The Use of AI for Video Game Development in Small Teams
- Gonzalo Frias
- Jan 13
- 2 min read
In July 2024, amidst the surge of new artificial intelligence tools, a question came to mind: How much will the implementation of AI impact my work? This thought led me to explore the most efficient ways to use these tools.
The idea was to create a horror video game where I could pour all my research and, relying solely on myself, see how far I could go in its development.
First, I needed a compatible design and theme for this experiment. I decided to create a retro-style game, aiming to evoke the nostalgia we feel when we see games from the early 2000s. Back then, developers achieved a lot with very little by employing various strategies to optimize resources and compress everything as much as possible.
I began experimenting with AI image generators, including Stable Diffusion, MidJourney, Leonardo AI, and Retro Diffusion. The results varied, and I quickly identified their limitations. While the technology has advanced rapidly in recent years, its pursuit of perfection has caused progress to plateau. The most valuable use I found was in generating different textures, which I later mixed and applied to my 3D models. Other outputs served as inspiration, helping me stay creatively fresh and explore design options for my game.
For text generation, I exclusively tested ChatGPT. It was incredibly helpful for solving minor programming challenges that arose and for brainstorming ideas for the game's lore.
Next, I worked with AI tools for sound generation: Stable Audio, ElevenLabs, and LANDR. These produced results that aligned well with my vision, although post-processing was necessary to refine them to my needs. Retro games often used heavily compressed sounds, especially in the horror genre, where the lack of fidelity evoked unease. I also experimented with generating voices and dialogue, which proved useful but limited due to their robotic tone and lack of emotional nuance.
Finally, I discovered Suno, an AI capable of creating music. This tool amazed me, as music is one of my hobbies—it felt like having a new toy. Inspired, I explored ways to integrate its output into the game, recalling the in-game radios from titles like Fallout, GTA, and Cyberpunk. These radios enhance immersion by reflecting the game world’s culture through its music. I began creating songs that narrated aspects of the game’s lore and assembled them into radio programs. Since the game’s concept revolves around post-apocalyptic life, I wanted to emphasize the theme of a lost culture, allowing players to traverse the ruined world accompanied by meaningful, captivating songs.
In summary, I worked closely with these technologies, leveraging them as tools. I realized that AI still relies heavily on human creativity to contribute to a final product like a video game. I concluded that AI tools are incredibly beneficial for small teams with big ideas, as they can significantly reduce development time and help creators achieve more than ever before.
Comments