A camera built to prove that its photos are real is wearing a 3D-printed body. In a blog post published October 29, the Prusa Research Content Team explains how the CAIM1 provenance camera, created by Josh Cowell, uses enclosures printed in Prusament PETG rather than injection-molded plastic. The reasoning is simple. The camera's internals are dense, hot, and still changing, and printing the shells lets the team redesign the housing and its cooling ducts overnight instead of committing to expensive steel tooling.

For makers, the project is worth a closer look less for the cryptography than for what it says about where printed parts now sit in hardware development. They are no longer only for prototypes. Here they are the body of a device whose whole job is to be trusted, though Prusa notes the camera is still in active development and testing, with a batch ready for production in 2027.

What the CAIM1 Actually Is

CAIM1 stands for Counter Artificial Intelligence Machine 1. According to Prusa, it is a portable camera that attests to the origin of an image or video at the moment of capture. The aim is a verifiable record that a frame came from a physical sensor pointed at a real scene, not from a generative model. Club386 covered the project in September, when Noctua announced its collaboration with Cowell. That report describes the camera as offering cryptographic proof that photos and video were not AI-generated.

The hardware list explains why the enclosure matters. Prusa's post lists 4K60 edge compute, a 4.3-inch touchscreen, a Micro Four Thirds sensor, and a Web3 hardware wallet built around Tropic Square's TROPIC01 secure element. That is a lot of active silicon in a handheld body, and Club386 notes that the TROPIC01-based internals get too hot without active cooling. A passive camera shell designed once and molded for years would not suit hardware like this.

The Case Against Steel Tooling

Injection molding is the default for consumer electronics enclosures for good reasons: low unit cost at volume, consistent surface finish, and fast cycle times. The catch is the up-front commitment. Prusa's post says molding would have locked the CAIM1's design into expensive steel tooling before the thermal engineering could be verified. Every change to a boss, vent, or snap fit after the tool is cut means rework or a new tool.

For a product still evolving around its thermal design, that commitment comes too early. Prusa says 3D printing lets the team iterate on the shell and its internal ducting overnight, adjusting geometry by fractions of a millimeter to optimize airflow. Cowell describes the practical side: if the team needs a replacement panel or gets feedback from a test device user, they can slice it and print it locally in a couple of hours. A cracked or redesigned part becomes a print job, not a supply-chain problem.

Overnight iteration may sound like a familiar maker slogan, but here it matters more because the enclosure and the cooling system are effectively the same part. Duct geometry decides whether air reaches the hot components, and in this camera the ducting is part of the shell itself. Each change to the airflow path means a new shell. With printing, that change costs some filament and a night of printer time.

Advertisement

Designing Around One Airstream

The CAIM1 uses a single airstream cooling duct driven by a Noctua NF-A4x10 5V PWM fan. Club386's report explains the engineering choice. The camera runs entirely within a low-voltage battery domain, so a standard 12V fan would have needed an extra step-up converter, which Noctua says would consume circuit-board space, reduce battery efficiency and add switching noise that could affect the image sensor. The team instead picked the 40 mm, 5V NF-A4x10 and mounted it just behind the sensor heatsink. Noctua's NV-AV4 mounts decouple the fan from the shell and prevent it from vibrating.

The vibration detail is easy to miss, but it matters in a camera. Cowell says the fans are quiet and vibration-free enough that they do not interfere with audio or image capture. Soft mounts that isolate the fan from the printed shell are a sensible way to get there, and they free the enclosure from having to damp vibration on its own.

A single airstream also makes the printed duct's job clearer. Per Prusa, air is drawn through front vents, routed across an alloy sensor backplate, pulled through the decoupled fan, blown across the CPU heatsinks, and expelled. Any maker who has designed a hotend fan shroud will recognize the problem: the duct is only as good as how accurately its walls print.

Advertisement

Why PETG, and Why Prusament

Prusa's post gives several reasons for using Prusament PETG specifically. The first is dimensional consistency. Prusament is cited for a filament diameter tolerance of ±0.02 mm, with each spool's diameter measured by optical laser sensors. Prusa says this gives consistent flow, avoiding gaps or layer swelling that could compromise dust seals or mounting tabs. That matters when panels have to mate cleanly around a lens mount.

The second is thermal stability near hot silicon. The CAIM1 packs edge compute and a secure element into a small actively cooled volume. Prusa says standard PLA would soften near the heatsink exhaust channels, while Prusament PETG handles temperatures up to 68–75 °C without deforming, keeping the ductwork's shape during long shooting sessions. The third is toughness: Prusa credits PETG with the impact strength and elasticity needed for outdoor field prototypes.

Then there is the color. The shells are printed in Prusament PETG in Noctua Beige and Noctua Brown, Noctua's signature palette. Prusa says styling was not the main driver of the material decision, but it shows that a printed enclosure can carry a deliberate visual identity.

What It Means for Makers

The CAIM1 is a useful reference for anyone deciding whether printed enclosures are good enough for a product in development. A few lessons carry over directly.

  • Treat the enclosure as part of the thermal system. When ducting is built into the shell, printing lets you tune airflow the way you would tune any other part, one revision per night.
  • Size components to your power domain. Choosing a 5V fan to avoid a step-up converter in a battery-powered device is the kind of decision that saves board space, heat, and parts count.
  • Isolate vibration mechanically. Decoupling mounts keep the fan from shaking the shell, which is especially important near an image sensor.
  • Filament consistency is a design parameter. Tight diameter tolerance supports the steady flow that mounting tabs, dust seals and mating panels depend on.
  • Delay the tooling decision. Steel molds make sense once a design stops changing. Until then, printed parts keep every option open.

The irony is hard to overlook. A device built to establish trust in digital images depends on one of the most adaptable manufacturing methods available, where the physical object can be revised as easily as a file. For the CAIM1 team, that adaptability is the reason to print. The camera's internals demanded an enclosure that could keep pace, and a spool of PETG kept up better than a steel mold would have.

Sources