Wild Raspberry Pi Camera Skips The Lens And Generates Photos With AI
AI-generated images have become somewhat of a norm as of late, with advancements in the technology bringing new features in photo editing such as Adobe's Generative Fill tool. Generative Fill can take an existing image, and then using AI create an expansive background surrounding it. Paragraphica is a context-to-image camera that takes AI-generated content to another level, by producing an image without any context other than location data and a few other prompts. To add to its allure, it can be used as a physical device, or as a virtual camera that you can try out now.
The physical device was designed and built by Bjorn Karmann and is a 3D-printed housing that consists of 3 dials, a 15-inch touchscreen, and custom electronics. In terms of software, it uses a combination of Noodl, python code, and the Stable Diffusion API.
The three dials are used to control the data and AI parameters in order to influence the appearance of the photo along with information such as weather and time of day. The first dial is similar to the focal length in an optical lens, however, it controls the radius of the area the camera searches for. The second dial is said to be comparable to film grain, and the third dial controls the guidance scale, or how closely the camera follows the paragraph it is given.
As far as the oddly shaped contraption on the front of the device, it is fashioned after a star-nosed mole and does not serve any purpose outside of a cosmetic. Karmann explained why he chose to add this to the device by stating, "This amazing animal became the perfect metaphor and inspiration for how empathizing with other intelligences and the way they perceive the world can be nearly impossible to imagine from a human perspective."
Karmann explains how the camera operates as "collecting data from its location using open APIs. Utilizing the address, weather, time of day, and nearby places." He adds that it combines all the data points and then composes a paragraph (seen below) that details a representation of the user's current place and time. The camera uses a text-to-image AI, allowing it to convert the paragraph into a "photo".
While Karmann has no plans right now of selling a consumer version of his Paragraphica camera, it can be taken for an online test drive by visiting his website. Karmann warns, however, that due to high traffic, it may not load for you, so be patient!