Project highlights

Worlds in
Action Hack

Two projects from the Los Angeles edition, each with a clip pulled from its demo. Both put a person inside a 3D space and let them decide something from in there.

Edition 02 · Los Angeles Projects 2 Builders 7

AFTERSHOCK

The decision layer between a damage grade and knowing who to send first.

AFTERSHOCK demo clip: Responder dashboard into the spatial damage map
Responder dashboard into the spatial damage map 00:35 to 00:40

What they built

On 24 June 2026 two major earthquakes struck Venezuela 39 seconds apart. Satellite damage assessments and seismic data were published quickly, but the team found the gap sat one step later: the hard part is not understanding what was damaged, it is deciding what to do next.

AFTERSHOCK is a spatial decision-support prototype built on real data from those earthquakes. Three agents each rank the same buildings differently. A medical agent orders them by structural damage severity for search and rescue. A machinery agent combines severity with spatial clustering, so equipment gets deployed where it can cover the most ground. A coordinator agent reconciles the two when they disagree.

The result is an interactive dashboard where a responder switches between perspectives and sees why each ranking came out the way it did, rather than being handed one universal priority list and asked to trust it.

LA48 CinemaScout

A spatial cinematography copilot for planning, previewing, and refining camera shots inside interactive 3D environments.

LA48 CinemaScout demo clip: Placing a virtual cinema camera inside the reconstruction
Placing a virtual cinema camera inside the reconstruction 00:16 to 00:20

What they built

Location scouting costs travel, repeat site visits, manual measurement and a lot of coordination before a single frame is shot. CinemaScout replaces the first passes of that with a reconstruction you can walk around inside at real-world scale.

The workspace is built on 3D Gaussian splat reconstructions. A user positions virtual cinema cameras, adjusts focal length, field of view and aspect ratio, previews the shot from any viewpoint, saves camera positions, and can ask for AI-assisted shot suggestions. A companion Mission Control dashboard collects the saved shots so a crew can review and organise them together.

The accomplishment the team singles out is rendering and controlling a virtual camera from inside a VR headset, with spline-based movement through the reconstruction, so a tracking shot can be previewed spatially instead of on a desktop screen.