The invention of the camera changed human society significantly. Cameras have made it possible to capture images in an objective manner. Prior to that, one had to rely on descriptions or drawings to get an idea of an object that was not directly experienced.
After the invention of the camera, people began to consider photographs as more reliable evidence. Now people can create images with artificial intelligence (AI). But deepfakes are also causing confusion. Taking advantage of this human interest in using AI, Bijoy Karman has developed a lensless camera. The camera, called Paragraphica, uses location data to generate images with the help of artificial intelligence.
When a user takes a photo, Paragraphica stores all the location information in its Stable Diffusion AI model. Paragraphica has GPS module to find current location. It then uses the OpenWeather website’s API to get local weather conditions. And using Mapbox’s API to get information about significant infrastructure features. Mapbox provides map services for various corporate websites.
With the dials above Paragraphica, the user can adjust the scope of the search, its noise seed, and a parameter determining how perfect the stable diffusion can produce an image of the desired object.
The picture that is presented as a result of combining these is actually a picture created by artificial intelligence, but it does not differ from reality. The characteristics of this image exactly match the image of the object in our memory. Hardware used in this paragraph—Raspberry Pi 4 Model B. It is basically a single board computer. There is an LCD touchscreen. Carman claims the screen is 15 inches. However, Hackster’s report mentions 5 inches. Maybe it’s a typo. The devices are housed in a 3D-printed box that looks like an old-school Leica camera.
For general users, Carman has created a web app called Paragraphica, which acts as a virtual version of this camera. Due to the high number of users on this website, it may take longer to load.