At Wilco Imaging, we love when our equipment ends up in the hands of someone deeply curious — someone who won't just test the specs, but push the technology somewhere genuinely interesting. Huxley Marvit, a researcher and builder based in the Bay Area, did exactly that when he borrowed one of our hyperspectral cameras and a set of precision lenses for an afternoon experiment with a deceptively simple premise: can spectral imaging detect a fake leaf hiding among real ones?
The answer is yes — and the way it works reveals something fundamental about what hyperspectral imaging can do that no conventional camera ever could.
What Is Hyperspectral Imaging, Really?
To appreciate what Huxley accomplished, it helps to understand what makes hyperspectral cameras fundamentally different from the cameras we use every day.
A standard digital camera — whether in your phone or an industrial machine vision setup — captures light across three broad channels: Red, Green, and Blue. Every pixel in a conventional image is essentially a weighted average of those three buckets. That's enough to produce a faithful color photograph, but it discards an enormous amount of information about the actual light present in the scene.
"Normal cameras capture three buckets: RGB. Hyperspectral cameras capture a bajillion buckets — across a way wider spectrum. Even within the normal visible range, we chop up each RGB bucket into many much smaller ones. Each one, only a few nanometers wide."
— Huxley Marvit, huxley.shA hyperspectral camera slices the light spectrum into hundreds of narrow, contiguous bands — typically spanning the visible range, near-infrared (NIR), and sometimes short-wave infrared (SWIR). Instead of a flat 2D image, the result is a 3D data cube: two spatial dimensions (X and Y) plus a full spectral dimension (wavelength). Every pixel carries its own spectral "fingerprint" across the entire measured range.
This fingerprint is the key to everything. Different materials absorb and reflect light differently at different wavelengths. Plastic and plant matter might appear nearly identical in the visible range — but at certain infrared wavelengths, their spectral signatures diverge dramatically. That's exactly the gap Huxley set out to exploit.
Hyperspectral images are called "data cubes" because of their 3D structure: width (pixels) × height (scan lines) × depth (spectral bands). The camera used in Huxley's experiment captured approximately 200 spectral bands, meaning each pixel carried 200 distinct intensity values — one per narrow wavelength slice.
Think of it as 200 separate grayscale images of the same scene, each shot at a slightly different wavelength, stacked into a single dataset. Processing that cube — scrubbing through the spectral axis, isolating specific wavelength ranges — is where the real analytical power lives.
A Rainy Afternoon and a Plastic Impostor
Huxley's experiment was refreshingly lo-fi in its setup — and all the more compelling for it. Working alongside collaborator Lucas Blakeslee, he collected a handful of real leaves and taped them to a spare 2x4 plank. Hidden among them: one plastic leaf. To the naked eye, the fake was indistinguishable from the real ones. The challenge: could a hyperspectral camera find it?
-
01
Build the scene
Real leaves and one plastic impostor were gaffer-taped to a 2x4. No labels, no markings — visually identical to a human observer.
-
02
Set up the line-scanning camera
Hyperspectral cameras of this type capture one spatial line at a time, building a full image by sweeping across the scene. In field use, the camera is typically drone-mounted. Huxley's ground-level solution: place the board on chairs and slide it past the stationary camera by hand during recording, then stitch the frames in post. Practical, and it worked perfectly.
-
03
Capture the hyperspectral data cube
The camera acquired approximately 200 spectral bands across the scene. Every pixel across every leaf — real and fake — was assigned a complete spectral profile.
-
04
Scrub the spectral axis
Huxley built an interactive slider to step through the spectral dimension band by band. In the visible wavelength range, all leaves looked essentially the same. But as the slider moved into infrared wavelengths, something remarkable emerged.
-
05
Find the separable range
At a specific range beyond what the human eye can perceive, the real leaves vanished entirely — their reflectance at that wavelength dropped to near zero. The plastic leaf remained fully visible, its spectral signature completely distinct from any organic matter in the scene. Camouflage: broken.
At a wavelength range beyond the visible spectrum, every real leaf disappeared from the image entirely — while the plastic leaf remained fully visible. A material undetectable to the human eye, isolated instantly through spectral signature analysis. Hello-world for hyperspectral imaging, complete.
This is a demonstration of material discrimination by spectral signature — one of the most powerful capabilities hyperspectral imaging offers. The same principle drives applications in agricultural crop health monitoring, pharmaceutical quality control, food safety inspection, and plastic sorting in recycling facilities.
From Leaves to the Real World: Where Hyperspectral Changes Everything
Huxley's leaf experiment is a "hello world" — a proof of concept that demonstrates a principle simply and clearly. But the real-world applications of that principle are remarkably broad. Wilco Imaging's hyperspectral cameras are built to bring this capability to demanding industrial and scientific environments.
🌾 Agriculture & UAV Remote Sensing
Detect crop disease, measure frost damage, identify weeds, estimate yields, and analyze soil conditions — all from a drone flyover, in real time.
♻️ Waste Sorting & Recycling
Identify and separate plastics, metals, glass, and textiles by spectral signature — including black plastics that confound color-based sorting systems.
🔬 Food Safety & Quality
Detect contaminants, verify freshness, and identify defects invisible to conventional cameras — non-destructively, at line speed.
🏭 Industrial Inspection
Surface and coating inspection, pharmaceutical QC, environmental monitoring, and geological analysis — anywhere material composition matters more than appearance.
The Wilco Imaging hyperspectral camera lineup — anchored by our patented Curved Volume Phase Holographic (VPH) Grating design — is built around these core demands: low distortion imaging, high diffraction efficiency, and a compact, lightweight form factor that integrates cleanly onto drones and production lines. The WilcoSpec720 offers high frame rates and efficient light collection purpose-built for industrial sorting and inspection workflows.
Whether your application is in the lab or the field, the principle Huxley demonstrated with a plastic leaf scales directly: spectral signatures don't lie, and hyperspectral imaging reads them with a clarity no conventional camera can match.
Huxley had generous words for the Wilco Imaging team who made this experiment possible — lending not just hardware, but their time, expertise, and genuine enthusiasm for the technology: "All the tech worked beautifully, the API was great, and everyone was so incredibly helpful. These folks are stellar."
We're grateful to Huxley and collaborator Lucas Blakeslee for putting our equipment to such creative use. You can read Huxley's original writeup — with an interactive spectral slider to explore the data cube yourself — at huxley.sh/projects/hyperspectral.
Ready to See Beyond the Visible?
Explore our hyperspectral camera systems — built for agriculture, industrial inspection, research, and beyond. Our team in West Sacramento is ready to help you design the right system for your application.
