Virtual window with a magnetic map selector

Oct 13, 2026 Last reply: 1 day ago 4 Replies
Jump to replies (4)

Most virtual window builds I have come across rely on a button panel or a phone app to switch feeds, and that always felt like the weak point of the idea. This one takes a different route: a framed world map mounted in the hallway acts as the control surface, and a model airplane placed on the map picks the location. The window frame itself sits on a door, so the selection and the display live in separate spots, which is a layout I had not seen before.



Under the map are five magnetometer modules, each with an indicator LED, and the D1 Mini behind it is an ESP8266 development board fed by micro-USB. Since the sensors run at 5V and the microcontroller at 3.3V, an eight-channel bidirectional logic level converter bridges the two. Move the plane to a spot, the magnetometers pick it up, the LED at that location blinks, and the stream changes to that city's live feed and audio. The five presets are Las Vegas, Florida, Amsterdam, Verbier and Shinjuku.



The part that stands out to me is the thermal sensor, tucked into a small framed picture labeled FLAMBOROUGH HEAD LIGHTHOUSE above the control area. It tracks where the viewer stands and applies a perspective parallax shift to the image, which should make the illusion far more convincing than a flat fixed feed.



On the software side, a Linux server scales and hosts the streams, VLC handles scaling and rotation through a Python script, MQTT carries the messages, and an Android app on a TV plays everything back. All the code is in a GitHub repository, with a process flow diagram on the maker's website. One open point: adding a stream is reportedly a couple of lines of code, but the maker says the hardest part was cutting a hole in the map, so scaling up the location count may be limited by that step.



Project by Alex Shakespeare. Watch it on their YouTube channel:

formatting link


Using a model airplane as the selector is a genuinely clever choice in my view. It gives the control surface a physical, tactile quality that buttons never have, and the plane visually doubles as the pointer on the map. The trade-off is precision, since five magnetometer zones only give you five presets, but for this use that seems like plenty.

How does the signal chain actually work between the map and the screen? With sensors at 5V and the ESP8266 at 3.3V, I am curious whether the level converter sits directly on each magnetometer line, and how the MQTT messages then reach the server side that feeds the TV.

From what the maker has shared, the five magnetometer modules sit behind the map, and an eight-channel bidirectional logic level converter connects the 5V sensor signals to the 3.3V logic of the ESP8266 board. MQTT is the protocol used for communication, a Linux server scales and hosts the stream, and an Android app on a TV plays it back. There is also a process flow diagram on the maker's website if you want the full picture.

An idea worth considering: since the thermal sensor already tracks viewer position for the parallax effect, the same readings could drive a smoother zoom or pan on the current feed, so the window reacts continuously rather than only when the plane moves. That would build on hardware already in the frame without needing extra sensors behind the map.

Join the Discussion

Have something to add? Share your thoughts — no account required.

Didn't find your answer?

Ask the community — no account required