Open source & low-cost spectroradiometry
Eyes detect light, so measuring the intensity and spectrum of light is crucial for much of my work in visual ecology. Standard spectroradiometers — the instruments that measure light — are expensive, difficult to use, and few are sensitive enough to work at real night-time light levels. This is a serious problem for working out how artificial light at night (ALAN) affects animal behaviour at night, or calibrating field or experimental conditions to test for the impact of ALAN.
So I built two open-source instruments to fill this gap: OSpRad, a low-cost, high-sensitivity spectroradiometer, and HOSI, a hyperspectral imaging system built around the same core sensor. Together they let researchers measure not just how bright a light is, but its full spectral range, down to levels far dimmer than most commercial equipment can register.
OSpRad: a high-sensitivity spectroradiometer
OSpRad is built around the Hamamatsu C12880MA, a miniature spectrometer chip originally designed for medical devices, repurposed here for ecology. On its own it’s just a bare component; OSpRad wraps it in an Arduino-controlled housing with a motorised shutter and a cosine corrector using 3D-printed parts, for a total build cost of around £250–300.
The result covers roughly 310–880 nm with about 9 nm resolution, and — critically — it’s sensitive enough to measure spectral radiance down to around 0.001 cd/m², and irradiance down to about 0.005 lx. That’s below moonlight and close to starlight, comfortably covering the range of real night-time light levels that matter for ALAN research, from distant streetlights to sky-glow.
Some key features make this far more useful than typical lab equipment:
- Plugs into a phone. The whole thing is controlled by a Python app that runs equally well on a laptop or an Android smartphone, making it properly field-portable.
- Automated measurement. The firmware ramps up integration time until it’s just short of saturation, then takes a matched dark-frame measurement to subtract sensor noise — the same trick long-exposure astrophotographers use, applied here in software, automatically, every time.
- Nothing is thrown away. Every measurement is saved with its raw sensor counts alongside the calibrated values, so data can be recalibrated later without repeating fieldwork. OSpRad was published in the Journal of Experimental Biology in 2023, and it’s already been used to study how artificial lighting reshapes the “landscape of fear” for an endangered shorebird — exactly the kind of question that needed a sensitive, affordable spectroradiometer to even ask.
Download OSpRad on GitHub
Citation: Troscianko, J. (2023). ‘OSpRad: an open-source, low-cost, high-sensitivity spectroradiometer’, Journal of Experimental Biology, 226(13), jeb245416 https://doi.org/10.1242/jeb.245416
HOSI: hyperspectral imaging
OSpRad measures one point at a time, which is perfect for many purposes but limiting if you want to understand a whole visual scene — say, how a streetlight, the sky, and a patch of vegetation compare spectrally, all at once. That’s what HOSI (the Hyperspectral Open Source Imager) is for.
HOSI takes the same core spectrometer chip and mounts it on a motorised pan-tilt gimbal, scanning it point by point across a scene to build a full hyperspectral image — where every single pixel carries a complete spectrum, not just a colour. It’s built from the same kind of off-the-shelf parts and 3D-printed housing as OSpRad, for a total cost of around £350, and runs from the same kind of lightweight desktop or smartphone interface.
Because each pixel is captured with its own independent exposure, HOSI achieves an extraordinary dynamic range — sample night-time scans have shown peak-to-peak ratios exceeding 50,000:1, letting it capture a scene containing both a bright streetlamp and near-total darkness without either blowing out or disappearing into noise. It shares OSpRad’s sensitivity (down to around 0.001 cd/m²) across the same 320–880 nm range, with a spatial resolution of about 0.2 degrees per pixel.
In practice, this means being able to point HOSI at a harbour, a street, or a patch of habitat at night and come away with a full hyperspectral panorama: not just an image, but a dataset in which any pixel’s exact light spectrum can be examined and compared, individual light sources identified, and the whole night-time light environment characterised in a way no standard camera can manage. HOSI was published in BMC Biology in 2025.
Download HOSI on GitHub
Citation: Troscianko, J. (2025). A hyperspectral open-source imager (HOSI), BMC Biology, 23(1), 5 https://doi.org/10.1186/s12915-024-02110-w
Free & open source
Both instruments take a sensor that’s normally locked away behind expensive, closed instrumentation, and rebuild it openly, cheaply, and with a user interface simple enough to run from a phone in a field site at midnight. All of the code, 3D-printable parts, wiring diagrams, and calibration data for both projects are free and open source, so any lab (or curious individual) can build their own.