Welcome to T.A.L.E., a groundbreaking tool for role-playing game enthusiasts! T.A.L.E. elevates your RPG experiences by transforming your session notes into an engaging website. This is achieved using the advanced capabilities of OpenAI's GPT-3.5-turbo and DALL-E-3 APIs, bringing a new dimension to your RPG narratives.

In short, this means you can turn your boring, plain notes into an aesthetic experience like this:

• Automatic Summarization: Transforms detailed RPG notes into succinct, engaging summaries.
• AI-Generated Imagery: Enhances summaries with captivating images generated by DALL-E-3, tailored to the context of your adventure.
• Web Generation: Automatically crafts a visually rich website to showcase your RPG sessions.

• Python 3.x
• Docker (Optional, recommended for a containerized environment)
• OpenAI API keys for GPT-3.5-turbo and DALL-E-3
• Roleplaying notes in a .txt-file.

1. Clone the repository:

   git clone https://github.com/your-username/TALE.git

2. Navigate to the cloned directory:

   cd TALE

3. Install dependencies:

   pip install -r requirements.txt

1. Insert your OpenAI API keys in secret.txt.
2. Put your roleplaying notes as .txt file in data/input/notes
   • You can add as many .txt files here as you want. Theoretically, the framework works with notes from different players in one session
3. For best results modify the file data/input/context.json. The JSON works like this:

   {
    "character1" : "prompt description of character",
     "place1" : "prompt description of place",
     ....
   }

Run the following command:

python tale/tale.py

This processes your RPG notes, creates summaries and images, and assembles a website in data/output.

The generated website will be in data/output. Access output.html to explore your RPG session stories and accompanying visuals.

• Regularly update your notes for accurate summaries.
• Monitor API usage to manage costs.
• Keep backups of original notes.
• Use context.json to ensure that character & place descriptions remain consistent
• Try being precise in your roleplaying notes i.e. always write down names of people & places of action

Dependent on OpenAI API availability and limits.

Sometimes the pipeline still breaks due to formatting issues. This is because the GPT-3.5 response is not always correct and can break by:

• Not outputting correct .json formats in-between

• Not matching images and text correctly

• Generating image prompts that might get rejected by the image generation API

• Generating outputs that are too long for the next steps

  We are aware of such issues and will work on improving them over time.

To obtain an API key, visit OpenAI's API page. Costs vary based on usage; see OpenAI's pricing page for details. Please take not that especially image generation can be pricey over time.

We are planning on eventually improving T.A.L.E. by including GUI components and making outputs more interactable as well as the pipeline more reliable. Please be aware that we are working on this project in our free-time so changes might take some time.

• OpenAI for the GPT-3.5-turbo and DALL-E-3 APIs
• All contributors and supporters of the T.A.L.E. project