🎞️ Seedance-2-prompts-datasets

🎞️ The ultimate Seedance-2 video prompt dataset (12GB+). 2000+ video generation prompts with full metadata and preview frames. Truly open source: No login, no ads, no redirection. Just pure data for AI video creators.

This project is a massive collection of prompts used for Bytedance's Seedance 2.0 and the resulting generated videos. The entire dataset exceeds 12GB and contains 2000+ videos, all structured into a comprehensive dataset.

Due to GitHub's limitations with large file storage, the full dataset is hosted on Hugging Face. The Hugging Face repository contains the generated videos (.mp4), cover images (.jpg), and a highly structured .jsonl file that holds all prompt metadata.

Download the full dataset here:
https://huggingface.co/datasets/GokuScraper/seedance-2-prompts-datasets

🌐 Online Viewer

No login required, lighting-fast response.
👉 View Online

📖 Introduction

Launched by GokuOpenLab, seedance-2-prompts-datasets is a prompt data infrastructure project created for developers and researchers.

In the current AI ecosystem, prompts are the new "productivity interface". However, reality reveals several issues:

• Prompt data is highly fragmented.
• There is a lack of unified structural standards.
• Retrieving and reusing prompts is difficult.
• They are not suitable for engineering and systematic utilization.

This project's goal is NOT just to "display prompts", but to:

Transform internet prompts into structured, calculable, and redistributable data assets.

🚫 Our Manifesto

Before building this project, we analyzed the absolute chaos in the open-source community and decided to firmly say "NO":

• NO to black-box prompt distribution: Closed systems that don't provide structured data and forbid secondary usage.
• NO to traffic-driven pseudo-open source: Repositories using GitHub as clickbait while hiding core data behind private platforms or logins.
• NO to uncomputable data formats: Simple text displays that cannot be parsed by programs or used for model training.

✨ Why Goku Prompt Hub?

1️⃣ 100% Open Data

All collected data is fully accessible. There are no "preview versions" or locked contents.

• Direct download via Hugging Face.
• Seamless online browsing.

2️⃣ Structured Data System (.jsonl)

Our prompt data is stored in a structured JSONL format. Every single prompt undergoes unified parsing and standardization, connecting the video assets to their exact configurations.

Example entry from our JSONL records:

{
  "version": "1.0",
  "id": "SD2_00133",
  "category": "Entertainment",
  "is_featured": false,
  "date": "2026-04-28",
  "slug": "glacial-tiger-vs-frost-serpent",
  "model_info": { "name": "seedance", "version": "2.0" },
  "raw_p": "Environment: A colossal glacial canyon under pale blue twilight...",
  "media": {
    "v": "seedance-2/videos/SD2_00133.mp4",
    "c": "seedance-2/covers/SD2_00133.jpg"
  },
  "spec": { "width": 1280, "height": 720, "ratio": 1.78, "duration": 15.12, "safety_rating": "Safe for Work" },
  "i18n": {
    "zh": {
      "t": "冰谷虎蛇战",
      "p": "环境：一座巨大的冰川峡谷，笼罩在淡蓝色的暮光之下...",
      "tags": ["冰川峡谷", "冰虎", "霜蛇"]
    },
    "en": {
      "t": "Glacial Tiger vs Frost Serpent",
      "p": "Environment: A colossal glacial canyon under pale blue twilight...",
      "tags": ["ice canyon", "frozen battle", "cinematic"]
    }
  },
  "platform": "x",
  "sourceLink": "https://x.com/LudovicCreator/status/2045419585491317186",
  "file_name": "seedance-2/videos/SD2_00133.mp4"
}

3️⃣ Developer Friendly

• Unified JSON Schema: JSONL perfectly formats logs into lines of JSON objects.
• Database Ready: Seamlessly import into SQLite, Supabase, or your local AI toolchains.
• One-line Python Integration: Load the dataset directly into a Pandas DataFrame in 1 second:

import pandas as pd

# Load Goku's dataset instantly!
url = "https://huggingface.co/datasets/GokuScraper/seedance-2-prompts-datasets/raw/main/metadata.jsonl"
df = pd.read_json(url, lines=True)
print(f"✅ Loaded {len(df)} structured video prompts!")

4️⃣ Open License (CC BY 4.0)

All data operates under the CC BY 4.0 License:

• ✔ Free to use
• ✔ Commercial use allowed
• ✔ Modification allowed
• ✔ Redistribution allowed
• ❗ Attribute original source required

📊 Data Overview

• Total Prompts: 2110+
• Languages: English / Chinese
• Target Models / Engines: Bytedance Seedance 2.0, Midjourney, Stable Diffusion, DALL·E 3, Flux, etc.
• Update Frequency: Continuous automated syncs.

🛡️ Disclaimer

The prompts and metadata in this repository are sourced from public internet communities and are strictly intended for learning, research, and data structuring purposes.
The copyright of the original generated content belongs to the original creators. This project only provides data curation, structured processing, classification, and indexing; it does not claim copyright over the original context.

If you are a copyright holder and believe there is an issue, please contact us via GitHub Issues or email. This project is not affiliated with Bytedance, OpenAI, Google, Anthropic, Midjourney, or any other specific model/platform.

🤝 Contributing

Welcome to Goku Prompt Hub!

• Submit an Issue: Report bad quality or broken prompts.
• Submit a PR: Contribute your high-quality prompt data.
• Star the repo: Show your support and push prompt data infrastructure forward!

---

Happy Prompting!