Image source: Generated by Unbounded AI

Shi Yuzhu did not break his promise.

Recently, Giant Network released the "Qianying QianYing" large audio game generation model, including the large game video generation model YingGame and the video dubbing large model YingSound, achieving a new breakthrough in audio and interactive game video generation.

As early as January this year at the Giant Network annual meeting, Chairman Shi Yuzhu directly expressed his emphasis on AI technology. "If game companies do not actively embrace AI, they will be eliminated sooner or later." He also publicly stated , the company has promoted AI to its core business, advancing hand in hand with its gaming business.

In August, Cai Haoyu, the founder of MiHoYo, who has always been invisible, stated on social media that AI will subvert game development and suggested that most game developers quickly change careers. His new entrepreneurial projects are also related to AI. Game related.

In 2024, there will be a substantial breakthrough in the integration of AI and games. AI is no longer limited to a tool for optimizing processes, but is integrated into the game mechanism itself. Major game companies have launched comprehensive AIGC catch up.

In the first half of the year, AI games such as "One Thousand and One Nights" and "Daddy" emerged in droves. In early November, the first real-time generated AI game triggered a carnival in the industry.

The "2024 Game Industry Status Report" shows that 49% of game developers say that generative AI has been applied to their work scenarios.

In addition, Elon Musk also officially announced his entry into the game. He posted on the X platform: "Most game studios are controlled by large companies. My subsidiary xAI will establish an AI game studio. Make gaming great again.”

However, the integration of AI and games has not yet passed through the fog and landed on a precise destination. How to get involved in the subdivision of game development has become a new topic facing the game industry. In this regard, major game companies have come up with different solutions.

Reducing costs and increasing efficiency, AI plays its first card

After two years of development, AI and gaming are the best partners The awkward period has passed, and AI has been widely embedded in every aspect of game production. Among them, cost reduction and efficiency increase are the most direct impact after the integration of AI and games.

The "China Game Industry New Productivity Development Report" released by Gamma Data mentioned that the companies of 99% of the game industry practitioners surveyed have introduced AI technology and tools, and nearly 60% of the leading games Enterprises have built AI production pipelines to enable virtual content production or intelligent marketing.

Sanqi Interactive Entertainment has maturely applied generative AI technology to 2D painting, intelligent customer service, localized translation, copywriting production, collaborative office and other modules.

According to Wang Chuanpeng, vice president of Sanqi Interactive Entertainment, the intelligent layout has achieved corresponding results. For example, the application of 2D pictures has improved efficiency by 60%-80%, with 35,000 pictures per week. Pictures in the field of distribution and research and developmentHas been very widely used.

“In terms of translation, AI technology plays a big role, controlling the modification rate to around 10%, which greatly saves the team’s translation costs. In addition, the company’s global products are all connected With the help of intelligent customer service, the problem resolution rate has also reached a certain level."

Lin Rui, director of Shengqu Game Art Center, believes that AI can not only save manpower, time and capital costs, but also help the team in their work. More new ideas will be found during the execution process, empowering the industry to achieve new development.

“From original painting design to concept creation, to promotion, character drawing and other aspects, AI tools have been deeply integrated into Shengqu’s workflow, realizing a comprehensive replacement of the traditional development process, and It has been deeply embedded in the production process, significantly improving work efficiency, especially in the mass production stage, the efficiency improvement can reach about 60% to 80%," Lin Rui said.

Picture source: Tuchong

Shengqu Games has established an AIGC team since 2022, with titles such as "Legend", "Dragon Nest", "Rainbow Island" and "Hungry Wilderness". Well-known IP products have been integrated with AI tools, which has achieved remarkable results in cost reduction and quality improvement. In addition, Shengqu has also improved the entire process of AI character creation, launched the digital human customer service "Xiaoqu", and built an art assets and AIGC creation management platform.

Kaying Network deeply integrates large model technology with all aspects of game development, covering game animation, map design, technology development and other aspects, making AI technology the core driving force of development.

The "Xingyi" large-scale model application it launched not only promoted the automation and collaboration of game research and development, but also provided a new way to improve efficiency for the game industry and reshaped the thinking and process of game production.

NetEase Games chooses to enhance the user experience in content creation. "Egg Boy Party" uses the "UGC+AI" creation model to lower the threshold for players to create content. "Ni Shui Han" uses related technologies from NetEase Fuxi Lab to launch a "crew mode" in the game. Players can use their wild imagination and game scenes plus AI input to generate actors, actions, and camera movements to create videos.

“AI has its place in the entire chain of the game industry. It is not simply about reducing costs and increasing efficiency, but it allows game R&D and operation personnel to free their hands and have more time to stimulate creativity. After all, games In addition to technical background, it is more of a world of creative dimensionality. Therefore, AI is not a hot topic for the game industry, but is similar to a technological revolution," Zhang Shule, a game industry analyst, told Times Finance.

Get rid of tool attributes and explore advanced versions of AI games

When talking about the combination of games and AI, cost reduction and efficiency increase are still the main arguments. However, how to mobilize the more powerful capabilities of AI and explore high-level development and production links has become an area where game companies need to take root in the long term.

“Emphasizing AI’s cost reduction, efficiency improvement and quality optimization, in factIt only emphasizes the tool attributes of AI; but in the coming tomorrow, we should pursue the value of AI advanced to more open and vivid scenarios such as game play, immersive experience and plot content, and emphasize AI as the future digital Content core drives the value of the engine. "Xie Fei, President of Century Huatong, said.

Take the real-time generated AI game Oasis as an example. Its presentation method is more like a single video generation model, which is equivalent to a series of interactions based on Sora.

“The experience of Oasis is far inferior to that of mature mobile games or stand-alone games. You may have been in the Colosseum a few minutes ago, and then switch to the Gobi Desert as soon as you turn around, lacking long-term memory support. "Li Feng (pseudonym), a senior game player, told Times Finance.

Photo source: Tuchong

A game development practitioner expressed a pessimistic attitude towards AI's participation in game creation." At this stage, AIGC's improvement in the game creation process is mainly focused on the texture of the picture, light and shadow, and scene details, but the effect on the fun of the game is not great. What we lack is a strong narrative and world view, and further polishing of game creativity and content. If a large number of AIGC parts that do not enhance the essential game experience become standard features of the game, it will make it difficult for more people to easily experience the joy of the game. ”

In fact, it is difficult for AI to be smoothly integrated into the game production process. Liu Bing, chief artist of Zulong Entertainment, shared on the forum the common problems that currently occur in AIGC applications: poor controllability of asset generation and improvement of efficiency At the same time, it is difficult to maintain high quality. AI cannot understand and execute the design intention, which also prevents subsequent steps from being implemented.

“The AI ​​scenario does not have much logic and is still in the stage of stacking materials. At first glance, it seems very complicated. Beautiful, but totally unbearable to look at. When the picture needs to show more detailed content, there will be obvious loopholes, such as summer and winter settings appearing at the same time. Li Feng told Times Finance that he believes that it is difficult for scenes built by AI to escape the eyes of experienced players.

Tan Kai, head of AI large models at Kaiying Network, believes that humans, large model technology, and vertical technology Ecological balance should be achieved, not who replaces who, but also the development of large model tools used in games. It is becoming more and more subdivided and becoming an auxiliary tool in every production link.

Zhang Shule said that for now, game scenarios that can better integrate AI are generally in two aspects. One is the research and development level. The voice and original painting generation has been used to a certain extent in game testing. Secondly, in the game experience, AI NPC has appeared, and it can assist users in generating original game content in UGC games.

“AI is now just a simple plug-in, almost like a toy. The reason why it is difficult to apply on a large scale lies in the entire game chain. The creative components and interactive content are relatively complex and extremely vertical, covering various digital technologies. AI still needs time to be refined. " Zhang Shule added.