scroll-world  by oso95

AI-powered 3D brand worlds

Created 5 days ago

New!

533 stars

Top 58.6% on SourcePulse

GitHubView on GitHub
Project Summary

This project provides an agent skill to generate immersive, scrollable 3D landing pages, transforming any brand or industry into a continuous "fly-through" world. Targeted at users of Claude Code, Codex, and other SKILL.md-compatible agents, it offers a novel way to create engaging web experiences akin to high-end product showcases. The core benefit is an AI-driven, seamless visual narrative that unfolds as the user scrolls.

How It Works

The skill leverages the Higgsfield platform for AI asset generation, utilizing GPT Image 2 for cohesive isometric scenes and Seedance for image-to-video camera flights. A custom, framework-agnostic vanilla-JS scrub engine drives the user experience, translating scroll position into camera movement through generated scenes and precisely matched connector clips. This approach ensures a continuous, cut-free flight, with the engine designed for easy integration into any web stack, including plain HTML, Next.js, Vue, or Python-served pages.

Quick Start & Requirements

  • Installation: Can be installed as a plugin for Claude Code (/plugin install scroll-world@scroll-world), via the skills CLI for Codex and others (npx skills add oso95/scroll-world), or manually by cloning the repository and copying the skill folder.
  • Prerequisites: Requires the Higgsfield CLI (authenticated with credits), ffmpeg/ffprobe for media processing, and optionally Python 3 with Pillow for background knockout.
  • Resource Footprint: Asset generation incurs Higgsfield credits and significant processing time.
  • Links: Project repository: https://github.com/oso95/scroll-world

Highlighted Details

  • AI-powered generation of isometric scenes and dynamic camera paths.
  • Framework-agnostic JavaScript engine for flexible web integration.
  • Seamless, continuous camera flight achieved through frame-identical scene connectors.
  • Automated pipeline for generating stills, dive-in clips, and transition assets.

Maintenance & Community

No specific details regarding maintainers, community channels (like Discord/Slack), or roadmap were provided in the README snippet.

Licensing & Compatibility

  • License: MIT License.
  • Compatibility: The MIT license permits broad usage, including commercial applications and linking within closed-source projects.

Limitations & Caveats

Asset generation is dependent on Higgsfield credits and can be time-consuming, with the skill managing background processes and polling for completion. The Higgsfield CLI must be authenticated and possess sufficient credits to operate.

Health Check
Last Commit

1 day ago

Responsiveness

Inactive

Pull Requests (30d)
1
Issues (30d)
0
Star History
532 stars in the last 5 days

Explore Similar Projects

Starred by Eric Zhu Eric Zhu(Coauthor of AutoGen; Research Scientist at Microsoft Research), Luis Capelo Luis Capelo(Cofounder of Lightning AI), and
1 more.

hyperframes by heygen-com

3.4%
34k
HTML-native video rendering for AI agents
Created 4 months ago
Updated 15 hours ago
Feedback? Help us improve.