To all the in-game characters
2021 Fall
Read more about the dev log︎︎︎
Read more about the dev log︎︎︎
To all the in-game character is an immersive narrative exploring modern game engines' potential as story devices for meta/recursive narratives. Working with OpenAI’s large natural language processing model GPT3︎︎︎ to present a tribute to all the in-game characters.
[ Unity, Openai api]
Conception and theme
The concept was set out to explore the following two domains:
- Collaboration with machine learning to generate story materials.
- Pushing modern game engines’ ability to tell a story about a self-aware in-game character.
Before coming up with a script for the narrative, the progress started out with the creation of space and using the space as creative prompts for the story and characters.
On collaborating with OpenAI’s language generation machine learning model GPT-3
The process started out with the goal to fabricate space from texts gathered from my interaction with a trained(fine-tuned) GPT-3.
︎GPT-3 is OpenAI’s newest language model, allowing users to train the model with curated materials and prompt the trained model with texts of their choice. The model will then try to complete the prompt.

The model in the diagram was trained with two books from Jorge Luis Borges (Library of babel and The garden of the forking paths). With the looping progress of recursively going into more detailed descriptions on each iteration.
Model outputs:
︎︎︎ Below is a example of my prompt and the out put from the model
︎︎︎ Below is a example of my prompt and the out put from the model

Texts in blacks are prompts written by me, while the reds are the model’s completion.
The second and thrid prompts are made in efforts to expand the scope of the space.
The fouth prompt are made in reaction to the clockwork appeared in the first completion, in hope to generate more detailed description about the clockwork.
The second and thrid prompts are made in efforts to expand the scope of the space.
The fouth prompt are made in reaction to the clockwork appeared in the first completion, in hope to generate more detailed description about the clockwork.

On the story
Inspired by Borges’ work and the generated texts, the narrative revolves around an in-game character player controls, who, through discourses with an entity called “Scene Manager” and interactions with the physical clues in the space, gradually realized that they’re in a game.

Separating the player and in-game protagonist
To tell a story about a gradually self-aware in-game protagonist, the narrative set out to provide a one-way dialog between the game and the silent protagonist.
To tell a story about a gradually self-aware in-game protagonist, the narrative set out to provide a one-way dialog between the game and the silent protagonist.
On meta/recursive narrative and game engines
Treating game engines as tools to crystalize and elevate literal metaphors, instill the same genuineness across all planes of existence.
Treating game engines as tools to crystalize and elevate literal metaphors, instill the same genuineness across all planes of existence.


The player goes into the back room, where they see a scaled model of this level, hinting at a recursive narrative and telling our "protagonist" something's off.


In the second level, a new character called "Scene Manager" is introduced, speaking on behalf of the game itself. The name "Scene Manager" is a direct reference to a common Unity game development practice, where developers often create and name a meta-object (usually not seen in-game) "Scene Manager" to control the events in-game. In the context of the narrative and to the protagonist, this is the first direct breach of the fourth wall.
As the scene manager reveals more meta-information, the player starts to separate themselves from the protagonist, ascribing a more integrated identity to the in-game protagonist.





An acknowledgement and celebration
The experience serves as an appreation for all the in-game characters, breaching the limitaion of metaphors through interations across different worlds and agents. Evoking empathy through the same authenticity.