Can AIGC recreate the next MiHoYo? probably let you down

Source: Jinghe

Author: Qian Hongyan

Image source: Generated by Unbounded AI tool

Medium-sized game manufacturers may be the most complicated group of people in this AIGC wave.

The efficiency improvement brought by AIGC tools makes manufacturers have to embrace or accept the changes brought about by new technologies, especially, especially in the field of 2D art, the integration of AI tools into workflow has become a common phenomenon.

The author also visited a number of game companies recently, and made some exchanges on the current pipeline production process, efficiency improvement, problems encountered, and demand for AI after using AIGC tools.

Among them, the 100-person small and medium-sized teams may benefit the most, but while enjoying huge dividends, they are also worried that with the continuous development of AI technology, whether the lack of underlying AI technology strength will make these manufacturers in the future AI wave was thrown far away.

In contrast, teams with fewer than 100 people and companies with thousands of people have less aggressive attitudes towards AI, or even just wait and see. But similarly, almost everyone has a positive attitude towards the long-term changes that AI will bring to the game industry.

They hesitate, they rejoice, they fear, they long... Gamers in the AI era are undergoing a revolution intertwined with creativity, emotion, and technology.

Cost reduction and efficiency increase indeed

Although it has been almost a year since the appearance of "Space Opera House", AI tools are still not accepted by many artists, which is not uncommon in the game industry.

But for some companies, compared with the huge temptation of cost reduction brought by AI, the opposition of artists seems not so important.

**Kim is the CEO of a game company with hundreds of employees. They mainly work in the MMORPG category. He told the author that art assets account for almost 60% of product development costs. After introducing AI art tools into the workflow, their The creation time and cost of 2D art assets have been optimized by about 50%. **

Another card manufacturer with a similar scale also said that their "cost reduction and efficiency increase" rate can probably reach 40%-50%, and the boss's attitude is more radical, and they will periodically ask about AI's cost reduction and efficiency increase. And it is required that all artists must be able to use AIGC art tools such as Stable Diffusion and Midjourney.

"AIGC art tools like Stable Diffusion are open source. About 5 or 6 of us use our own assets to train, and finally we can meet some stylized art requirements," Kim said. "This process actually costs nothing. , and the quality is not bad, sometimes even better than artificial ones, and some graphic artists can use them directly with a little modification.”

Almost every company will have AIGC art researcher recruitment needs

After visiting, the author found that almost every manufacturer has built a set of art production tools with Stable Diffusion, and some manufacturers have also specially established corresponding lexicons. The category and quality requirements are different.

For example, the two manufacturers mentioned above do not need that kind of very top-level art assets. AI can fulfill their needs efficiently and quickly, and naturally becomes their "good helper".

And another team with less than 100 people is currently developing a 3A-level full-platform game. Now the team's use of AI is still in the reference stage and will not participate in the production process.

The person in charge of the team said: "It's like cooking. If you go to a high-end restaurant to eat, whether the dishes made by hand make you feel better or the dishes made by machines are more advanced. I think it should be the former." However, he is also optimistic that AI can bring positive changes to the game industry in the future.

** As for whether it will lead to the unemployment of some artists, I can only say that it does. But at present, most of the people who operate these AIGC art tools are still the original artists, because the AI drawing still needs to be modified. It is not simply replacing the art with AI, but more like equipping the artists with a new painting tool. **

It is undeniable that the cost reduction and efficiency increase of AIGC tools is very obvious, which is also the core reason why manufacturers actively embrace AI tools. As for the negative impact of AI, such as the stifling of creativity, homogenization of style, and poor communication process that many artists and players worry about, how do manufacturers consider it?

AI kills creativity? uncertain

From Leo, the technical director of a thousand-person manufacturer, the author did get some different answers.

Leo said that their process of promoting AI art tools is still relatively soft. For them, the advantages of AI art tools are two aspects, and cost reduction and efficiency increase are naturally one of them. On the other hand, instead of killing creativity, the use of AI tools can sometimes stimulate it.

"Because what a person creates is more or less affected by his own experience, knowledge, style, and ability, and AI brings together the wisdom of thousands of people, it can be more diverse, richer, and more Unconstrained style,” Leo said, “Some of the pictures generated by AI, our artists said that they really couldn’t think of the possibility of such a composition.”

Epoch-making "Space Opera House"

Of course, the author is not advocating that AI is the same, but I just want to express that the advantages of AI in terms of improving efficiency and providing inspiration have been generally recognized by the industry.

Now let's talk about the shortcomings of AI.

From the perspective of technical directors, CEOs and other management positions, the overall cost reduction and efficiency increase of AI is in front of us, but when it comes to individual artists, the instability of AI tools has also become an annoyance in their work.

Due to the widespread use of AI tools, some manufacturers have greatly reduced the delivery time of artists. It may be that the original painting that was completed in ten days now requires two days to be delivered. Some art complained that it took him a day to adjust, but the AI still failed to generate a satisfactory picture. In the end, he had to go home from get off work and work overtime to draw by hand.

However, the managers are also aware of this instability, so they will not make the delivery time too tight. The manufacturers are still pursuing a "win-win" in terms of quality and time.

**In addition, in some 3D game productions, since 3D AI generation technology is still in the exploratory stage, 3D modeling such as characters and scenes still has to be manually operated by modelers. However, since the original painting referenced during modeling is generated by AI rather than completely original by the artist, it is inevitable that there will be details in the communication that neither the 2D artist nor the modeler can understand. **

In response to this kind of situation, Leo told the author that this is also the reason why they emphasize creative reference rather than cost reduction and efficiency increase when using AI tools.

** "In our opinion, cost reduction and efficiency increase are the result rather than the goal. If the modified AI-generated graph can meet or even exceed our standards, then it will be used smoothly; if not, then we are also willing to spend more Time for the artists to paint themselves," Leo said. **

**In order to prevent the frequent occurrence of the above situation, Leo's company basically requires the artist to self-interpret and modify each element of the AI-generated drawing, so that the artist can participate in and understand the conception and creation of the entire painting. **

**Although it sounds like the art does not cost much effort, but from the perspective of the entire production process, a 3D character that takes 3-6 months to produce can probably reduce the creation time by one month by using AI tools, and the optimization is mainly It is 2D art, that is, the original painting setting stage. **

It can be seen that due to differences in categories, project quality, company size, and audience groups, different manufacturers have different attitudes towards AI tools. It is the AI methodology of manufacturers to dig out the good ones, eliminate the bad ones, and find their own core needs. .

Myths in the AI era

The author said at the beginning of the article: "Medium-sized game manufacturers may be the most complicated group of people in this AIGC wave."

On the one hand, they are small in scale, relatively easy to manage, and the products they produce are not extremely demanding on art quality, making them the biggest beneficiaries among game manufacturers.

On the other hand, they do not have the technical foundation to train the underlying large models, so they also worry that with the rapid development of AI technology, they will be left behind by big manufacturers with their technical advantages.

For the industry, the most in-demand and immature technology is the AI technology for 3D model generation. There are indeed some 3D AIGC technologies on the market that can help optimize some links in the production of 3D models, such as UV display, LOD deletion, skinning and bone binding, etc. It's still a long way from actually going into game production.

NVIDIA GET3D 3D reconstruction solution

Interestingly, more than half of the manufacturers that the author visited at the beginning have a background in large manufacturers, so they have relatively few worries about this aspect. It is only the difference between early use and late use.

As for independent medium-sized manufacturers, they have the goal of becoming a medium-to-large manufacturer, but they do not have the corresponding technical strength at present.

Looking at it now, in the game industry (except for third-party service providers) those who have enough strength to get involved in AIGC's underlying technology are basically traditional Internet companies such as Tencent, NetEase, Byte, Ali, and Kuaishou. Among pure game companies, Mihayou may be the most There is a chance.

The demands of medium-sized manufacturers are also very direct. They hope that open source 3D generation models or products will appear on the market for them to use, or that third-party service providers can provide customizable 3D generation commercial products. As long as the effect is good, they are willing to pay for it. use.

Therefore, the previous expectation that AI can help small and medium-sized teams compete with large companies or even the assertion that "AIGC era can run the next Mihayou" is difficult to achieve now.

But in the short term, AI can help small teams achieve tasks that required a large number of people to complete in the past, so that the best presentation of developers' ideas may be achievable. Although it is difficult to recreate "Original God", being able to recreate "Vampire Survivor" is not a small success.

The author understands that in addition to art production, some manufacturers are also trying to use AI to explore some native AI gameplay, making AI-driven the real core gameplay, but almost all of them are still in the prototype stage. If you can run out of a more mature gameplay system, it is not impossible to create a brand new gameplay category.

Perhaps, under the wave of AI, this is the real technological dividend of the game industry.