polym-eval-generate-sora

Generate Sora videos from text prompts or reference images via OpenAI or ByteDance AIDP.

8|Updated May 13, 2026
One-click install
npx skills add https://github.com/byteplus-sa/polym --skill polym-eval-generate-sora
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: polym-eval-generate-sora
Source: https://github.com/byteplus-sa/polym/tree/main/skills/polym-eval-generate-sora
Command: npx skills add https://github.com/byteplus-sa/polym --skill polym-eval-generate-sora

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires requests, Pillow, python-dotenv, tos, and includes scripts (resource) components.

What problem does it solve?

Generate shareable video outputs from text prompts or reference images without manually operating Sora backends or handling long-running generation workflows.

Core Features & Use Cases

  • Text-to-Video (T2V): Create videos from a natural-language prompt for Sora 2/Pro and supported AIDP models.
  • Image-to-Video (I2V) / First-Frame Reference: Animate an initial frame by passing a reference image, with automatic aspect-ratio handling.
  • Dual Backend Support: Run generation via OpenAI (public API) or ByteDance AIDP (azure-sora2 recommended; azure-sora is legacy T2V-only).

Quick Start

Run this skill to generate an 8-second Sora T2V clip by telling your agent: "Generate a video of a serene lake reflecting autumn trees at sunset for 8 seconds using polym-eval-generate-sora with the default OpenAI backend."

Frequently Asked Questions about polym-eval-generate-sora

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I generate an MP4 video from text prompts using Sora?▼

To generate an MP4 video from text prompts using Sora, you provide a natural language description and the skill calls deterministic scripts to poll completion and download the output. You can specify duration and use the OpenAI or ByteDance AIDP backend.

Can I animate a first-frame image to create a video with Sora?▼

Yes, you can animate a first-frame image to create a video with Sora by passing a reference image. The skill handles Image-to-Video generation with automatic aspect-ratio handling to produce an MP4 output.

What is the difference between OpenAI and ByteDance AIDP backends for text-to-video?▼

The OpenAI backend uses the public API for text-to-video, while the ByteDance AIDP backend recommends azure-sora2 for full functionality. The legacy azure-sora on AIDP only supports text-to-video generation.

Do I need to upload my generated Sora MP4 to TOS to get a network URL?▼

No, uploading to TOS is optional. The skill automatically downloads the generated MP4 output locally, and can optionally upload to TOS to produce a network URL for sharing.

What environment variables are required to run Sora video generation scripts?▼

Running Sora video generation scripts requires setting specific environment variables for the chosen backend. The skill enforces these required variables before calling scripts to ensure successful text-to-video or image-to-video workflows.

How do I select video duration when generating text-to-video clips?▼

You select video duration when generating text-to-video clips by specifying the length in your prompt request. The skill fits duration selection into prompt iteration workflows for both OpenAI and AIDP backends.