fltman/claude-code-skill-sora-video/sora-video

Sora Video Generation

Generate high-quality AI videos using OpenAI's Sora model.

Capabilities

• Text-to-video: Create video clips from text descriptions
• Image-to-video: Animate a reference image into a video clip
• Remix: Create variations of previously generated videos

Requirements

Before using this skill, ensure:

1. Python packages are installed:

``bash pip install openai ``

1. API key is set:

``bash export OPENAI_API_KEY="your-openai-api-key" ``

Get your key at: https://platform.openai.com/api-keys

Usage

Text-to-Video

Generate a video from a text description:

python skills/sora-video/scripts/generate_video.py --prompt "A drone shot flying over a misty forest at sunrise" --output forest.mp4

Image-to-Video

Animate a reference image:

python skills/sora-video/scripts/generate_video.py --prompt "The scene comes to life with gentle wind" --input photo.jpg --output animated.mp4

Higher Quality (Pro)

Use the pro model for production-grade quality:

python skills/sora-video/scripts/generate_video.py --prompt "A cinematic sunset over mountains" --model sora-2-pro --size 1792x1024 --seconds 12 --output cinematic.mp4

Script Arguments

Available Sizes

Prompt Tips

For best results:

1. Be cinematic: Describe camera movement, lighting, and atmosphere
2. Specify style: "photorealistic", "anime style", "watercolor animation"
3. Describe motion: "slowly panning", "time-lapse", "tracking shot"
4. Set the mood: "dramatic lighting", "golden hour", "neon-lit"

Example Prompts

Nature:

A drone shot flying over a misty forest at sunrise, golden light filtering through
the trees, birds rising from the canopy, cinematic quality

Urban:

A time-lapse of a busy Tokyo intersection at night, neon signs reflecting on wet
pavement, crowds crossing in all directions, cyberpunk atmosphere

Abstract:

Liquid metal flowing and morphing into geometric shapes, iridescent reflections,
dark background, smooth motion, high-end product visualization style

Pricing

Note: You are charged for every generation attempt, including failed ones.

Workflow

1. Understand what the user wants to create
2. Craft a detailed, cinematic prompt (camera angles, lighting, mood, motion)
3. Choose appropriate model, duration, and resolution
4. Run the generation script
5. If the user provides an image, use image-to-video mode with --input
6. Present the generated video to the user
7. Offer to iterate or adjust based on feedback

Troubleshooting

"OPENAI_API_KEY not set"

• Set the environment variable: export OPENAI_API_KEY="your-key"

"openai package not installed"

• Install it: pip install openai

"Video generation failed: content_policy_violation"

• The prompt was flagged by content moderation
• Remove references to real people, copyrighted characters, or inappropriate content
• Rephrase the prompt and try again

"Video generation timed out"

• Pro model and higher resolutions take longer (up to 5 minutes)
• Increase timeout: --timeout 900
• Try with sora-2 (faster) instead of sora-2-pro

"Rate limit exceeded"

• Wait a moment and try again
• Standard API tier allows ~10 requests per minute

Output

The script will:

1. Print progress messages with status and percentage
2. Save the MP4 video to the specified output path
3. Print "SUCCESS" with the file path on success
4. Return exit code 0 on success, 1 on failure