A better screen is the room
Guillaume Ardaud's Lightbulb Computer combines projection, computer vision, voice, and pointing. Its demos place recipes on a kitchen counter, maps on a table, and information on physical objects. Ardaud says the videos came from a real but bulky prototype, not a finished consumer device.[1]
In the Hacker News discussion, Ardaud described the current setup as custom software on a Mac, a consumer 4K laser projector, and a basic webcam. Apple's hand-tracking framework handles the gesture input. He also called it a research and design prototype, with the experience worked out before the hardware is reduced to a product shape.[2]
The shared view is the breakthrough. Shared sensor state has to come with it.
The bulb has history
MIT Media Lab's 1998 I/O Bulb proposed the same essential pairing. It projected high-resolution information and collected live video from the surface below. The larger Luminous Room concept coordinated several bulbs so the room itself became an interface.[3]
The old idea has new consequences. Modern vision models can turn a camera frame into text, labels, locations, or actions. A useful projection can be obvious to everyone while the frame's destination and lifetime remain invisible. A bright image on the table does not explain whether analysis happened in the fixture, on a nearby computer, or across a network.
Put the contract on the fixture
The Lightbulb Computer article calls privacy a first-order concern and names on-device processing, a hard-wired camera indicator, and a physical cover as possible safeguards.[1] In the discussion, Ardaud said he was more interested in a local-processing, open-source version than a large-platform version.[2]
That direction needs controls that a visitor can read without opening an app. Use a light that is electrically tied to the sensor power path. Label whether analysis stays local or crosses a network. Print the retention limit beside the status. Put a physical shutter or cutoff on the housing, then make its effect visible on the projected surface.
NIST's IoT device baseline gives useful engineering anchors. It treats device configuration, data protection, restricted access to interfaces, software updates, and cybersecurity state awareness as separate capabilities. It also says the baseline is a starting point, not a complete design for every use case.[4] A room camera needs the same separation in plain sight.
Prototype the social contract too
The projector prototype tests whether voice, pointing, and shared surfaces feel good. The next prototype should test whether a guest can tell when the camera is live, whether a child can stop it, whether a network outage changes the state, and whether deletion finishes when the light says it did.
The shutterboard below drafts that test contract. It does not control a camera, inspect network traffic, prove local processing, or delete data. Its completed state means four review sections were selected for a real hardware and software test.