This lensless AI camera uses only textual prompts to 'take' a photo

The 'Paragraphica' camera model uses geolocation and textual prompts to generate an AI image.
Sejal Sharma
Creator Bjørn Karmann holding Paragraphica
Creator Bjørn Karmann holding Paragraphica

Bjorn Karmann 

A picture is now worth only 20 words.

That’s right. There exists a new camera in the market that ‘clicks’ a photo using only the location data of the user and the textual prompts they feed the artificial intelligence (AI)-powered camera. Once the camera gets hold of all the information it needs, it processes that information and spouts out an AI-generated photo. It doesn’t even have a lens.

This lensless AI camera uses only textual prompts to 'take' a photo
An example from Bjørn Karmann's tweet

Netherlands designer, Bjørn Karmann, introduced his passion art project ‘Paragraphica’ in a tweet last week. “The prototype is fully working and has physical dials that let you control the range of data and AI image settings, similar to an analogue camera,” he said.

Similar to a traditional camera, Paragraphica has three dials to control the data and AI parameters that influence the appearance of the photo. The user can set the radius (in meters) in which the camera will search for places and data around it using the first dial. The second dial is to add or subtract noise from the picture. With the third dial, the user can decide how closely they want the camera to follow their textual instructions/prompts.

No lens since it doesn’t need light to operate

The feedback has mostly been positive, with some comments regarding privacy concerns since the camera knows the whereabouts of the person taking the photos. The camera needs the internet to operate, access weather data, and uses GPS to track the user’s location. The model, the first of its kind, is currently using Stability AI’s text-to-image Stable Diffusion model. The model generates images based on textual prompts.

The resulting "photo" is not just a snapshot but a complex and nuanced reflection of your location and perhaps how the AI model "sees" that place, said Karmann in his blog post.

This lensless AI camera uses only textual prompts to 'take' a photo
Another example of how the camera works

The photos are able to capture reminiscent emotions and moods from the location but never really look exactly like the location the camera has been fed.

Karmann clarifies that he has no intention of commercializing the product or challenging photography. His product rather questions the role of AI in a time of creative tension. 

This lensless AI camera uses only textual prompts to 'take' a photo
A flowchart showing the workings of the camera

The front of the camera looks a bit uncanny, with its red tentacles spewing out. Karmann says he drew inspiration from the star-nosed mole, which lives and hunts underground and finds light useless. The animal has evolved to perceive the world through its finger-like antennae, granting it an unusual and intelligent way of "seeing," elaborated the blog post.

Add Interesting Engineering to your Google News feed.
Add Interesting Engineering to your Google News feed.
message circleSHOW COMMENT (1)chevron
Job Board