Designer Bjørn Karmann created an unusual art project called Paragraphica. It’s a compact device that mimics a camera, but it doesn’t have a lens.
The “camera” works by collecting data about its location using open APIs. By combining information about the address, weather, time of day, and nearby locations, Paragraphica creates a paragraph of text that describes in detail the current location and moment of shooting. Using artificial intelligence to convert text into images, the device creates a “photo”.
Introducing – Paragraphica! 📡📷
A camera that takes photos using location data. It describes the place you are at and then converts it into an AI-generated "photo".
— Bjørn Karmann (@BjoernKarmann) May 30, 2023
The resulting image is not just a snapshot, but a complex representation of where the user is and how the AI model “sees” him.
Location, text description and Paragraphica generated image:
Paragraphica is a completely hand-assembled device that is built on a Raspberry Pi 4 using custom electronics, a touch screen and a 3D printed housing.
In place of the lens in the “camera” there is an antenna, the design of which resembles the antennae of a mole.
“The star-nosed mole, which lives and hunts underground, finds light useless. Consequently, it has evolved to perceive the world through its finger-like antennae, granting it an unusual and intelligent way of “seeing”, writes Björn Karmann.
In addition to the touch screen, Paragraphica is controlled using three rotary dials.
The first dial behaves similar to the focal length in an optical lens, but instead controls the radius (in meters) of the area in which the camera searches for locations and data. The second dial can be compared to film grain, as a value between 0.1 and 1 creates a noise grain for the AI image diffusion process.
And the third dial adjusts the aiming scale. Increasing the aiming forces the AI to follow the paragraph more precisely. By analogy with a traditional camera, the higher the value, the “sharper” and the lower, the “blurrier” the photo corresponding to the focus.
The author of the project used Noodl, Python code and Stable Diffusion API to make the Paragraphica software work. He also launched a site where you can “take a picture” of your location directly through the browser, but now it cannot cope with the influx of visitors and works with interruptions.
While Bjørn Karmann does not plan to turn the device into a commercial product, there is a possibility that other manufacturers may be interested in the idea.