2024 At the beginning of the new year, a new word has become popular in the film and television circles, technology circles, and capital circles—sora. On February 16th, Beijing time, the artificial intelligence company Openai launched a new model sora that can instantly generate short videos based on text instructions. At this point, it’s just over a year since the company released chatbot model chatgpt.
In the demonstration video produced by openai's latest product sora, a woman wearing sunglasses and a leather jacket and red skirt is walking on the streets of Japan. She looks back from time to time, with a mysterious temperament. The neon street scene in the distance, the reflection of the water on the ground nearby, and the condition of the heroine's skin in the close-up are all clearly visible.
In the traditional film and television industry, shooting and producing such a video requires the creation of storyboards, site selection, service, lighting, filming, live-action interpretation, and post-editing... But through the application of sora, all this only requires The fact that it can be achieved with a single text command is undoubtedly exciting for people from all walks of life. Zhou Hongyi, founder of
360, expressed his opinion on social networks, saying, "The birth of Sora means that the realization of AGI (general artificial intelligence) may be shortened from 10 years to one or two years." Some people therefore assert that Sora will change our judgment and understanding of the world. From now on, "seeing is not necessarily believing" and "the real world will no longer exist"; the shocked film and television people will inevitably fall into "unemployment anxiety" and self-doubt: "Will our jobs eventually be replaced by AI?" Some people are also worried that with the promotion and application of Sora, the threshold for video fraud will be significantly lowered, and a series of moral, ethical and legal supervision issues will arise... How "magical" is
Sora? What kind of benefits or impact will it bring to the film, television and entertainment industries? What mentality should we adopt to welcome the advent of the AI era? Nandu reporters recently visited a number of practitioners to discuss the impact of sora on the film and television entertainment industry. What is the strength of
1
sora?
The name
sora, a "world simulator" that can create 60-second videos, comes from the Japanese word "empty" (そら sora), which means sky, to show its unlimited creative potential. "This thing is very imaginative. As practitioners, we have very complicated emotions." Director Wei Qi, co-founder of Virtual Pictures, lamented to reporters.
According to the official introduction, sora can create 60-second realistic videos based on user text prompts. Not only does the image presented be detailed and vivid, but more importantly, it can understand how objects exist in the physical world, thereby deeply simulating reality. Physical world, generating complex scenes with multiple characters and specific movements. Therefore, sora is also known as the "world simulator".
In fact, before sora, there were already many products that can generate high-definition videos through text or pictures. The more well-known products include runway, free pika, which are already commercially available, as well as google lumiere and meta make-a-video, which are still in the improvement stage. etc. Compared with these previous products, what are the highlights and strengths of sora? Why did it explode on social networks as soon as it was released? Video case launched by
sora.
Liu Jun, vice president of Unilumin Group, summarized the three characteristics of the sora model in an interview with Nandu reporters. "The first is that the duration of the video it can generate is relatively long; the second is that its simulation capabilities are very strong. It can not only simulate dynamic visual effects, but also capture deep-seated aspects that are consistent with our daily life experience. Interactive mode. For example, in the video of 'Woman Walking on the Street' released by Sora this time, even the reflection of the water on the road after it rained (very accurate), including the height and overall height of the woman. The contrast relationship of spatial structure, etc. (all accurate). So it can actually simulate this complex physical space. The third is that in terms of speech understanding and video generation, it has a long text parsing technology, which can be based on User text can be analyzed. It can also accept us uploading some dynamic images. For example, if I want to make some extensions on existing videos, the content it adds will be close to the style of your original video."
Of course, according to the official Introduction, sora still has some "flaws".For example, because its model does not rely on an internal physical simulation engine but relies on large-scale data drive, there will be places in the videos it generates that do not conform to real physical laws. This problem is currently difficult to eradicate.
2
may replace traditional tools and "tool people", leaving practitioners "surprised and anxious"
sora has made huge breakthroughs in video duration, imaging quality, analysis and simulation capabilities. According to predictions by International Data Corporation, it will be the first to be applied in media fields such as short videos, advertising, interactive entertainment, and film and television production. So, can the sudden emergence of Sora replace video and film and television workers? In which positions will employees be affected and face an "unemployment crisis"? In an interview with a reporter from Nandu, industry insiders said that Sora is likely to replace traditional CG tools and related low-tech jobs, greatly improving production efficiency in video previews, basic editing, and secondary processing and creation of existing materials. and quality.
Liu Jun revealed that UniMing Technology has obtained Microsoft Independent Software Vendor (ISV) certification and obtained the official OpenAI access license. However, the company has not yet tested Sora and can only speculate on the possibility through officially disclosed information. He said: "The first impression is that AI is progressing very quickly. If we give it enough time to improve, it can really replace some of the current creative tools and the work of some basic 'tool people' For example, video previews, such as in the industrial field, medical field, etc., we need to use a lot of video content for teaching. You only need to enter the requirements, sora can simulate it. In this way, it can replace a large number of traditional CG-related positions, and What it outputs will be better."
Liu Jun said frankly, "Once the AI model is exposed to a large amount of data, it can continue to learn and self-fission, and its upper limit is immeasurable. We should be very happy with this result. Surprised, but also quite anxious. Surprised because the application of AI in certain fields can indeed save manpower and is both fast and efficient." Not only does it greatly improve production efficiency, Sora can also lower the production threshold and make video creation more efficient. Popularity and convenience. But on the other hand, it will indeed have an impact on traditional tools and low-tech jobs, causing some people to lose their jobs.
Regarding this kind of technical anxiety, Wei Qi, co-founder and director of Virtual Film Industry, said that practitioners should continue to learn and improve themselves: "As the same saying, we maintain our imagination, but be prepared for everything. We must continue to learn, If the old technology is not innovated, even if it is not Sora but another new technology, (we) will be eliminated sooner or later." Liu Shuangjian, director of Virtual Film Industry, also showed a positive attitude in an interview with Nandu reporters: "Since AI It is a tool, and it naturally needs people who use it. So what we should think about is how to use it and make it a better creative tool."
3
"AI is just an auxiliary tool and cannot replace creative talents"
in While marveling at the powerful functions of sora, practitioners also clearly realize that as an auxiliary tool, it also has limitations in creation. Especially in terms of innovative thinking in film and television works and video scripts, humans are still irreplaceable.
"AI can only assist everyone in creation, it cannot replace our creative talents." Liu Jun took the production of the movie "Walking Alone on the Moon" as an example, "You can let AI generate 'a kangaroo walking in a space capsule' material, but what exactly does the image of a kangaroo look like? How tall? How strong? Is it cute or strong? It is difficult to design a specific image and style. It still requires the creative ideas of the director, art and other creative personnel , to outline the outline of the kangaroo, and then use AI tools to generate it."
Liu Jun said that AI has a large material library, and its role is to help creators make secondary edits based on existing materials, but It’s hard to “make something out of nothing.” “If the creator wants to find some existing materials for secondary creation, AI can improve his creative efficiency, and it can do the creative execution work.However, the generation of creativity and creative ideas are still inseparable from our human subjective initiative. "
For short videos, compared to later editing technology and visual optimization, a novel and interesting script idea and an idea that can hit the emotions of the audience are the "soul" of creation, which is what sora is not. has the ability. In the face of the creation of longer and more emotionally complex film and television dramas, Wensheng video tools such as sora appear to be more "weak and weak". In addition to producing scripts, creativity is required, as well as the laying out and narrative of a complete story line. Rhythm control, atmosphere enhancement, character creation, emotional expression... these complex processes are far beyond the capabilities of current AI technology. Liu Jun mentioned: "There are so many scenes and scenes in movies and TV series. For the story line, AI may be able to generate pieces of material, but it is currently difficult to string together the entire film, and the style and tonality of the scenes are also random and may not necessarily maintain coherence. "
When the craze for new concepts subsides and returns to calmness and rationality, practitioners have to return to the discussion of the most essential issues - how to improve creative capabilities? How to tell a story well? Whether it is sora or any other high-tech , currently cannot replace creators with innovative thinking and profound expression skills.
Written by: Nandu reporter Zhu Wenyi Yu Xiaoyu
