OpenAI Sora

OpenAI Sora is a text-to-video model capable of generating high-quality video clips up to 20 seconds long. It also supports image generation (stills). Sora is accessed through sora.com or integrated into ChatGPT.

Platform Overview

• Provider: OpenAI
• Specialty: Video generation, with image stills
• Access: Requires ChatGPT Plus or Pro subscription
• Supported Output Types: Video, Image

Lazer Detection

URL Patterns

The extension detects Sora on these domains:

• https://sora.com
• https://sora.chatgpt.com
• https://chat.openai.com/...sora
• https://chatgpt.com/...sora

What Is Auto-Detected

When you generate on Sora, Lazer extracts:

• Prompt - From the chat-style textarea or generation card
• Model - "Sora" or version variant (when displayed)
• Aspect Ratio - From ratio selector buttons (e.g., 16:9, 1:1, 9:16)
• Duration - From duration control (if visible)
• Resolution - From resolution selector
• Output URL - Video <video> element src or download link
• Thumbnail - Video poster attribute

Detection Quality

• Prompt extraction: High - Works in both chat and dedicated creation views
• Settings extraction: Medium - Some settings hidden in UI
• Output extraction: High - Reliable video and image URL detection

Capture Workflow

Standard Workflow

1. Open Sora generation page
2. Write your prompt in the textarea
3. Adjust aspect ratio and duration settings
4. Click "Generate"
5. Wait for video to complete (progress bar)
6. Open Lazer extension side panel
7. Detection banner shows "OpenAI Sora" with green dot
8. Click "Auto Fill" in Capture mode
9. Review filled fields (prompt, outputUrl, settings)
10. Click "Save Capture"

Multiple Variations

Sora generates one video at a time. For multiple variations:

1. Generate first variation
2. Capture as above
3. Edit prompt slightly
4. Generate again
5. Click "Refresh Detection" in extension
6. Click "Auto Fill" again
7. Capture second variation

Stills (Images)

If you generate stills instead of video:

1. Asset type is auto-detected as "IMAGE"
2. Image URL is extracted from <img> elements
3. All other fields work the same

Capture Tips

Best Practices

• Wait for completion - Don't capture while "Generating..." indicator is active
• Check video quality - Sora sometimes generates lower quality on first attempt
• Save thumbnails - Thumbnail URLs are critical for grid preview in web app
• Note model version - Model field may show "Sora" or version (e.g., "Sora Turbo")

Common Issues

Prompt not detected:

• Sora hides the input after generation in some UIs
• Manually copy prompt text before generating
• Or extract from generation card if displayed

Video URL expired:

• Sora uses temporary CDN URLs (oaiusercontent.com)
• Extension syncs immediately to avoid expiry
• Verify sync success in Queue mode

Duration not captured:

• Duration control may be hidden after generation
• Manually note video length and add to metadata field

Rights and Licensing

Ownership

Per OpenAI's terms:

• ChatGPT Plus/Pro users - You own the outputs you generate
• Commercial use - Allowed under your plan terms
• Attribution - Not required, but recommended

Restrictions

• Outputs must comply with OpenAI Usage Policies
• Cannot use for illegal or harmful content
• Subject to content policy violations and safety filters

Lazer Rights State

The extension sets rightsState to:

• LICENSED - If capturing with valid subscription (inferred)
• UNKNOWN - If plan status cannot be determined

You can manually update rights state in the web app after reviewing your plan terms.

Known Limitations

UI Changes

Sora's interface updates frequently:

• Detector uses multiple fallback selectors
• Report detection failures with screenshots

Chat Integration

When Sora is accessed via ChatGPT:

• Prompt is in the chat message, not a dedicated input
• Detection extracts from last user message
• Settings may be in conversational responses rather than UI controls

Temporary URLs

Sora video URLs are temporary:

• CDN links expire after several hours
• Lazer fetches and stores immediately
• Always verify successful sync in Queue mode

Batch Generation

Sora does not support batch generation:

• Must generate and capture one at a time
• Use prompt reuse from History for iterations

Metadata Fields

Sora-specific metadata captured in JSON:

{
  "duration": 5.2,
  "aspectRatio": "16:9",
  "resolution": "1080p",
  "modelName": "Sora",
  "platform": "openai-sora"
}

Example Capture

A typical Sora capture in Lazer:

• Platform: openai-sora
• Asset Type: VIDEO
• Prompt: "A cat wearing sunglasses walks through a futuristic city at sunset"
• Model: Sora Turbo
• Output URL: https://cdn.oaiusercontent.com/video_abc123.mp4
• Thumbnail: https://cdn.oaiusercontent.com/thumb_abc123.jpg
• Metadata: {"duration": 5, "aspectRatio": "16:9"}
• Status: GENERATED
• Rights State: LICENSED

Troubleshooting

Detection Not Working

• Verify you're on a generation or results page (not homepage)
• Check that prompt textarea is visible or generation card is present
• Click "Refresh Detection" after page loads
• Grant host permissions if prompted

Output URL Not Found

• Wait for video to fully load (check progress bar)
• Look for download button and manually copy link
• Right-click video element and "Copy video address"

Sync Failures

• Check that output URL is valid (starts with https://)
• Verify API token is valid and not expired
• Check Queue mode for error messages
• CDN URLs may expire quickly - capture immediately after generation

Related Resources

• OpenAI Sora Homepage
• Sora Technical Report
• ChatGPT Plus

Next Steps

• Capture workflow guide
• Understanding rights and licensing
• Compare Sora with other platforms