image-to-video

What this skill does

> **Install the belt CLI skill:** `npx skills add belt-sh/cli`

# Image to Video

Convert still images to animated videos via [inference.sh](https://inference.sh) CLI.

## Quick Start

> Requires inference.sh CLI (`belt`). [Install instructions](https://raw.githubusercontent.com/inference-sh/skills/refs/heads/main/cli-install.md)

```bash
belt login

# Generate a still image
belt app run falai/flux-dev-lora --input '{
  "prompt": "serene mountain lake at sunset, snow-capped peaks reflected in still water, golden hour light, landscape photography",
  "width": 1248,
  "height": 832
}'

# Animate it
belt app run falai/wan-2-5-i2v --input '{
  "prompt": "gentle ripples on the lake surface, clouds slowly drifting, warm light shifting, birds flying in the distance",
  "image": "path/to/lake-image.png"
}'
```


## Model Selection

| Model | App ID | Best For | Motion Style |
|-------|--------|----------|-------------|
| **Wan 2.5 i2v** | `falai/wan-2-5-i2v` | Realistic motion, natural movement | Photorealistic, subtle |
| **WAN-I2V (Pruna)** | `pruna/wan-i2v` | Economical, fast, 480p/720p | Natural, efficient |
| **Seedance 2.0** | `bytedance/seedance-2-0` | Up to 1080p, sync audio, all input types | Versatile, high quality |
| **Seedance 2.0 Fast** | `bytedance/seedance-2-0-fast` | Fast variant, same capabilities | Versatile, fast |
| **Fabric 1.0** | `falai/fabric-1-0` | Cloth, fabric, liquid, flowing materials | Physics-based flow |
| **Grok Imagine Video** | `xai/grok-imagine-video` | General animation, text-guided | Versatile |

### When to Use Each

| Scenario | Best Model | Why |
|----------|-----------|-----|
| Landscape with water/clouds | **Wan 2.5 i2v** | Best at natural, realistic motion |
| Portrait with subtle expression | **Wan 2.5 i2v** | Maintains face fidelity |
| Product with fabric/cloth | **Fabric 1.0** | Specialized in material physics |
| Flag waving, curtain flowing | **Fabric 1.0** | Cloth simulation |
| Illustrated/artistic image | **Seedance 2.0** | Matches stylized content |
| General "bring to life" | **Seedance 2.0** | Good all-rounder, up to 1080p |
| Quick test/iteration | **Seedance 2.0 Fast** | Faster generation |

## Motion Types

### Camera Movement

| Movement | Prompt Keyword | Effect |
|----------|---------------|--------|
| Push in / Dolly forward | "slow dolly forward", "camera pushes in" | Increasing intimacy/focus |
| Pull out / Dolly back | "camera pulls back", "slow zoom out" | Reveal, context |
| Pan left/right | "camera pans slowly to the right" | Scanning, following |
| Tilt up/down | "camera tilts upward" | Revealing height |
| Orbit | "camera orbits around the subject" | 3D exploration |
| Crane up | "camera rises upward" | Grand reveal |
| Static | (no camera movement prompt) | Subject motion only |

### Subject Motion

| Type | Prompt Examples |
|------|----------------|
| Natural elements | "water rippling", "clouds drifting", "leaves rustling in breeze" |
| Hair/clothing | "hair blowing gently in wind", "dress fabric flowing" |
| Atmospheric | "fog slowly rolling", "dust particles floating in light beams" |
| Character | "person slowly turns to camera", "subtle breathing motion" |
| Mechanical | "gears turning", "clock hands moving" |
| Liquid | "coffee steam rising", "paint dripping", "water pouring" |

## Prompting Best Practices

### The Golden Rule: Subtle > Dramatic

AI video models produce better results with **gentle, subtle motion** than dramatic action. Requesting too much movement causes distortion and artifacts.

```
❌ "person running and jumping over obstacles while the camera spins"
✅ "person slowly walking forward, gentle breeze, camera follows alongside"

❌ "explosion with debris flying everywhere"
✅ "candle flame flickering gently, warm ambient light shifting"

❌ "fast zoom into the eyes with dramatic camera shake"
✅ "slow dolly forward toward the subject, subtle focus shift"
```

### Prompt Structure

```
[Camera movement] + [Subject motion] + [Atmospheric effects] + [Mood/pace]
```

### Examples by Scenario

```bash
# Landscape animation
belt app run falai/wan-2-5-i2v --input '{
  "prompt": "gentle camera pan right, water reflecting moving clouds, trees swaying slightly in breeze, warm golden light, peaceful and slow",
  "image": "landscape.png"
}'

# Portrait animation
belt app run falai/wan-2-5-i2v --input '{
  "prompt": "subtle breathing motion, slight head turn, natural eye blink, hair moving gently, soft ambient lighting shifts",
  "image": "portrait.png"
}'

# Product shot animation
belt app run bytedance/seedance-2-0 --input '{
  "prompt": "slow 360 degree orbit around the product, gentle spotlight movement, subtle reflections shifting, premium product showcase, smooth motion",
  "image": "product.png",
  "generate_audio": true
}'

# Fabric/cloth animation
belt app run falai/fabric-1-0 --input '{
  "prompt": "fabric flowing and rippling in gentle wind, natural cloth physics, soft movement",
  "image": "fabric-scene.png"
}'

# Architectural visualization
belt app run falai/wan-2-5-i2v --input '{
  "prompt": "slow dolly forward through the entrance, slight camera tilt upward, ambient light filtering through windows, dust particles in light beams",
  "image": "building-interior.png"
}'
```

## Duration Guidelines

| Duration | Quality | Use For |
|----------|---------|---------|
| 2-3 seconds | Highest quality | GIFs, looping backgrounds, cinemagraphs |
| 4-5 seconds | High quality | Social media posts, product reveals |
| 6-8 seconds | Good quality | Short clips, transitions |
| 10+ seconds | Quality degrades | Avoid unless stitching shorter clips |

### Extending Duration

For longer videos, generate multiple short clips and stitch:

```bash
# Generate 3 clips from the same image with progressive motion
belt app run falai/wan-2-5-i2v --input '{
  "prompt": "slow pan left, gentle water motion",
  "image": "scene.png"
}' --no-wait

belt app run falai/wan-2-5-i2v --input '{
  "prompt": "continuing pan, clouds shifting, light changing",
  "image": "scene.png"
}' --no-wait

# Stitch together
belt app run infsh/media-merger --input '{
  "media": ["clip1.mp4", "clip2.mp4"]
}'
```

## The Full Workflow

### Still-to-Final-Video Pipeline

```bash
# 1. Generate source image (best quality)
belt app run bytedance/seedream-4-5 --input '{
  "prompt": "cinematic landscape, misty mountains at dawn, lake in foreground, dramatic clouds, golden hour, 4K quality, professional photography",
  "size": "2K"
}'

# 2. Animate the image
belt app run falai/wan-2-5-i2v --input '{
  "prompt": "gentle mist rolling through the valley, lake surface rippling, clouds slowly moving, birds in distance, warm light shifting",
  "image": "landscape.png"
}'

# 3. Upscale video if needed
belt app run falai/topaz-video-upscaler --input '{
  "video": "animated-landscape.mp4"
}'

# 4. Add ambient audio
belt app run infsh/hunyuanvideo-foley --input '{
  "video": "animated-landscape.mp4",
  "prompt": "gentle nature ambience, distant birds, soft wind, water lapping"
}'

# 5. Merge video with audio
belt app run infsh/video-audio-merger --input '{
  "video": "upscaled-landscape.mp4",
  "audio": "ambient-audio.mp3"
}'
```

## Cinemagraph Effect

A cinemagraph is a still photo where only one element moves (e.g., waterfall moving in an otherwise frozen scene). To achieve this:

1. Generate the still image with the motion element clearly defined
2. Prompt for motion only in that specific element
3. Keep to 2-4 seconds for seamless looping

```bash
belt app run falai/wan-2-5-i2v --input '{
  "prompt": "only the waterfall is moving, everything else remains perfectly still, water cascading smoothly, rest of scene frozen",
  "image": "waterfall-scene.png"
}'
```

## Common Mistakes

| Mistake | Problem | Fix |
|---------|---------|-----|
| Too much motion requested | Distortion, artifacts, warping | Subtle > dramatic, always |
| Wrong model for content type | Poor results | Use selection guide above |
| Clips too long (10s+) | Qualit