GhostPackDemo/readme.md

👻🎬 GhostPackDemo · Apache 2.0

Generate cinematic 30 FPS AI videos from a single image—runs locally on your GPU.

---

⚡ Showcase

---

1 · Overview

GhostPackDemo bundles HunyuanVideo + FramePack into a single-script GPU pipeline (ghostpack.py) with an intuitive Gradio UI.
Upload any starter image, enter a prompt, and receive a 30 FPS MP4 clip with live denoising preview.

---

2 · Feature Matrix

✅ Feature	Description
Image → Video diffusion	Full diffusion loop conditioned on image and text
Live progress & preview	Real-time denoise frames + progress bar
VRAM-aware execution	Automatic slicing/offload when free VRAM < 60 GB
Offline HF cache	Models cached once in hf_download/
One-click installer	install.py builds venv, installs deps, logs in to HF, downloads models
Cross-platform	Linux 🐧 · WSL · Windows 🪟 (PowerShell)
Permissive license	Distributed under Apache 2.0—commercial & non-commercial use allowed

---

3 · Architecture