banner

Blog

May 24, 2023

Check Out This Raspberry Pi Camera That Takes Photos With AI Instead of Lens

What comes to your mind when you think of a Raspberry Pi camera? A Pi connected to the Global Shutter Camera, HQ Camera, or Camera Module 3, correct? You might even think of the Otto a GIF creation camera, which dates back to 2015. Well, the latest project is a lot different than what you might have seen in the past.

Created by Bjørn Karmann, the Raspberry Pi camera project does not come with any lens. In fact, the creator does not even call it a camera. Instead, it goes by Paragraphica. Why does it have an unconventional name? It utilizes AI to generate an image!

You might be unfamiliar with what traditional Raspberry Pi cameras are and how they work. Well, at the core, these setups consist of a Pi module with a camera sensor. You can find a barebone kit and build one for yourself, which will offer you the ability to capture photos and videos.

When it comes to features, a traditional Raspberry Pi camera can typically take VGA, 720p, and 1080p videos. And the high-resolution sensors for Pi can take high-quality photos as well.

However, the features will differ from one module to another. And in case you are wondering, there are several Pi camera models available officially from Raspberry. That includes the original 5MP module, 8MP camera module 2, and the latest camera module 3. Each one of these modules brought upgrades over the predecessor and made DIY Raspberry Pi Cameras more relevant for current applications.

That said, these traditional setups utilize the Raspberry Pi's Camera Serial Interface for connection. That bus connector integrates with a flexible ribbon cable. And to make the whole setup work, you will need to use the built-in terminal or Python (a coding language).

In short, these traditional setups are basically DIY cameras. You can use them for a variety of projects. That includes live camera streaming, time-lapse photography, and home security modules. And the great part is that there are tons of in-depth guides available, which makes it easier for beginners to get into such projects.

At first glance, the project by Bjørn Karmann will seem nothing more than a classic camera. But instead of a lens, the Raspberry Pi Camera has a huge cover, which looks like a nose of a star-nosed mole. For the viewfinder, Bjørn Karmann has utilized a screen where you can adjust the settings for getting the AI-generated picture.

The camera also comes with a series of dials, which you can adjust traditional camera parameters. For example, with the dials of this Raspberry Pi camera, you can adjust the film grain of the picture. They also allow you to tune the sharpness of the AI-generated picture.

But the magic happens only when you press the capture button. This button triggers the AI program and makes it generate a photograph based on the parameters you have selected.

The data that this Raspberry Pi camera utilizes is from various open APIs. That includes information such as the location, time, date, and nearby landmarks. The Paragraphica utilizes all of these pieces of information to process the final photo. And as you might have guessed, it relies on a text-to-image AI for the image.

Now, the finalized picture this Raspberry Pi camera offers differs from what you would get from a regular camera. The Paragraphica will not include any subject that might be right in front of you. Instead, the final image will contain the information specifically provided in the prompt.

In other words, this Raspberry Pi camera is not an alternative to a traditional camera. After all, it does not have a lens to capture the scene in real time. Instead, it is more for getting AI-generated images of the area you are in. As Bjørn Karmann, the creator, states, the "camera offers a (new) way of experiencing the world around us, one that is not limited to visual perception alone."

To power the project, Bjørn Karmann has utilized a Raspberry Pi 4. He created a custom 3D-printed shell to house the components inside. Paragraphica utilizes a touchscreen for the viewfinder that enables the user to customize the input. And to get location data, this project has a GPS module inside.

When it comes to the software, this Raspberry Pi camera runs on a code that's written by Karmann. According to him, the software utilizes Stable Diffusion API and Python to create the final image. Everything works together through Noodl, which is a visual tool that developers use to create web apps.

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".

See more here: https://t.co/Oh2BZuhRcfor try to take your own photo here: https://t.co/w9UFjckiF2 pic.twitter.com/23kR2QGzpa

— Bjørn Karmann (@BjoernKarmann) May 30, 2023

You will find the detailed diagram of the project and a breakdown of the process at Karmann's official website. And if you want to create something similar with a Raspberry Pi, Karmann has a dedicated blog post that you can check out. Alternatively, you can test the software by yourself in a browser through his website. Yes, you do not need any hardware to get a taste of this brand-new experience.

SHARE