Sora Video Generation

SKILL.md

---
name: sora
description: |
  Generate videos from text prompts or reference images using OpenAI Sora.
  
  ✅ USE WHEN:
  - Need AI-generated video from text description
  - Want image-to-video (animate a still image)
  - Creating cinematic/artistic video content
  - Need motion/animation without lip-sync
  
  ❌ DON'T USE WHEN:
  - Need lip-sync (person speaking) → use veed-ugc or ugc-manual
  - Just need image generation → use nano-banana-pro or morpheus
  - Editing existing videos → use Remotion
  - Need UGC-style talking head → use veed-ugc
  
  INPUT: Text prompt + optional reference image
  OUTPUT: MP4 video (various resolutions/durations)
---

# Sora Video Generation

Generate videos using OpenAI's Sora API.

## API Reference

**Endpoint:** `POST https://api.openai.com/v1/videos`

### Parameters

| Parameter | Values | Description |
|-----------|--------|-------------|
| `prompt` | string | Text description of the video (required) |
| `input_reference` | file | Optional image that guides generation |
| `model` | `sora-2`, `sora-2-pro` | Model to use (default: sora-2) |
| `seconds` | `4`, `8`, `12` | Video duration (default: 4) |
| `size` | `720x1280`, `1280x720`, `1024x1792`, `1792x1024` | Output resolution |

### Important Notes

- **Image dimensions must match video size exactly** - the script auto-resizes
- Video generation takes 1-3 minutes typically
- Videos expire after ~1 hour - download immediately

## Usage

```bash
# Basic text-to-video
uv run ~/.clawdbot/skills/sora/scripts/generate_video.py \
  --prompt "A cat playing piano" \
  --filename "output.mp4"

# Image-to-video (auto-resizes image)
uv run ~/.clawdbot/skills/sora/scripts/generate_video.py \
  --prompt "Slow dolly shot, steam rising, warm lighting" \
  --filename "output.mp4" \
  --input-image "reference.png" \
  --seconds 8 \
  --size 720x1280

# With specific model
uv run ~/.clawdbot/skills/sora/scripts/generate_video.py \
  --prompt "Cinematic scene" \
  --filename "output.mp4" \
  --model sora-2-pro \
  --seconds 12
```

## Script Parameters

| Flag | Description | Default |
|------|-------------|---------|
| `--prompt`, `-p` | Video description (required) | - |
| `--filename`, `-f` | Output file path (required) | - |
| `--input-image`, `-i` | Reference image path | None |
| `--seconds`, `-s` | Duration: 4, 8, or 12 | 8 |
| `--size`, `-sz` | Resolution | 720x1280 |
| `--model`, `-m` | sora-2 or sora-2-pro | sora-2 |
| `--api-key`, `-k` | OpenAI API key | env var |
| `--poll-interval` | Check status every N seconds | 10 |

## API Key

Set `OPENAI_API_KEY` environment variable or pass `--api-key`.

## Prompt Engineering for Video

### Good prompts include:

1. **Camera movement**: dolly, pan, zoom, tracking shot
2. **Motion description**: swirling, rising, falling, shifting
3. **Lighting**: golden hour, candlelight, dramatic rim lighting
4. **Atmosphere**: steam, particles, bokeh, haze
5. **Mood/style**: cinematic, commercial, lifestyle, editorial

### Example prompts:

**Food commercial:**
```
Slow dolly shot of gourmet dish, soft morning sunlight streaming through window, 
subtle steam rising, warm cozy atmosphere, premium food commercial aesthetic
```

**Lifestyle:**
```
Golden hour light slowly shifting across mountains, gentle breeze rustling leaves, 
serene morning atmosphere, premium lifestyle commercial
```

**Product shot:**
```
Cinematic close-up, dramatic lighting with warm highlights, 
slow reveal, luxury commercial style
```

## Workflow: Image → Video

1. Generate image with Nano Banana Pro (or use existing)
2. Pass image as `--input-image` to Sora
3. Write prompt describing desired motion/atmosphere
4. Script auto-resizes image to match video dimensions

## Output

- Videos saved as MP4
- Typical file size: 1.5-3MB for 8 seconds
- Resolution matches `--size` parameter