Description
AI GENERATED CONTENT DISCLOSURE
The developers describe how their game uses AI Generated Content like this:
As a pre-GPT game that started in 2020, the motivation behind this experimental game is to explore creative narrative gameplay based on conversational AI.
When we started this game as an art and academic project, our guiding principle was to use AI only to create novel gaming experiences that were previously impossible, not to replace human creativity in any way. Inspired by the ancient Persian folklore, the game’s soul comes from us, the human creators, not the AI technology itself. AI is just the means by which this vision is brought to life and shared with you. As human artists and researchers, we hope this game serves as an example of how technology can be an actor, while human creators become the directors who shape it in artistic and innovative ways.
All main game assets are manually made by artists in the team, and only used AI visuals to assist minor components, e.g. far background sprite and carpet textures. Please see our official website for the detailed credits.
The main mechanism is: You narrate stories to the King, who is powered by large language model (the LLM in our free trial is GLM, and you can also use GPT 3.5 or 4 to play the game), and he continues your tale. You must guide him to mention specific weapon keywords. Each weapon is unique, with descriptions generated based on the stories you and the King create together. As you collect more weapons, a world emerges within the story and on the weapon cards.
This real-time imagery is generated through Stable Diffusion based on the stories you tell. We make a conscious effort to avoid using specific artists’ styles to prevent infringement, maintain a general style, and ensure transparency. The main goal is to bring the stories to life in real-time while keeping it consistent with the game’s artstyle. This means we ensure the real-time generated images in our game won’t be mistaken for those created by particular artists or be overly similar to any artist’s style. The base model we use is Cheese Daddy, combined with Pixelization algorithm. These are all opensourced. The workflow is mostly explained in the academic papers we published.
Currently, we’re a small and self-funded indie team without any funding or income for this project. We are awared that the stochastic and source-consuming generative AI technology still has ethical problems in terms of environment and authorship. With appropriate future funding and support, we aim to address and reduce potential harm from generative AI technologies. We welcome any concerns, suggestions, or questions about our AI-generated content and processes at adaeden1001[at]gmail.com and our Discord. Your feedback will help us improve and explore, and become more considerate.
Reviews
There are no reviews yet.