There was once a race to put out cameras with ever higher numbers of megapixels to snare customers eager to take the highest quality digital photos. These days, we know that things like optics, processing, and finer qualities of an image sensor are all very important beyond pure resolution. But, for a time, companies behaved as if megapixels mattered over all else.

But what if you could go farther—shooting not millions, but billions of pixels in a single image? That’s precisely what [Yannick Richter] came to Hackaday Europe to talk about, covering his Project Gigapixel build.

More Pixels

When it comes to building a consumer camera with higher resolution, manufacturers achieve this by creating an image sensor with a greater number of sensing elements. This, of course, can get expensive and difficult the farther you want to scale, particularly if you’re trying to fit more sensing elements into a given standard sensor size.

However, there are other ways to capture images in greater resolution that don’t require a larger image sensor. Namely, you can actually use quite a simple image sensor of limited resolution, and simply move it to various positions, capturing light all the while. Then, all you need to do is stitch the output together and you have a remarkably high resolution image. It might sound complicated, but as [Yannick] explains, it’s a perfectly cromulent way to build a gigapixel image.

[Yannick’s] project began with an old Epson flatbed scanner. This made the perfect donor for such a project, as it came with a linear image sensor with quite good resolution for scanning photographs and documents. The only problem is that it needs to move in a straight path in order to capture a full image. The goal was to build it into a scanner-style camera that was truly portable, which required some reverse engineering and creative design to make it into a practical tool for real-world photography. [Yannick] didn’t want to just stuff the existing scanner in a bodged-together camera body, either. He wanted to interface the sensor directly and build a custom linear-scanning camera from the ground up, with the high-resolution linear sensor mounted behind a nice medium-format Pentax lens.

The key was that the scanner in question—an Epson V370—used a Sony CCD scanning element, rather than a cheaper CIS element. A proper CCD sensor is more expensive, but produces better output, and is more suitable for the sort of imaging [Yannick] was trying to do with this build. Namely, by running the scanning element behind a medium format lens to capture incredibly high-resolution images at up to 40800 x 80000 pixels, or 3.2 gigapixels if you multiply it out.

The talk covers all the work that [Yannick] did to make this a fully functional camera. That included doing a deep dive into Epson documentation to figure out how to interface the sensor at all. Thankfully, service manuals provided enough detail on how the 12-line RGB sensor works to get the project over the line. Interfacing the sensor was achieved via reusing the ADC and timing generation hardware from the scanner itself, hooked up to a Raspberry Pi 5 and a RP2350.

Plenty of work was required to figure out how to properly offset all the R, G, and B pixels to line up properly into a coherent color image. [Yannick] also dives into the mechanical design, regarding how the sensor was assembled on a 100-millimeter linear drive to scan it behind the lens assembly to capture images. There were also issues generating a live preview that you’d get on any other digital camera, which is not exactly practical to generate from a scanning image sensor. Instead, a standard Raspberry Pi camera was included in the build for live preview to help with lining up a shot.

Perhaps the best part of the talk is when [Yannick] shows off the final results. The simple fact is that 3.2 gigapixel images capture a ton of detail when they’re taken well and focused correctly. A simple shot of some benchtop instruments doesn’t look like anything special, until [Yannick] shows that cropping in will let you read the codes off a 0603 SMD resistor. Obviously, the camera is limited when it comes to speed and it can’t really capture moving objects well. However, when it comes to grabbing very high resolution shots of still scenes, it’s a great performer. If taking gorgeously detailed landscapes or stunning architectural shots is your thing, you might find such a build an appealing proposition for your own needs.

Even for those of us that are quite technically minded, we spend precious little time thinking about the cables that carry our signals and do all the important work we need them to do on a daily basis. A great deal of theory and engineering goes into making things like telephone lines and HDMI cables work, but we mostly just plug them in and get on with whatever we’re doing.

If this is your experience, you might find the Hackaday Europe talk from [Michael Wiebusch] to be particularly interesting. He dives into transmission line theory from an accessible standpoint, explaining how two disparate signals can go in opposite directions on the very same wire. Then he demonstrates the theory by building a cable modem… well, sort of!

Signal

Michael begins his talk by discussing the Telegrapher’s Equation, but only as a fakeout. Given the limited time on offer, he decided a quicker, easier explanation of the physics involved would be more appropriate. Key to this was explaining the difference between cables and transmission lines. To create a true transmission line, by his definition, he explains that there is a necessity to have two conductors that are relatively close together. Such a transmission line is effectively a distributed network of inductances and capacitances all the way down, though often we talk about “lossless” transmission lines for modelling purposes. He also covers the point of coaxial cables, wherein one conductor is wrapped around another to shield a signal from external noise, and to prevent signal from leaking out.

There are several basic facts to remember about transmission lines. They are fundamentally just channels down which EM signals can travel. It’s also good to remember that they delay signals. To a human, the signal may appear to travel instantaneously, but it does take time. This also has other impacts; for example, coax cables are filled with plastic, a material in which the speed of light is roughly 66% of the speed of light in a vacuum.

This slows the rate at which the field of an EM signal can travel to this fundamental limit. [Michael] also notes that transmission lines, as a wave medium, essentially allow waves travelling in different directions to pass each other, much like ripples spreading on the surface of a pond. This is why it’s possible to have bidirectional communication on a single transmission line. It’s also important to terminate a transmission line properly, such that the wave you’re transmitting down it ends where you want it to—at the receiver. Fail to terminate your transmission line, and you’ll have that wave bouncing back and forth which is undesirable for clear transmission.

[Michael] demonstrates basic transmission line theory by building a sort of cable modem out of an Arduino and some supporting hardware. He notes it’s not really a modem—there is no modulation or demodulation going on. Instead, he’s simply squirting TTL signals into either end of a cable and receiving them on the other end. The “black box” that couples the signals into and out of the transmission line is a simple directional coupler. Built out of resistors and an op-amp, it allows sending a signal down a transmission line, as well as receiving a signal coming the other way. The design works all the way down to DC logic level signals, which let [Michael] use it to send TTL signals up and down 50-ohm and 75-ohm coaxial cables. He notes this has very obvious practical applications where it’s desirable to reduce cable counts when sending signals in multiple directions, relating this directly to his professional work on science experiments.

If you’ve ever wanted to get two devices talking over a single cable in a relatively easy fashion, then [Michael’s] talk may be valuable to you. At the very least, it’s a great way to learn some of the basics of transmission lines and better understand what’s going on when you shoot a signal down a random bit of wire. It’s all good stuff.