amolkulkarni95/realistic_anime_movie_generator

Realistic AI Anime Movie Generator (MVP)

This is an MVP slideshow-style AI Movie Generator that converts a short story into a narrated animated slideshow (images + narration).

Features (MVP)

• —Splits the story into several scenes
• —Generates images for each scene using the Hugging Face Inference API (Stable Diffusion XL)
• —Synthesizes narration for each scene using gTTS
• —Stitches images + narration into an MP4 using MoviePy
• —Streamlit UI for input and preview

Setup & Deploy

1. 1.Create a new repository and add these files or upload the ZIP content to your Streamlit app folder.
2. 2.Add a Hugging Face API token (required) into Streamlit Secrets with key HF_TOKEN and optionally set HF_SDXL_URL if you use a custom endpoint.

Example secrets.toml:

HF_TOKEN = "your_hf_token_here"
HF_SDXL_URL = "https://api-inference.huggingface.co/models/stabilityai/stable-diffusion-xl-base-1.0"

1. 1.Install requirements and run locally:

pip install -r requirements.txt
streamlit run app.py

Notes & Limitations

• —Using HF Hosted inference may consume credits and has rate limits.
• —This MVP creates a slideshow video — for real frame-by-frame animation consider integrating AnimateDiff or Runway for motion generation.
• —Replace gTTS with ElevenLabs for more realistic voices.