Image-Blaster open source: turn a photo into a 3D scene, Claude’s skills can be done in 5 minutes

Core conclusion

On May 15, 2026, the open source project Image-Blaster appeared on the homepage of Hacker News, gaining 125 points of popularity. This project is a picture-to-3D scene conversion tool based on Claude's skills, which can automatically generate explorable 3D environments, mesh models, physical sound effects, and Gaussian Splatting scenes within 5 minutes of inputting an image. It currently has 2057 stars on GitHub.

Key Points

• Event Time: 2026-05-15 (HN Popular), project development started on 2026-04-21
• Core Capabilities: From single image → 3D environment (.glb/.obj model + .spz scene + .mp3 sound effect)
• Technology Stack: Claude Skills + World Labs Marble + Hunyuan 3D + ElevenLabs SFX
• Meanings for automation practitioners: AI is expanding from "writing code" to "creating 3D assets", and the boundaries of AI Agent's capabilities are rapidly expanding.

Background and Project Overview

Image-Blaster was created by developer neilsonnn and is an open source Claude skill. Users only need to provide an image, and Claude will automatically call multiple AI models to complete the complete process from environment modeling, object extraction to sound effect generation.

Underlying logic: It is essentially an orchestration layer that connects multiple professional AI models into an automated "picture → world" pipeline. Each link is handled by the best models: World Labs Marble to build the environment, Hunyuan 3D to build objects, and ElevenLabs SFX to generate sound effects.

The project is open source based on the MIT license, the code is hosted on GitHub, and the main language is TypeScript.

Core function disassembly

Implications for AI automation practitioners

Image-Blaster represents three important trends in the AI Agent capability paradigm:

1. Model orchestration becomes core competitiveness

Image-Blaster is essentially a practical application of Claude Skill. It does not model or render by itself, but serves as an "AI orchestrator" to string together professional models such as World Labs, Hunyuan, and ElevenLabs. This is completely consistent with the concept of orchestration tools such as n8n and LangGraph: the value of AI does not lie in how strong a single model is, but in how they are combined to solve practical problems.

2. The boundaries of AI capabilities are extending from programming to creative production

From writing code with Claude Code to doing 3D modeling and sound design with Image-Blaster, the "hands" of AI Agent are getting longer and longer. For content creators, game developers, and indie developers, this means one person + AI Agent can do things that used to take a team of 5 people.

3. Claude Skills ecosystem is taking shape

Image-Blaster is an implementation case of Claude Skill system. Similar to ChatGPT's GPTs, Claude Skills allows developers to write reusable toolsets and package them for Claude to use.

How to use Image-Blaster

Prerequisites: Requires Claude CLI installed and API Key (World Labs and FAL).

git clone https://github.com/neilsonnn/image-blaster
cd image-blaster
claude

# Claude API Key， input/

# blast it and confirm each step with me

3D Unity、Unreal、Godot、Blender Three.js 。

Claude、OpenAI、ElevenLabs、n8n、LangGraph

Reference sources

• Image-Blaster GitHub
• HN
• World Labs
• FAL AI

Internal link guidance

• Want to learn how to use AI Agent for automation? Watch: AI Agent-Driven Content Automation: n8n MCP Building Guide from Scratch
• Someone used Claude Code to build a complete automated money-making system: He Built an AI Automation Stack with Claude + n8n — $4K to $12K/mo in 6 Months
• Practical method to save 90% of costs with Claude Code: DeepSeek V4 vs Claude Code: 90% Cheaper, Same Quality