How-We-Use AI
The story, dialogue, sound design, game design, and overarching narrative themes of Embersteel is original and handcrafted. It is, however, proofread and edited by machines, for humans (and machines).
The majority of Embersteel's content and design is hand written and hand crafted by our two person development team. We start with original ideas and concepts of our own before utilizing these tools to streamline our own creative workflows. Most of the time, we are using AI to proofread our text content and create placeholder images or descriptions that are later changes, iterated upon, and improved upon with biological creativity.
We do not claim any of Embersteel's generative AI art assets as copyright. Artists, worldbuilders, and creatives of the community are encouraged to help us update generative placeholders through hand-drawn content submitted through official guidelines and pipelines. As a studio, we own the narrative, characters, code, and overarching content written between the story and game of Embersteel as part of the Tales of Ayora Universe - a series of fictional Tales and interactive narrative video game experiences we have been hand writing and planning for over six years. Embersteel is only the beginning to our series of games, each vastly different from the previous experience, but still part of our grander cosmic narrative.
Welcoming New Workflows
If you find yourself tinkering with any of the generative tools listed below, consider contributing to the game by trying to generate new content for Embersteel! You may submit your content through our GitHub pull request/submission process. Our favorite submissions, prompts, characters, and stories will be selected as official canon based on aethetic, worfklow, vision, and thematic cohesion. This is your chance to contribute to official lore, characters, items, or entities that can be canonized into the game! For more information, check out our Help Wanted page.
Generative Art
We utilize Stable Diffusion, MidJourney, ControlNet, TemporalKit, WarpFusion, EbSynth, and StableDiffusion XL + ComfyUI to experiment with generative art worfklows within video games, combining potential multi-modal capabilities between LLM enhanced generative NPC dialogue and gameplay paired with aided visual generation and representation of dynamically generated asset content unique to Embersteel's time loop mechanic. Beyond those experimental features, we utilize generative art to draft quick prototype placeholder mockups of visual ideas that convey some of the key expressions we hope to capture with later commissioned, hand-drawn art promoted from artists within the community.
Generative Text
We utilize locally hosted LLMs like Llama2 and other open-source models to research and develop experimental gameplay features. We experiment with data frameworks like Llamaindex to investigate long-term NPC memory and dialogue. We also explore the usage of Langchain in video games for emergent generative agent gameplay, utilizing ChatGPT's API to experiment additional features with closed source models (we prefer open-source, but trial and test any model we can get our hands on).
Generative Audio
We do not currently utilize any audio-based generative AI system or integration (and do not have plans to use generative audio in Embersteel).
Commissioning Human Artists
As a commitment to empowering humans: a permanent portion of Embersteel's profits is dedicated to commissioning real art from real humans, replacing generative work with that of artists within our community. Royalties and commissions will be offered to all artists employed or contracted with the studio. With this we aim to create streams of income and opportunity for any artist willing to draw, commission, or contribute to the thematic we've outlined with generative models.
We understand there are many unknown and emerging impacts AI has on the world. We strongly advocate and believe in a world where every person has access to overcome personal hurdles utilizing these emerging systems as tools to give themselves super human capabilities. We embrace the challenges of advocating for this change and hope to continue to seek ways that create net positives for all creatives, artists, and developers alike. Without placeholder images and working with a $0 budget we would not have been able to present Embersteel to you today. The fact we can and may create artist jobs along the way is something we see as a step in the right direction, especially if those jobs may not have been possible without the initial opportunity to present Embersteel the way we have today.
All artists, digital, generative, and traditional alike will undergo the same quality review process through community polls and processing procedures.
We are a pro-modding company, meaning with enough support our frameworks will allow you to pick and choose which art styles you prefer on your run, giving you chances to opt-in to other experimental gameplay features we think are exciting and innovative, empowering and entertaining the human element with AI, not replacing it.