Sora User Guide

Product Overview

Sora is OpenAI's text-to-video model. It generates high-quality, coherent video from text descriptions or static images. Key features include:

• Long single-clip duration
• Strong physics and motion
• Support for multiple resolutions and aspect ratios

Access Methods

1. API Access

Developers can call Sora via the OpenAI API. Typically you need:

• A valid OPENAI_API_KEY
• Sora access enabled on your account
• An SDK or client that supports Sora

2. Basic Request Format

# Example: Text-to-video request (check official docs for exact parameters)
import openai
client = openai.OpenAI()
response = client.videos.generate(
    model="sora",
    prompt="An orange cat napping on a sunny windowsill, occasionally blinking, with leaves swaying gently outside",
    duration=5,  # seconds
    aspect_ratio="16:9"
)

3. Image-to-Video

If you have a static image, use it as the starting point for motion:

response = client.videos.generate(
    model="sora",
    image=open("scene.png", "rb"),
    prompt="Clouds drift slowly, water ripples gently"
)

Parameter Reference

| Parameter | Description | |----------------|--------------------------------------| | prompt | Text description, core input | | duration | Video duration in seconds | | aspect_ratio | Aspect ratio, e.g. 16:9, 9:16 | | resolution | Resolution (if supported) | | image | Input image for image-to-video |

Check OpenAI's official documentation for exact parameters; they may change with updates.

Usage Tips

Prompt Writing

• Be specific: Clearly define subject, action, environment, lighting
• Keep it concise: Avoid long or contradictory descriptions
• Split when needed: Break complex scenes into multiple generations

Duration and Resolution

• Start with shorter clips (e.g., 5 seconds) when experimenting
• Choose aspect ratio based on platform (16:9 for landscape, 9:16 for portrait)

Errors and Retries

• If generation fails, simplify the prompt or adjust parameters and retry
• Check API error codes and messages for permission, quota, or other issues

Workflow Example

Plan: Define theme, style, pacing

Write prompt: Organize as "subject + action + environment + style"

Call API: Send request and wait for generation

Review: Download video, assess quality and coherence

Iterate: Refine prompt or parameters based on results

Summary

Sora provides powerful text-to-video and image-to-video capabilities via API. Understanding access, parameters, and prompt techniques helps you generate video that meets your expectations more efficiently.