Produced by: Hitansh Gaur
Artificial intelligence is giving a tough competition to old technology every day. After taking on the jobs of writers, coders and even artists, the next target of AI seems to be photographers. A Danish artist has invented an AI-based camera which might change your photography experience completely and it’s called ‘Paragraphica’.
Touted as the world’s first context-to-image device, Paragraphica visualises an image with the help of AI and the location data. The device is available in the physical form and virtual form as well for the users to try.
According to its website, this AI-based camera collects the location data with the help of APIs (application programming interfaces). Then it uses information like the weather, time of the day and the places around the location. It then writes the paragraph based on all this information.
Paragraphica then uses the text-to-image AI and converts the paragraph into a photo. But it is not just an AI-generated image, instead the website suggests that it is a reflection of how the AI sees that place. The images carry the same tones and moods but they are not exactly similar to the place.
The physical device consists of a viewfinder displaying real-time descriptions of the location where the user is at that point of time. Once a user presses the trigger, it will create a scintigraphic interpretation (a process that generates scans of the human body interior) of the paragraph written in the camera.
The physical AI camera also has three dials attached to it, which help the user to control the location data and AI parameters to change the appearance of the images. This process makes it somewhat similar to how traditional cameras work.
The first dial is used to decide the radius in metres for the area in which the camera should search for places and data. The second dial is to determine the noise seed necessary for AI image diffusion. The third dial is to change the scale of guidance. Higher guidance will mean more in-depth assessment of the paragraph to create a more accurate and appropriate image.
The camera has a system based on Raspberry Pi 4, one large touchscreen and a 3D printed housing. The camera also has custom electronics. On the virtual front, the Paragraphica software is backed by Python Code, Noodl and Stable Diffusion API.
The physical camera has been created by a Danish artist named Bjørn Karmann. He unveiled the prototype of the Paragraphica camera on May 30.