GitHub repo image-blaster turns one image into an explorable scene

A viral X thread surfaced the GitHub repo, which claims meshes with physics, background splatting and ambient audio from one image in about five minutes.

By ·

Why it matters

Single-image to 3D pipelines, if reliable, collapse content-creation costs for games, AR/VR and virtual production. Open-sourcing invites fast iteration and forks from the developer community.

GitHub repo image-blaster turns one image into an explorable scene

Neilson (@neilsonks) has open-sourced image-blaster, a GitHub repo that takes a single photo and generates an explorable 3D scene. The project was spotlighted in a thread on X from @CopyRebeldia.

https://x.com/CopyRebeldia/status/2055083368870817818

According to that thread, the repo outputs meshes with physics, a background splat, and ambient audio, aiming to go from one input image to a navigable world in roughly five minutes. The post framed it simply: "Una imagen entra. Un mundo sale," attributing the original work to Neilson and linking the code.

The post has drawn more than 500,000 views on X at time of writing, with developers in the replies asking how to try it. The code is publicly available on GitHub; RuntimeWire has not independently evaluated the repo or its performance.

If the pipeline works as described, it could compress a workflow that typically spans photogrammetry, manual modeling, and sound design into a single automated pass, opening up rapid world-building for game prototypes, AR/VR scenes, and interactive demos.

Reader comments

Conversation for this story loads after sign-in.