---
title: 🎬 Image-to-Video AI
emoji: 🎬
colorFrom: indigo
colorTo: purple
sdk: gradio
sdk_version: "5.34.2"
suggested_hardware: zero-a10g
app_file: app.py
pinned: true
license: apache-2.0
short_description: Real AI animation from any image
tags:
  - image-to-video
  - cogvideox
  - video-generation
  - animation
  - diffusion
  - zero-gpu
models:
  - THUDM/CogVideoX-5b-I2V
---

# 🎬 Image-to-Video AI Studio

**Transform any image into a real AI-animated video — things actually move!**

> ⚡ Free ZeroGPU · CogVideoX-5B · Real Diffusion Animation · Not Zoom Tricks

## ✨ What This Does

Upload any image + describe what should animate → **CogVideoX-5B generates actual video frames** where things move naturally: water flows, hair blows, people walk, clouds drift.

This is a **5-billion parameter diffusion model** generating real new frames — not zoom/pan/warp tricks.

## 🎛️ Controls

| Parameter | Description |
|-----------|-------------|
| **📸 Image** | Upload any photo — landscapes, portraits, animals, cities, art |
| **✍️ Prompt** | Describe what should move: "the river flows, birds fly overhead" |
| **🎥 Camera Motion** | Static, Zoom, Pan, Tilt, Orbit, Dolly, Crane, Handheld |
| **🎭 Style** | Cinematic, Realistic, Dreamy, Dynamic, Slow-Mo, Nature Doc, Anime |
| **⏱️ Duration** | ~3 seconds (24 frames) or ~6 seconds (49 frames) |
| **🔄 Steps** | 15 (fast) → 30 (balanced) → 50 (best quality) |
| **🎯 Guidance** | How strictly to follow your prompt (1-12) |

## 📊 Technical Details

- **Model:** [CogVideoX-5B-I2V](https://huggingface.co/THUDM/CogVideoX-5b-I2V) — 5 billion parameters
- **Resolution:** 720×480 (landscape) or 480×720 (portrait) 
- **Output:** 8 FPS MP4 video, ~3-6 seconds
- **Generation time:** ~2-4 minutes on ZeroGPU
- **Hardware:** ZeroGPU (free A10G access on HuggingFace)