TaleWeaver-StoryGen

Overview

TaleWeaver-StoryGen is an AI-powered storytelling project that generates stories with a unique title, continuous narrative, and moral lesson.
It is a multi-modal (text + image) AI project, meaning it produces story text and a cover image for the story.

This project demonstrates:

• Generative AI concepts
• API usage
• Running models locally via pipelines
• Integrating text and image generation into one workflow

---

Multi-Modal AI / Text & Image Integration

• Text Generation: Uses mistralai/Mistral-7B-Instruct-v0.2 via Hugging Face API to generate story text.
• Image Generation: Uses Stable Diffusion pipeline to create a cover image based on the story title.
• PDF Creation: Combines generated text and cover image into a professional storybook PDF.

This demonstrates how AI can handle multiple modalities (text + image) simultaneously.

---

Features

• Generates stories based on topic, genre, and style.
• Parses stories into Title, Story, and Moral.
• Generates a cover image automatically.
• Creates a PDF storybook with cover, formatted paragraphs, and moral.
• Uses Hugging Face API for text generation.
• Uses local Stable Diffusion pipeline for image generation.
• Supports GPU acceleration for fast local image inference.

---

Usage

1. Set story parameters:

topic = "The Lion and the Crystal Lake"
genre = "Fantasy"
style = "Magical and Heart-touching"
length_pages = 3

2. Run the notebook (TaleWeaver.ipynb) or script.

---

Installation

1. Clone the repository

git clone https://github.com/imhnor/TaleWeaver-StoryGen.git
cd TaleWeaver-StoryGen

---

Hugging Face API Token

To use hosted text generation models:

1. Create an account on Hugging Face.
2. Go to Settings → Access Tokens → New Token.
3. Copy the token.
4. Set it in your code:

import os
os.environ["HF_TOKEN"] = "your_huggingface_token_here"

Or directly in the client:

from huggingface_hub import InferenceClient
client = InferenceClient(token="your_huggingface_token_here")

The token is required to generate text from Hugging Face hosted models.

---

Learning Purpose / What You Learn

This project is designed to help understand how Generative AI works:

• Prompt Design: Craft prompts for structured story generation.
• Multi-Modal AI: Learn to integrate text and image generation.
• API & Local Models: Use Hugging Face API and local pipelines together.
• Content Processing: Convert AI outputs into professional PDFs.
• Creative Experimentation: Understand AI storytelling and explore model behavior.

---

References

• Hugging Face Inference API
• Diffusers: Stable Diffusion Pipeline
• ReportLab Documentation