Billy Tolley swings a Microsoft Kinect around an abandoned room in sudden, jittery movements. “Whoa!” he says. “Dude, it was so creepy.” On the display, we see an anomaly of arrows, spheres, and red lines that disappears almost as soon as it arrives. For Tolley and Zak Bagans, two members of the Ghost Adventures YouTube channel, this is enough to suggest they should leave the building. Because for this team and other similar enthusiasts, that seemingly innocuous blotter of white arrows means something more terrifying: a glimpse at specters and phantoms invisible to the human eye.
Fifteen years after its release, just about the only people still buying the Microsoft Kinect are ghost hunters like Tolley and Bagans. Though the body-tracking camera, which was discontinued in 2017, started as a gaming peripheral, it also enjoyed a spirited afterlife outside of video games. But in 2025, its most notable application is helping paranormal investigators, like the Ghost Adventures team, in their attempts at documenting the afterlife.
The Kinect’s ability to convert the data from its body-tracking sensors into an on-screen skeletal dummy delights these investigators, who allege the figures it shows in empty space are, in fact, skeletons of the spooky, scary variety. Looking at it in use — the Kinect is particularly popular with ghost-hunting YouTubers — it’s certainly producing results, showing human-like figures where there are none. The question is: why?
With the help of ghost hunters and those familiar with how the Kinect actually works, The Verge set out to understand why the perhaps most misbegotten gaming peripheral has gained such a strong foothold in the search for the paranormal.
Part of the reason is purely technical. “The Kinect’s popularity as a depth camera for ghost hunting stems from its ability to detect depth and create stick-figure representations of humanoid shapes, making it easier to identify potential human-like forms, even if faint or translucent,” says Sam Ashford, founder of ghost-hunting equipment store SpiritShack.
This is made possible by the first-generation Kinect’s structured light system. By projecting a grid of infrared dots into an environment — even a dark one — and reading the resulting pattern, the Kinect can detect deformations in the projection and, through a machine-learning algorithm, discern human limbs within those deformations. The Kinect then converts that data into a visual representation of a stick figure, which, in its previous life, was pumped back into games like Dance Central and Kinect Sports.
The Kinect isn’t always seeing what it thinks it is
When it was released in 2010, the first-gen Kinect was cutting-edge technology: a high-powered, robust, and lightweight depth camera that condensed what would usually retail upward of $6,000 into a $150 peripheral. Today, you can find a Kinect on eBay for around $20. Ghost hunters, however, typically mount it to a carry handle and a tablet and upsell it for around $400-600, rebranded as a “structured light sensor” (SLS) camera. “The user will direct the camera to a certain point of the room where they believe activity to be present,” says Andy Bailey, founder of a gear shop for ghost hunters called Infraready. “The subject area will be absent of human beings. However, the camera will often calculate and display the presence of a skeletal image.”
Though this is often touted as proof we’re all bound for an eternity haunting aging hotels and abandoned prisons, Bailey urges caution, telling would-be ghost hunters that the cameras are best paired with other equipment to “provide an additional layer of supporting evidence.” For this, Ghost Hunters Equipment, the retail arm of haunted tour operator Ghost Augustine recommends that “EMF readings, temperature, baseline readings, and all of that are essential when considering authentication of paranormal activity.”
That’s because the Kinect isn’t always seeing what it thinks it is. But what is it actually seeing? Did Microsoft, while trying to break into a motion-control market monopolized by the Nintendo Wii, accidentally create a conduit through which we might glimpse the afterlife? Sadly, no.
Photo by Joe Raedle/Getty Images
The Kinect is actually a straightforward piece of hardware. It is trained to recognize the human body, and assumes that it’s always looking at one — because that’s what it’s designed to do. Whatever you show it, whether human or humanoid or something entirely different, it will try and discern human anatomy. If the Kinect is not 100 percent sure of its position, it might even look like the figure it displays is moving. “We may recognise the face of Jesus in a piece of toast or an elephant in a rock formation,” says Jon Wood, a science performer who has a show devoted to examining ghost hunting equipment. “Our brains are trying to make sense of the randomness.” The Kinect does much the same, except it cannot overrule its hunches.
That suits ghost hunters just fine, of course: the Kinect’s habit of finding human shapes where there are none is a crowd-pleaser. The Kinect, deployed in dark rooms bathed in infrared light from cameras and torches, wobbling in the hands of excitable ghost hunters as it tries to read a precise grid of infrared points, is almost guaranteed to show them what they want to see.
Much of ghost hunting depends on ambiguity. If you’re searching for proof of something, be it the afterlife or not, logic suggests you’d want tools that can provide the clearest results, the better to cement the veracity of that proof. Ghost hunters, however, prefer technology that will produce results of any kind: murky recordings on 2000s voice recorders that might be mistaken for voices, low-resolution videos haunted by shadowy artifacts, and any cheap equipment that can call into question the existence of dust (sorry, spirit orbs) — bonus points if battery life is temperamental.
“I’ve watched ghost hunters use two different devices for measuring electromagnetic fields (EMF),” Wood says. “One would be an accurate and expensive Trifled TF2, that never moves unless it actually encounters an electrical field. The other would be a £15 [$18], no-brand, ‘KII’ device with five lights that go berserk when someone so much as sneezes. Which one was more popular, do you think?”
Glitches aren’t tolerated — they’re encouraged
Given the notoriously unreliable skeletal tracking of the Kinect — most non-gaming applications bypass the Kinect’s default SDKs, preferring to process its raw data by other, less error-prone, means — it would be stranger if it didn’t see figures every time it’s deployed. But that’s the point. Like so much technology ghost hunters use, the Kinect’s flaws aren’t bugs or glitches. They’re not tolerated — they’re encouraged.
“If a person pays good money to enjoy a ghost hunt, what are they after?” Wood asks. “They prime themselves for a ‘spooky encounter’ and open up to the suggestion of anything being ‘evidence of a ghost’ — they want to find a ghost, so they make sure they do.”
If it were just the skeletal tracking that ghost hunters were after, better options are now possible with a simple color image. But improved methodology wouldn’t return the false-positives that maintain belief, and so skeletal tracking from 2010 is preferred. None of this is likely to move the needle for those who believe towards something more skeptical. But we do know why the Kinect (or SLS) returns the results it does, and we know it’s not ghosts.
That said, even if its results are erroneous, maybe the Kinect’s new lease on afterlife isn’t a bad thing. Much as ghosts supposedly patrol the same paths over and over until interrupted by ghost hunters, perhaps it’s fitting that the Kinect will continue forevermore to track human bodies — even if the bodies aren’t really there.
