StillTale is an AI-powered Slideshow video generation platform that converts text prompts or full stories into complete videos with consistent characters, scene visuals, and voiceover narration.

Watch the Demo

1. Character Reference Images: When the story mentions characters like "Luna with silver hair," we call Bria's text-to-image endpoint to generate a reference portrait with Detailed description. This image is stored and reused to keep the character looking consistent.

2. Scene Generation: For each scene, we send a descriptive prompt to Bria. If the scene includes a known character, we pass the character's reference image using Bria's image-to-image feature, ensuring visual consistency throughout the video.

Backend:

• Python, FastAPI, SQLite
• Google Gemini AI (story generation, scene creation)
• Bria AI API (image generation)
• gTTS (text-to-speech)
• OpenCV, FFmpeg (video processing)

Frontend:

• React, Tailwind CSS
• Axios, React Router

• Python 3.10+
• Node.js 18+
• FFmpeg installed and in PATH

Create a .env file in the root directory:

GOOGLE_CLOUD_PROJECT=your-project-id
GOOGLE_CLOUD_LOCATION=project-region
GOOGLE_GENAI_USE_VERTEXAI=True
BRIA_API_TOKEN=your-bria-api-token

# Install dependencies
pip install -r requirements.txt

# Create .env file with your API keys
cp example.env .env
# Edit .env with your credentials

# Run backend
python app.py

cd frontend
npm install
npm start