This repository contains a NextJS sample app built on top of the Sora Video API and OpenAI SDK. It provides a simple UI for experimentation, using text prompts and optionally image inputs to generate and remix videos.

Features:

• Generate new videos or remix existing runs through the /videos endpoints.
• Queue multiple variations in one click with Versions Count and follow progress via automatic polling.
• Auto-title runs so finished clips stay recognizable.
• Automatically optimize prompts.
• Generate starter imagery with gpt-image-1 to use as image inputs.
• Preview, download, remix, or retry completed runs directly from the history panel.
• Persist preferences and run history in localStorage so refreshes do not wipe context.

This app is meant to be used as a starting point to build Sora-powered video experiences that you can customize to your needs.

1. Get access to the Sora Video API:

   • Sora is currently in limited availability. Request access through your OpenAI account team or the Video API waitlist.
   • Once approved, review the video generation guide to understand supported parameters and limits.

2. Set the OpenAI API key:

   2 options:

   • Export the OPENAI_API_KEY environment variable globally on your machine.
   • Create a .env.local file at the project root (see below)

   OPENAI_API_KEY="your-openai-api-key"

3. Clone the repository:

   git clone https://github.com/openai/sora-starter-app.git

4. Install dependencies:

   Run in the project root:

   npm install

5. Run the app:

   npm run dev

   The app will be available at http://localhost:3000.

This sample app highlights common Sora workflows and wraps each one in a small UI helper to speed up iteration.

• Fill in your prompt, choose a Sora model (sora-2 or sora-2-pro), aspect ratio, and duration (4s, 8s, or 12s).
• Optionally upload an image; the app crops it to the chosen aspect ratio before sending it along as input_reference.
• Submit one or many variations at once via Versions Count. The sidebar polls /videos/{id} until each job completes.

• Paste an existing video_id into the Remix section or click Remix on a completed run.
• Adjust prompt, model, size, or duration and resubmit. The history keeps both parent and remix entries linked for fast comparison.

Use Generate prompt to call the Responses API to optimize your prompt for the selected model, size, and duration.

• Let gpt-image-1 propose stills aligned with your prompt. Choose the one you like and send it as the next input_reference.
• Swap between generated and uploaded imagery without losing form state.

• Preview, download, remix, or retry finished runs without leaving the page.
• Remove unwanted runs from the history and bulk download completed clips.

• Enter a descriptive prompt such as A sweeping drone shot over neon-drenched Kyoto at night.
• Select sora-2-pro, set duration to 8 seconds, and bump Versions Count to 3.
• Click Generate. Watch each run stream through queued → processing → completed states, then preview the outputs inline.

• Pick your favorite completed clip and hit Remix.
• Add guidance like with a vintage film grain feel and submit. Compare the remix to the original using the preview overlay navigation.

• Click Suggest prompt after entering a rough idea—e.g., A playful golden retriever in the snow—and apply the refined version.
• Ask for image ideas, select one of the generated stills, and run a new generation to see how the reference shapes the motion.

You are welcome to open issues or submit PRs to improve this app, however, please note that we may not review all suggestions.

This project is licensed under the MIT License. See the LICENSE file for details.