teamday-ai/agents/image-to-video

Image to Video Skill

Generate videos from static images using FAL AI's video generation models.

Available Models

Prerequisites

• FAL AI API key set as FAL_KEY environment variable

Usage

bun .claude/skills/image-to-video/scripts/image-to-video.ts \
  --image "https://example.com/image.png" \
  --prompt "A character waving and smiling" \
  --output "./output.mp4"

Parameters

Default Negative Prompt

blur, distort, low quality, low resolution, error, worst quality, defects

Examples

Animate an Avatar

bun .claude/skills/image-to-video/scripts/image-to-video.ts \
  --image "/path/to/avatar.png" \
  --prompt "Character nodding and blinking naturally" \
  --duration 5 \
  --output "./avatar-animation.mp4"

Generate with Audio (Kling)

bun .claude/skills/image-to-video/scripts/image-to-video.ts \
  --image "/path/to/character.png" \
  --prompt "Character says 'Hello, welcome!'" \
  --audio \
  --model kling \
  --output "./speaking-character.mp4"

Tips for Avatar Animation

1. Subtle motions work best: "blinking", "nodding slightly", "breathing"
2. Avoid large movements: Keep the character mostly in place
3. Loop-friendly prompts: End state similar to start state
4. Square images: 1:1 aspect ratio works best for avatars

Cost Considerations

• Kling V2.6 Pro: Higher quality, higher cost (~$0.35/5s video)
• Wan V2.6: Good balance of quality and cost
• 5-second videos are more cost-effective than 10-second