Recently, NVIDIA announced on its website that it will offer a custom AI model foundry service, Avatar Cloud Engine (ACE) for Games, to develop AI models that run in the cloud and on PCs, which developers can use to build and deploy custom voice, dialogue and animation AI models in their software and games.
The impact of Nvidia's proposed AI model foundry service on the game industry is more reflected in the opportunity given to middle and waist companies to reduce costs and improve efficiency and develop new categories, said Ma Axin, a cultural consumption industry analyst at ECOSYS, to the Huaxia Times.
"For example, in ACE for Games, enterprises can choose all the NPC customization, voice recognition and model-generated real-time expressions, or they can customize and choose the features they need. As an example, the development of products such as MMO category is no longer limited to head enterprises, with good ideas and concepts, middle and waist enterprises can also rely on AI tools and services to create excellent and playable products." Ma A-Xin said.
Make NPC more intelligent
ACE for Games is a custom AI model foundry service that middleware, tool and game developers can use to build and deploy custom voice, dialogue and animation AI models, according to the report. It empowers non-player characters (NPCs) with smarter and evolving conversational skills, enabling them to answer players' questions with lifelike personalities.
"Generative AI has the potential to revolutionize the way players interact with game characters, significantly enhancing their in-game immersion, and NVIDIA, with its AI expertise and decades of experience working with game developers, is pioneering the adoption of generative AI in games," said John Spitzer, vice president of NVIDIA's Developer and Performance Technology Division. said John Spitzer, vice president, Developer and Performance Technologies.
In NVIDIA's view, the ACE for Games foundry service will help developers retool models for their games and then perform real-time inference via NVIDIA DGX Cloud, GeForce RTX PC or local deployments. These models are optimized for latency and help optimize the game's immersive interactive experience.
According to NVIDIA's website, numerous overseas manufacturers are already using NVIDIA ACE game development versions, including European game company GSC Game World's game Chernobyl Heart, independent developer Fall Leaf's Fort Solis, and virtual social company Charisma.ai is also leveraging its animation generation capabilities.
NVIDIA's introduction of custom AI models accelerates the integration of AI with the gaming industry. "We believe that with the launch of iterative upgrades of tool products incorporating AIGC functions and the rapid exploration of applications by game companies, the process of comprehensive AI empowerment for games is expected to accelerate." Huatai Securities research report shows that the application of AIGC is expected to change the way players interact with game characters and achieve a significant increase in game immersion.
Huatai Securities points out that the huge demand contained in the game market itself is a good application scenario for technologies such as AIGC, and the progress of related technologies will further enhance the product quality and comprehensive competitiveness of China's game industry, thus driving the long-term growth of the industry.
It has become the consensus of global game companies
As early as Feb. 15, NetEase (09999.HK) announced that its "Inverse Water Cold" hand game was loaded with the first "Game GPT", which allows NPCs to talk freely with players and give logical behavioral feedback based on the content of the conversation. The solution comes from NetEase's Fuxi Lab, which was established at the end of 2017, and is underpinned by a set of proprietary models that have reinforced and learned a lot of martial arts novels, history books and poems and songs.
The person in charge of the You network (002174.SZ) told the Huaxia Times that for the game industry, artificial intelligence will bring substantial improvement to the overall quality and R&D efficiency of the game. Among them, the threshold of game production will become lower and lower, the number of games will become more and more, and the competition in the whole industry will intensify. But at the same time, there are also more positive effects. Benefiting from more natural interaction, AI will really realize the game experience of thousands of people; the lowering of the development threshold will also promote the rise of independent game producer groups, especially those developers who master AI tools will have more advantages.
"From the development side, the overall R&D efficiency will be improved, while the R&D cost will be reduced, especially the output of art-side content, which can see a clear trend of reduction. From the user side, thanks to the powerful analysis and summarization ability of artificial intelligence, users can get a more personalized game experience." The person in charge of Yuyuan said.
According to his introduction, "In terms of Yuyu, we rely on the ability of AIGC, in terms of planning, art and program, we already have more content output, such as original painting design, image super-score (note: image super-resolution reconstruction), fine coloring, plot text generation, TTS dubbing, small language translation and other links have relevant practical applications. At the same time, we are also actively cooperating with other companies that have the ability to develop AI technology, and are expected to introduce AI big model technology in games in the future."
Kaiying Network (002517.SZ) also told the reporter of "Huaxia Times" that in the field of artificial intelligence, some of the company's strategic investment enterprises have also formed their own AIGC capabilities, and some of the war investment enterprises have self-researched AIGC tools under Unreal Engine, which have been able to form high-precision character and scene rendering maps in batch, greatly reducing the art costs in the field of research and development, and fully Applying the AI tool MidJourney to generate scenes and characters in large batches can effectively reduce the creation time.
The reporter of "Huaxia Times" noted that the application of AI has become the consensus of the global game industry. Recently, Activision Blizzard launched internal deployment of generative AI game development tools "Blizzard Diffusion", for in-game assets such as "World of Warcraft", "Diablo" and "Overwatch" trained image generator to reduce some of the design and development of tedious work.
Not long ago, European and American SLG major Paradox likewise made its embrace of AI clear in an earnings call, "We're using AI across all departments at Paradox, from marketing to development, from graphics to design, from game planning to everyone. We want to use AI to help us become a better game studio."
According to the calculation of Guotai Junan Securities, the actual sales revenue of the domestic game industry in 2022 is 265.9 billion yuan, and the expected R&D expenses account for 15%, assuming that the proportion of future AIGCable is 50%, then the market size of AIGC in the current game industry application can reach about 20 billion yuan. Thus, the integration with AIGC is expected to enhance the market value of the game industry since.
It is worth noting that although "AI+game" is all the rage, it also hides many unknown challenges while it brings new opportunities.
"The current game review mechanism actually limits the possibility of uncontrollable content. Existing games are numerically balanced models, and new game types and gameplay are needed to take full advantage of AI." Sun Hui, CEO of Fuzzy Jiujiu, who has been engaged in game development for more than 19 years, told the Huaxia Times.
An Guangyong, an expert of the Credit Management Committee of Allied M&A Guild, told the Huaxia Times reporter that, firstly, the development of AI requires a lot of data and arithmetic support, which may put forward higher requirements for game developers. Secondly, the game industry needs to solve the problem of data privacy and security to ensure that the application of AI models will not violate the rights and interests of users. In addition, AI algorithms and decision-making processes need to be transparent and explainable so that developers and users can understand and trust these systems.
"Currently, there are a number of bottlenecks in the development and application of AI in the gaming industry. One of them is the further improvement of algorithms to improve the performance and effectiveness of AI models. Another bottleneck is the quality and size of datasets, with more high-quality training data needed to support the training and validation of AI models. In addition, game developers need to address the technical challenges of model deployment and integration to ensure AI models can run efficiently in games." An Guangyong said.