AI Briefing
KO

image-blaster, 3D Environment Generation

·2026.08.04 09:30

Key point

image-blaster converts a single image into a 3D environment, objects, and sound effects.

1 / 2

Details

image-blaster is an open-source working environment that uses Claude Code skills and agents to generate an explorable 3D environment, individual 3D objects, and sound effects from a single image.

After cloning the repository, running claude, and placing an image in the input/ directory, it automates the step-by-step process from image analysis to result organization. Actual generation uses the World Labs and FAL APIs.

The main processing flow is as follows.

  • Analyze the image for separable object candidates
  • Generate a clean plate with objects removed
  • Generate a static environment and Gaussian Splat using World Labs marble-1.1
  • Generate individual 3D meshes using FAL's hunyuan-3d
  • Generate environmental ambience and collision sounds using ElevenLabs SFX
  • Check results in a React/Three.js-based viewer

The outputs are provided as .spz for the environment and .glb for colliders, .glb/.obj for objects, and .mp3 for audio. They can be used in Unity, Unreal, Godot, Blender, Maya, and Three.js projects.

However, the background is generated as a Gaussian Splat rather than a mesh, and the reproducibility of generation results or precise dimensions and topology are not guaranteed. World Labs and FAL accounts and API costs are also required, making it more suitable for early-stage concepts for games and film, and building robot simulation environments, rather than for precise CAD work or fully local execution.

This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.

Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.