At the Global Game Developers Conference (GDC 2023), which ends in March 2023, the topic of "Application of AI technology in games" has become the focus of most attention, and almost everyone is discussing it. Sam Altman, the founder of OpenAI, said in a recent interview: "AI is one of the rare things that has been seriously underestimated after being seriously hyped." This sentence is more appropriate in the game industry. It can be said that this round of AIGC technology The rapid development will bear the brunt of the game industry, which will naturally give birth to a wave of change in the entire game industry.
first level title
1. AI + globalization trend
In fact, the application of AI in the game industry is no longer in the discussion stage—the game industry has always been the main driving force for the development of AI technology. From smart battles to virtual worlds, from smart customer service to AI detection of cheating behavior, AI technology is widely used used in games.
Now, driven by the wave of AIGC, the game industry is ushering in a new round of AI transformation nodes. The main trends can be summed up in four words:Reduce costs and increase efficiency.
secondary title
1.1 AI creates 3A-level high-quality animations
secondary title
1.2 Automatically create 3D game scenes based on text descriptions
Opus.ai has developed a new, text-driven approach to 3D world generation that enables users to create animated characters with dynamic lighting, camera controls, terrain, trees and animals, buildings, roads, and humanoid forms through text input . Simply put, the user only needs to type, and AI can create 3D game scenes and various components and dynamic effects in the scenes (such as: leaves swaying slightly in the wind).
secondary title
1.3 AI NPC creates the "Western World" in the game
GAEA is a technical system that has the ability to build a complete set of NPC ecosystems. At the current stage when everyone is still immersed in exploring how to use AI technology to improve R&D efficiency, GAEA may be the first set of solutions to deeply integrate gameplay and change the game content ecology, and may even inspire the product form of the next generation of games. Not long ago, a domestic hyperparameter technology company used the GAEA technology system to implement a technology demo called "Living Chang'an City". The NPCs in it have no preset scripts and are all controlled by AI. "Memory" will "remember" what you see and do and affect subsequent actions. Each AI NPC has its own goals and reasons for action, and together they form an evolving small "society". "That smell?
Coincidentally, just a while ago, the AI NPC experiment of Stanford Google's new research also caused a sensation-they created 25 AI NPCs and placed them in a simulated town. Participated in complex behavior (including throwing a Valentine's Day party, etc.).
Some social media commented that the performance of these NPCs is considered to look more "real" than real-life role-playing:
secondary title
1.4 AI generated NPC dialogue
As games got bigger and bigger, writers were faced with the challenge of making NPCs personal and realistic. With hundreds of NPCs, how do you keep each interaction with them unique? That's the problem that Ghostwriter, an in-house AI tool developed by La Forge, Ubisoft's R&D arm, solves.
secondary title
1.5 AI quickly generates game assets
first level title
2. The breakthroughs of major game manufacturers
secondary title
2.1 Development phase
secondary title
2.2 Operation stage
In the operation stage of the game, in addition to the official use of AIGC artificial intelligence to produce various materials for dissemination, not limited to images, copywriting, videos, etc., it also lowers the threshold for UGC content creation, making it easier for participating players to produce Players are also more motivated to actively spread high-quality secondary content.
Here are a few examples:
Adverse water cold
In the latest update, Nishuihan end game realizes the UGC content generation of AI-generated poems for the occasion when taking screenshots of the game, and the effect is quite amazing. The project team has set up AI lyric check-in points for more than 40 popular classics in the game. Given a keyword, it can generate Song lyrics that fit the scenery, neatly rhetoric, and reflect the theme of the keyword, with the player's signature! One-click sharing Moments, doesn't this bring countless likes?
Genshin
Using stablediffusion+LORA to produce high-quality coser photos that are difficult to distinguish between real and fake, which coser can beat the efficiency and quality of using AI technology? Adhering to the principle of joining if you can't beat it, some cosplayers directly choose to train their own LORA models. With the help of ControlNET, they use their own faces to quickly generate high-quality pictures of various characters and actions.
Not to mention the beautiful pictures of the two-dimensional characters:
A certain overseas company
Using ChatGPT for translation to achieve multilingual versions is fast and authentic, and it is many times cheaper than finding a specialized translation company before. Using Midjourney to quickly generate 100 sets of operation and delivery material diagrams, coupled with the (English) copywriting designed with Chatgpt, overseas operating costs have been greatly increased. What’s even more outrageous is that it used to cost tens of thousands of budgets to ask someone to compose, write, and sing old-fashioned fan songs. Now there are AI tools, and with a little training, you can generate an 80-point saliva song, including singing and accompaniment... …
In general, the capabilities of AIGC's current tools have allowed many major manufacturers to change the workflow of game production, and they have penetrated deeper and deeper into various links such as game production, operation, and launch of major manufacturers. Naturally, the budget for outsourcing has been slashed, and the efficiency of internal processes has also been greatly improved. I vaguely remember that when <Red Dead Redemption 2> was released, the total cost reached an astonishing 5.6 billion RMB, and the development time was as high as 8.5 years. The main money was spent on labor costs—in order to make the whole open world more real, it cost It takes unimaginable manpower to write settings, stories, and lines for each NPC... If it is put into the present, develop anotherfirst level title
3. The significance of AI for small teams
secondary title
3.1 Lowering the threshold of game development
secondary title
3.2 Reduce content costs: AI quickly outputs 2D/2.5D content materials
secondary title
3.3 Use AI technology to create unique and creative game works
Domestic game AI company City From Naught and Web 3 creator community Magipop DAO have jointly developed a detective RPG game based on LLM and generative AI, code-named "DetectiveGPT", players can interact with AI NPC free and in-depth interaction to explore the truth layer by layer, is expected to be released on Steam in April this year. The 3D model textures in this game are all completed by the 3D texture automatic generation tool developed by Nowhere City Studio. In traditional decryption games, NPCs have preset dialogues. Players can obtain clues by choosing question options. In this decryption game made by AI tools, players can freely ask witnesses through voice or text. NPC, witnesses will answer the player's questions accordingly, immersive crime solving experience, sense of substitution ++ Do you have any!
There is a foreign game developer called DevBanana, who used various AI tools such as New Bing and ChatGPT to make a Whack-a-Mole World! The AI is complete, and you can find DevBanana on YouTube to watch him make the game.
The development of AIGC technology has indeed brought endless possibilities to independent game developers - no code? ChatGPT can help you; can't draw? Midjourney, SD for you to choose; Can't write copywriting? Let CharGPT generate it. UI, game dubbing, and sound effect AI can also be competent. All you need is creativity and the ability to describe what you want to AI
(Speak quietly, when you have no ideas, you can also ask ChatGPT for inspiration!)
first level title
4. Written at the end
When the AIGC revolution was just emerging, many people were questioning and watching, and resisted learning new methods because they were afraid of substitution and did not want to leave the familiar comfort zone. However, the wheels of technology are relentlessly rolling forward——
At that time, someone said: AI painters are not good. Not long after, midjourney V 5 was released, and there was no problem with drawing hands.
At that time, someone said: The Asian women drawn by AI are too ugly to look at. Not long after, Lora, who has trained many Japanese and Korean beauties, came and easily generated all kinds of beautiful internet celebrities
At that time, some people said: AI drawing can only be used to provide inspiration, and it cannot control the movement of characters, so it is not easy to use. Not long after, stablediffusion has a plug-in Control Net, and the motion of the picture generated by AI can also be controlled
At that time, many people said: the icons and logos drawn by AI are not vector graphics, which are not easy to use. Not long after, VectorizerAI came out, converting JPG, PNG and other bitmaps into vector SVG format with one click
At that time, someone said: AI can't generate the specified text, just look at AI with garbled text! Now, the new version of SD supports posting text
……
AIGC is developing too fast. Today, a month later, how far will AI develop? I dare not think about it... But anyway, I know that it is time to let go of those prejudices and actively embrace AI technology with an open mind.
This article is over. Next "AI+WEB 3 Chain Games"
