is consistent with Du Xiaomeng's point of view . The research report believes that in the short term, sora and similar products can significantly improve the efficiency of image and short video production, change the creative production and marketing workflow, and improve the productivity of short video products. For long-form videos and games with more complex formats, the models are still unable to accurately understand causal relationships and other technical difficulties. At this stage, the main focus is to provide art inspiration support.
In the medium and long term, Zheshang Securities stated that sora and similar products will participate in the process of changing the two major links of information production and distribution. PGC (professional production content) will widely use AI tools to assist production, and UGC (user-generated content) PGC will be gradually replaced with the help of AI tools. During this time, the commercialization of AI-generated video tools will accelerate.
A short video practitioner also believed in an interview with Red Star News that head short video creators already have a large professional team. The role of sora is not particularly big at present. Instead, it depends on the writing ability of waist short video creators. The editing and directing ability and video presentation ability may not be that strong, so sora can become a powerful toolbox to create difficult pictures and improve the exquisiteness and completeness of the pictures.
"Replacement does not come so fast"
's widespread implementation is more than a technical issue
"Sora can be said to be another chatgpt moment." An industry expert commented.
As a screenwriter, Wang Hailin said frankly that the screenwriting industry is indeed facing challenges from chatgpt. In fact, last year’s super general strike by Hollywood actors and screenwriters in the United States targeted artificial intelligence. Especially for screenwriters below the mid- to low-level level, they protest that their unit labor value will be reduced to the extreme.
But Wang Hailin also pointed out that after the emergence of new technologies such as chatgpt, many creative workers are afraid of their jobs or being replaced; but after a period of time, it seems that it is not that fast yet, and those who are replaced are often low-level, simple of labor. “I originally thought it would be very soon, but after a few years, some of the things I worried about at the time have not materialized."
A film and television producer pointed out to the Red Star News reporter that the emergence of CG technology once panicked the film and television industry. It has also become a contributor to visual effects blockbusters, but everyone has found that it cannot replace the effects of real shots. At present, film and television still rely on real shots. Mainly shooting, only using CG when it is impossible to actually present it or to save money. Some directors advocate complete real-life shooting without using CG. Generally speaking, he believes that for producers, the emergence of sora is a good thing.
Taking the threat to the art industry from the emergence of high-quality image generation engines as an example, Mr. Zhou also pointed out that the current demand for artists is still huge, and painters have not lost their jobs. There are still original painters, and there are some 2D animators. Unemployed, but "if the east is not bright, the west is bright", three-dimensional efficiency is higher, and the effect is better.
Wang Hailin said that technical problems are never just purely technical problems. At present, the problems of laws, regulations and industry support have not yet been solved. This is also the problem that Du Xiaomeng believes that sora needs to be solved to achieve real commercial application - a single technology requires the support of a series of technologies.
Du Xiaomeng believes that at present, sora can be implemented in limited scenarios like chatgpt. In her view, the metaverse simulation scene is very promising, but whether this scene can have a relatively large development depends on, in addition to itself, breakthroughs in hardware technologies such as VR and MR related technologies, headsets, and glasses. In combination, a single technology has limitations; if related technologies produce a better synergy, in addition to the metaverse scene, the game and pan-video industries will bring about greater breakthroughs, including advertising, media, film and television, and self-media. .
Red Star News reporter Hu Yiwen
Editor Li Binbin
