Skip to content

Veritasium Science Physics Security Health

The Science Behind Dogs' Incredible Sense Of Smell

Veritasium visits a NIST fluid dynamics lab where researcher Matt Staymates photographs air with mirrors, lights and lasers. The hook is dogs: a dog smells by breathing out and in about five times a second, and the turbulent backward exhale is what pulls a fresh scent sample toward its nose, a bellows action that extends reach by roughly a factor of 18. Staymates 3D printed dog style nostrils onto a commercial vapor detector and improved its detection performance 16 to 18 times. From there the video becomes a tour of flow visualization: Schlieren imaging, shadowgraphs that read a bullet's speed and loudness from its shock cone, and laser sheets that reveal invisible gunshot residue and drug contamination. It closes on trace detection, Locard's principle, and mask research that shows exactly how much air escapes a face mask.

Published Dec 24, 2022 21:48 video 25 min read Added Jul 7, 2026 Open on YouTube →

At a glance

The title promises dogs, and dogs are the hook, but this Veritasium film is really a tour of a US government lab that photographs air. Derek Muller visits a fluid dynamics group at NIST, the National Institute of Standards and Technology, where researcher Matt Staymates uses mirrors, lights and lasers to make the invisible visible: the heat off a hand, the breath of a dog, the shock wave of a bullet, the residue of a gunshot, the plume of a person leaking skin cells into a room.

The dog part is the payoff of a real experiment. A dog does not smell by inhaling steadily. It breathes out and in about five times a second, and the rapid turbulent exhale is what lets it reach out and pull a fresh scent sample toward its nose. Staymates measured that this in and out bellows action extends a sniffer's reach by roughly a factor of 18, then 3D printed a set of dog style nostrils, bolted them onto an ordinary commercial vapor detector, and improved its detection performance by a factor of 16 to 18. The machine got dramatically better at its job just by learning to sniff like a Labrador.

From there the video becomes a masterclass in flow visualization. Three techniques do the work: Schlieren imaging, its simpler cousin the shadowgraph, and the laser light sheet. With them the lab reads the speed of a bullet off the angle of its shock cone, watches gunshot residue flood a room, tracks how a clandestine drug cook contaminates himself and everything around him, and tests exactly how much air escapes around the edges of a face mask. The closing idea is the through line: seeing the air is how you build the standards that keep the public safe.

A crime lab that photographs air

The film opens cold, inside the lab. In this US government facility they study air flow to solve crimes. Using mirrors, lights and lasers they can illuminate the tiniest differences in air temperature and density. With that ability they can track how drug powder settles through the rooms of a house, work out which of two people fired a gun, and spot particles quietly escaping from a sealed package. Those three claims are the spine of everything that follows, and by the end each one is demonstrated on camera.

The premise Staymates opens with is that our best tool for picking up trace substances has barely changed in centuries, because nature already built the best chemical detector there is. It is the dog's nose. Everything the lab does is, in one way or another, an attempt to understand that nose well enough to copy it and to build the measurement standards around it.

Bubbles, and how a dog actually sniffs

Meet Bubbles: a 3D printed, anatomically correct model of a female Labrador Retriever. The team had tried to film a real police canine in front of the mirror, but as Staymates puts it, dogs get hungry and cranky very quickly, and after about four seconds of usable footage he decided a printed nose that never gets bored would do the job better.

Here is the key fact the whole segment turns on. A dog sniffing for a faint, far away scent does not simply breathe in. It breathes out and in rapidly, around five times a second. The thing that lets it detect a scent from a distance is not the inhale at all, it is the fast, turbulent exhale.

When the dog exhales, a turbulent air jet shoots out of each nostril, the same two jets you feel if you put a hand under your nose and breathe out. But a dog on the ground is nose down, and those jets are vectored backward, aimed toward its rear. By blowing air backward the dog does something clever: it pulls air toward itself from up ahead. Staymates gives the human version. Blow air away from you and, by conservation of momentum, you also reach out and draw air in behind the jet. The dog fires its exhale backward, that jet entrains fresh air from in front of its nose, then the dog reverses the flow and inhales, capturing a brand new sample and handing it to the olfactory region upstairs for analysis. Then it does it again. Five times a second, a repetitive sine wave of push and pull. That rhythmic sampling is what makes the dog such an extraordinary detector.

Figure 1. How a dog reaches out for a smell. The nose fires turbulent jets backward on the exhale, which entrains and pulls fresh scented air in from ahead. Then it inhales that captured sample and repeats the cycle about five times a second. The video makes the counterintuitive point that the exhale, not the inhale, is what gives a dog its long range.

One detail Staymates stresses: the breed barely mattered. Whatever the dog, evolution seems to have converged on the same timing, roughly one fifth of a second between puffs. And the reason the rhythm matters is reach. If a dog had infinite lung capacity and simply inhaled without stopping, its reach would be short, because a pure inhale only draws from the air immediately at the nose. The out and in bellows action is what stretches that reach out, by roughly a factor of 18.

Teaching a machine to sniff like a dog

That number is where the science turns into engineering. The lab has a shelf of commercially available vapor detectors: some tuned for explosives and drugs, some for chemical and biological hazards. They are good instruments, but they all share one crude habit. They just inhale air, steadily, the way a dog never would.

So the team designed and 3D printed what are essentially a pair of nostrils that plug onto the front of one of these detectors and make it sniff like a dog, firing rhythmic jets instead of pulling a constant stream. Then they ran controlled experiments, moving the vapor source farther and farther away to test the range. On average, the biomimetic nostrils improved the detection performance of the instrument by a factor of 16 to 18. Derek's reaction on camera is the honest one: whoa. The machine got that much better at catching a faint, distant scent for one reason only, because it was made to sniff like a real dog. That result is published work, the study Biomimetic Sniffing Improves the Detection Performance of a 3D Printed Nose of a Dog and a Commercial Trace Vapor Detector.

	Ordinary detector	Dog style sniffer
Airflow	Steady, continuous inhale	Rapid out and in, about 5 times per second
Why it reaches	Only draws from air right at the intake short range	Backward exhale jets entrain and pull a fresh sample from ahead long range
Effective reach	Baseline	Up to about 18x farther (the bellows effect)
Detection performance	Baseline	Improved by a factor of 16 to 18 in the tests
What was added	Nothing, it inhales as sold	A pair of 3D printed nostrils bolted onto the same instrument

Figure 2. The headline experiment. Bolting printed dog nostrils onto an off the shelf vapor detector, and driving it to breathe out and in like a Labrador rather than inhale steadily, multiplied both its reach and its detection performance by roughly 16 to 18 times. Same sensor, better breathing.

Schlieren imaging: knifing the light

How do you photograph a dog's breath, or the heat coming off a hand? With Schlieren imaging, a technique Veritasium has covered before. Staymates walks through his rig piece by piece.

The light source is an ordinary automotive headlamp. Its light passes through a condenser lens that focuses it down to a single point. From that point the light diverges again, travels out, and fills a large concave mirror. Think of the light as a bundle of arrows, all flying dead straight. When one of those arrows passes through something with a different refractive index, the warm air off a hand, or a pocket of gas at a different density, it bends. It refracts, just slightly, and its path shifts a hair off from the others.

All the light then reflects back off the mirror to a two way mirror, which turns the returning beam 90 degrees. And here is the trick, the whole reason the method works. Sitting exactly at the focal point of the mirror is a razor blade. Every undisturbed arrow of light comes to a focus right at that edge. By sliding the knife edge in, Staymates cuts off precisely the beams that got shifted, the ones that passed through a density change, and blocks them. With the blade pulled out, nothing is blocked and you just have a very expensive mirror doing nothing. As he pushes the blade in, the turbulence in the air suddenly appears, and Derek can hold his hand in front of the mirror and watch its heat pour upward like smoke. The bent rays get knifed, so gradients in the air print as bright and dark on the image.

Figure 3. The Schlieren knife edge. Undisturbed rays focus exactly on the razor blade and are trimmed. A ray bent by a density change in the test air focuses slightly off the edge and slips past to the camera, so invisible heat and gas print as bright and dark. Slide the blade out and the effect vanishes and you are left with an expensive mirror doing nothing.

Shadowgraph: Schlieren's little brother

Next to the Schlieren rig sits a shadowgraph. Staymates calls it Schlieren's little brother. It is not as sensitive, but it is far easier to build: all you really need is a flashlight and a white wall, and a shadowgraph appears on its own. The physics is simpler too. Light passes through a hot flame or plume at a different density and refractive index, and the plume casts a shadow of itself onto the wall. That shadow is the image, which is why it is called a shadowgraph.

Simple as it is, a shadowgraph can pull out real numbers. From footage of guns being fired, the lab can read off both the speed of the bullet and how loud the bang was. In the slow motion you can see the bullet whizzing away trailed by a very small bow shock, the sign that it is supersonic. For a Smith & Wesson handgun round the shock is tiny, roughly Mach 1.05, just barely above the speed of sound. Set that beside the infamous AK47 rifle round and two things jump out.

First, loudness. Comparing the two side by side, the line of the shock is darker for one gun than the other. The Smith & Wesson is the quieter firearm, so its shock line is less pronounced. The reason is physical: the hot gases released when a gun fires expand outward and create a shock wave. Across that shock the pressure, temperature and density of the air all change abruptly, and the bigger the jump, the darker it prints in the shadowgraph. So the darkness of the shock is a readout of how loud the shot is.

Reading a bullet's speed from its shock cone

Second, speed. The AK47 round comes off far faster, around Mach 2 to 2.5, and its shock cone is a narrow, sharp wedge. The underlying idea is general. Any object moving through air makes pressure waves that spread out at the speed of sound. When the object outruns its own sound, those waves pile up and compress into a single conical shock wave trailing behind it, a tiny sonic boom. And the geometry of that cone encodes the speed. As Staymates states the rule, the sine of the cone's half angle equals the ratio of the speed of sound to the speed of the object. Measure the angle and you have the Mach number.

The relation means a slow, barely supersonic round makes a wide, weak, nearly flat shock, while a fast round makes a tight, narrow, violent one. A round at Mach 1.05 opens a cone of about 72 degrees, almost perpendicular to its path, whereas one at Mach 2.25 draws a sharp cone of about 26 degrees. Wide and faint means slow and quiet, narrow and dark means fast and loud, and the shadowgraph lets you read both straight off the image.

Figure 4. Reading speed off the shock. A supersonic bullet drags a conical shock wave whose half angle mu satisfies sin mu equals the speed of sound over the speed of the bullet, one over the Mach number. A fast AK47 round cuts a narrow, dark cone, a barely supersonic Smith and Wesson round makes a wide, faint, nearly flat one, so the picture alone gives you both the speed and a sense of the loudness.

Laser light sheet: making invisible plumes glow

The third technique swaps refraction for illumination. A laser light sheet lights up fine particles directly, so you can see plumes of gun powder residue in the air. The setup is almost absurdly simple: a laser beam is steered through a cylindrical glass rod, just a piece of glass shaped like a cylinder, and the rod spreads the beam into a flat, two dimensional wall of laser light. Whenever particles or theatrical fog drift across that sheet, they light up. Staymates wears laser safety glasses and can barely see the sheet himself; it shows up clearly only through the camera.

With the sheet running, the lab studies gunshot residue, the burned and unburned propellant thrown out whenever a firearm is discharged. Laser sheets plus high speed cameras capture the plume of residue that erupts after a firing event, and reveal the ventilation inside a real gun range. The footage is startling: an invisible cloud of residue you would never see with the naked eye, five shots in a few seconds throwing off enormous amounts of material. A lot lands on the shooter's hands, but a lot more disperses into the surrounding room, and it travels much farther than Derek expected. That leads straight to a forensic question the lab is actively working on: can you tell apart residue actually deposited by the shooter from residue picked up by a bystander who wandered in two minutes later?

Staymates is careful about what these tools are. The Schlieren, shadowgraph and laser sheet methods are qualitative, or at best semi quantitative. They do not replace hard chemical measurement; they illustrate what is happening so you know where and how to measure. The table below lays the three side by side.

	Schlieren	Shadowgraph	Laser light sheet
Makes visible	Faint density and temperature gradients in clear air	Sharper density changes, cast as shadows	Actual particles and fog crossing a flat light plane
How it works	Bent rays trimmed by a razor at a mirror's focal point	A plume casts a shadow of its own refraction on a wall	A laser beam fanned by a glass cylinder lights up whatever passes through
Sensitivity	Highest, catches the faintest gradients	Lower, but far easier to build	Shows particulates, not gradients
In this video	Dog breath, hand heat, mask leakage	Gun shocks, bullet speed, loudness	Gunshot residue, talc, drug mixing plumes

Figure 5. Three ways to see air. Schlieren is the most sensitive and prints the faintest gradients, the shadowgraph trades sensitivity for a build so simple it needs only a flashlight and a wall, and the laser sheet lights up real particles rather than gradients. The lab reaches for whichever one fits the question.

Trace detection and Locard's principle

Now the public security half of the lab, which is about trace contraband detection. It starts with a creepy demonstration. Every one of us gives off heat, so every one of us carries a warm rising column of air, the human thermal plume. Under Schlieren you can watch it stream up off a person. And that plume is full of you: skin cells, shed constantly at a rate Staymates puts at thousands an hour. From a trace detection standpoint, that constant shedding is not a bad thing at all, it is the whole game.

The principle underneath is Locard's exchange principle from forensic science: every contact leaves a trace. Spend time in the lab and you leave part of yourself behind, and you carry part of the lab away, whether you like it or not. Turn that on a criminal and it becomes a detection strategy. A bomb maker, or someone cooking illegal fentanyl in a basement, inevitably contaminates himself with the bulk material, and that contamination takes the form of extremely small particles. Modern chemical detection systems are so sensitive they can flag a single particle of explosive residue.

So imagine a bomb sealed in a package at a screening facility, say a checkpoint for items arriving from overseas. You have a sniffer, a commercial vapor detector, and ten seconds to sample the package. Where do you aim it? The lab answers that with an experiment they call burping the package. They put acetone inside a box, and when you squeeze it, acetone vapor puffs out through every gap and seam. Those gaps are where the inside leaks to the outside, so aiming the detector at the seams and corners, where vapor is actually permeating through, gives you the best chance of catching what is inside. Aim at a flat, sealed face and you get nothing.

The drug angle is even more vivid. To simulate illicit manufacturing, Staymates uses talc powder. He shakes a container and squeezes, and to the naked eye nothing happens, yet in the laser sheet the room fills with powder. He pours one substance into another and, invisible to the eye but glaring in the laser, a cloud rolls off the moment the lid comes off and keeps going. The point lands hard: a basement cook is spreading contamination across every surface in the room without seeing a thing. Staymates says he cannot share the actual numbers, only that they are pretty startling. There is, in his words and Derek's, a lot.

The particle count graph: how much you breathe in

The laser sheet does more than show where powder goes; it lets you count exposure in real time. As a person works with the material, a plot in the top corner of the frame tracks the real time count of particles being generated around them. That graph is essentially the person's inhalation exposure, the load of material they are breathing in moment to moment. Derek does the grim math out loud: if that powder were fentanyl and the person had no mask on, they are gone. To trace where contamination ends up, the lab also spikes the material with fluorescent powder so the spread lights up and can be mapped.

Alongside the visual methods, the lab runs a quantitative one: swabbing surfaces around the house to measure what settled where. It is exactly what happens at the airport when security asks for your hands and wipes them with a swab, hunting for trace explosive particles that would cling to anyone who had handled a device.

Then a forward looking idea. Staymates points at a drone. Its four propellers throw off an interacting prop wash, and the question is whether that fluid dynamics could do the sampling for you. Sending human Hazmat crews into a suspected meth or fentanyl lab is slow, expensive and dangerous. Instead, fly a drone in, let its prop wash stir particles up off the surfaces, catch them with a collector on the drone's belly, fly back to base, and run the chemical analysis. If it comes back clean, nobody had to suit up; if it finds something, only then do you send in the Hazmat team. It is an idea, not a finished product, but it shows how the lab thinks.

Mask research: seeing what a face mask really does

All of this sits under one umbrella, public safety and security, and when the Covid pandemic hit, the lab switched gears to it. Staymates built a machine that breathes like a human, tuned to his own measured breathing rate: a pneumatic system driving a set of artificial lungs, fed by a fog generator so each breath is visible. On camera it looks exactly like something taking a drag off a cigarette and exhaling.

The payoff is quantitative. Put a mask on the breather and it looks like almost nothing gets through, but in fact millions and millions of particles are pushing at the fabric, and some do make it, because no mask is 100 percent. Staymates wrote an image processing program that counts the white pixels of escaping fog. Run an N95 respirator through it and guess what fraction of pixels light up white. Derek guesses 5 percent, and he is right, 5 percent.

Then Derek wears masks himself while the Schlieren rolls. Breathe in and the image darkens; breathe out and it lightens, because the exhaled air is warmed by your lungs and shows up as a different density. A good mask that seals shows almost no color change at all, which is the difference between sealing and merely filtering. A thinner mask changes color much more dramatically, because more heat transfers straight through it, and you can watch a lot of air escaping over the top edge where the fit is poor. Seeing it is the whole point.

Why show it at all: communication and standards

Staymates ties the mask work back to a public failure. Early in the pandemic the message lurched around: masks do not work, do not wear one; then masks do work, wear a cloth mask; then no, cloth masks do not work. There was a lot of confusion. What the lab took from that is that the communication of mask effectiveness could have been much better, and that is exactly why he made the Schlieren videos. The average person is never going to sit down and read a scientific journal article, but they will happily watch a 90 second clip of a scientist coughing with a mask on and off. So that is what he made.

He also explains how this lab fits into the wider system. NIST has a distinctive relationship with other federal agencies. A three letter agency in the security world arrives with a concrete need, say, we want to sample people's shoes for explosives. The lab then works out which sampling methods are good and which are poor, what the measurements need to look like to evaluate a shoe sampling device that does not even exist yet, and what standards would support those measurements. They package all of that up and hand it back to the sponsoring agency, which takes it to industry, and industry gets a head start because NIST already did the heavy lifting.

The closing note is that the application space is enormous. Because the Covid visualization worked so well, the same tools now point at indoor air quality, and Staymates imagines a bigger mirror that could show two people interacting and exactly how breath transfers from one to the other. The take home message he leaves with is simple: flow visualization is a critical tool at NIST, because it is one thing to run quantitative analysis on surfaces, and another thing entirely to actually see where the particles are and where they go.

The video closes with a sponsor read for Caseta by Lutron, smart home lighting control, on the theme that good lighting matters for more than photographing air.

Where it stands

Nothing here is speculative fringe science; it is a working measurement lab, and most of the striking claims are grounded. The dog nose result is peer reviewed: the roughly 16 to 18 times improvement in detection comes from the published biomimetic sniffing study, and the physics of shock cones, refraction and thermal plumes is textbook. The honest caveats are the ones Staymates flags himself. The imaging methods are qualitative to semi quantitative, so they guide measurement rather than replace it. The single particle sensitivity and the drone sampling concept are real research directions, not deployed products, and he is explicit that some numbers, on drug contamination, cannot be shared. Read that way, the video is less a story about dogs and more a clear window into how a standards lab turns something invisible into something you can measure and regulate.

Key takeaways

A dog does not smell by inhaling steadily. It breathes out and in about five times a second, and the turbulent backward exhale is what pulls a fresh scent sample toward its nose.
That out and in bellows action extends a sniffer's reach by roughly a factor of 18 over a constant inhale, and the timing is nearly the same across breeds.
Bolting 3D printed dog style nostrils onto a stock commercial vapor detector improved its detection performance by a factor of 16 to 18, a peer reviewed result.
Schlieren imaging makes invisible density and temperature gradients visible by focusing light to a point and trimming the bent rays with a razor blade at the focal point.
A shadowgraph is simpler and less sensitive, needing only a light and a wall, yet it can measure a bullet's speed and loudness from the shape and darkness of its shock.
A supersonic object trails a conical shock whose half angle obeys sin of the angle equals the speed of sound over the object's speed, so the cone's geometry gives the Mach number.
A laser light sheet, made by fanning a laser through a glass cylinder, lights up particles directly and reveals otherwise invisible gunshot residue and drug powder plumes.
Locard's exchange principle, every contact leaves a trace, underpins trace detection: people and criminals both shed particles that sensitive detectors can catch down to a single particle.
The lab makes air visible not for spectacle but to build the measurement standards and public communication, from where to aim a package sniffer to how much air escapes a face mask.

Chapters

00:00 Intro 00:26 How dogs sniff 03:14 Schlieren Imaging 04:57 Shadow Graph 07:24 Laser Sheet 09:28 Trace Detection 13:53 Trace Detection Graph 15:56 Mask Research 16:56 Breathing 17:39 Masks 20:12 Sponsor

Notable quotes

"So it turns out nature has already provided us with the best chemical detector, and that is the dog's nose." (00:31)
"What they do is breathe out and in rapidly, around five times a second." (00:50)
"But when the dog is down on the ground, those air jets are vectored back towards its rear. The dog is pushing air back, and by doing that, it's pulling air from ahead of it." (01:11)
"But because of this in and out bellows effect, its reach goes up by roughly a factor of 18." (01:43)
"We were able to, on average, improve the detection capabilities of these by roughly a factor of 16 to 18. Just by making them sniff like a real dog." (02:01)
"I cut off those arrows of light, those beams of light that shifted a little bit, and I cut 'em off, I block 'em." (04:26)
"It's Schlieren's little brother. It's not a Schlieren system, but it's very close. Shadowgraph is not as sensitive, but it's easier to build." (05:07)
"The bigger that change, the darker this shows up in the Shadowgraph. So you can get a sense of how loud it is by how dark that shock appears." (06:00)
"The sine of the angle is the ratio of the speed of sound to the speed of the object." (07:15)
"It's all based on this fundamental principle, it's called Locard's Exchange Principle. It's that every contact leaves a trace." (10:52)
"The chemical detection systems that we have available now are so sensitive, they can detect a single particle of an explosive residue." (11:36)
"That's crazy, 'cause if that's fentanyl, and that person doesn't have a mask on, they're gone." (15:20)
"The average American is not gonna sit down and read a scientific journal article. But they will sit down and watch a 90 second video of me coughing with a mask on and off." (18:52)

Resources mentioned

Veritasium, the channel, hosted by Derek Muller.
NIST, the National Institute of Standards and Technology, where the lab is based.
Matt Staymates, the NIST research fluid dynamicist featured throughout.
Biomimetic Sniffing Improves the Detection Performance of a 3D Printed Nose of a Dog and a Commercial Trace Vapor Detector, the published study behind the dog nose result.
Schlieren photography and the shadowgraph technique.
Mach number, shock waves and the sonic boom that set a bullet's shock cone.
Gunshot residue and Locard's exchange principle from forensic science.
Dog olfaction and the Labrador Retriever that Bubbles was modeled on.
N95 respirators and face masks during the Covid pandemic.
Caseta by Lutron, the smart lighting sponsor of the video.

Full transcript

In this US Government lab, they study air flow to solve crimes. Using mirrors, lights and lasers, they can illuminate the tiniest differences in air temperature and density, and track how drug powder settles in the rooms of a house, determine which person fired a gun, or spot particles escaping from a sealed package. A portion of this video was sponsored by Caseta by Lutron. Our best tool for picking up trace substances has been the same for centuries. So it turns out nature has already provided us with the best chemical detector, and that is the dog's nose. So this is Bubbles. She's a 3D printed, anatomically correct model of a female Labrador Retriever. We tried to get one of the police canines to come in here and get in front of the mirror, but dogs get hungry and cranky very quickly. And we got about four seconds of good footage, and I'm like, okay, we can do better than this. Dogs can sniff very faint scents and from far away, but they don't do it just by breathing in. What they do is breathe out and in rapidly, around five times a second. What I want you to notice is this pulsating motion by the, we'll call it the doggy treat, right? This is our doggy treat. If you notice this pulsating motion. And what allows them to detect scents from far away is the rapid and turbulent exhale. When the dog exhales, there's a turbulent air jet that comes out of each nostril, just like when you and I exhale. If you put your hand under here and just exhale out of your nose, those are two turbulent air jets coming out of the dog's nose. But when the dog is down on the ground, those air jets are vectored back towards its rear. The dog is pushing air back, and by doing that, it's pulling air from ahead of it. So imagine me, I'm blowing air this way, and when I do that, I actually reach out and pull air towards me. The dog reverses flow, inhales, right? Gets a new sample, analyzes it with an amazing chemical detector upstairs, right? Its olfactory region, and does it at five times a second. On average, a dog when it's sniffing for something, five times a second, repetitive sine wave. So that's what makes the dog this incredible sampling system. It kind of didn't matter the breed of dog. Somehow their evolution allowed them to figure out, okay, one over five seconds is what I need before I take the next puff or the next inhale. If the dog had infinite lung capacity, and it just inhaled constantly, it has a very limited reach. But because of this in and out bellows effect, its reach goes up by roughly a factor of 18. We have a number of commercially available vapor detectors, okay? Because some of these are really good for various explosives and drugs, some are very good for chemical and biological hazards, but they all just inhale air. What we did is we designed and printed what are essentially nostrils that plug onto this thing and make it sniff like a dog. And then by doing a set of controlled experiments, by moving your vapor source farther and farther, we were able to, on average, improve the detection capabilities of these by roughly a factor of 16 to 18. Whoa. Just by making them sniff like a real dog. We're able to see this dog breath using Schlieren imaging. I've actually done a video on this before. This is an automotive headlamp. After that, I have a condenser lens, so it takes this light and focuses it down to a point. So from there our light diverges, and it comes up and it fills our mirror. Think of these light beams as arrows, okay? And there are a bunch of arrows that are coming up, they're straight as can be, but when they pass something of a different refractive index, which could be the heat from my hand, or a different density, which could be a gas of a different density, those light rays, they shift a little bit, they shift very slightly, okay? They refract, they bend. Then they all come back. So we're coming back here, and right here we have a two way mirror. So light can travel through the mirror this way, but when it comes back, it turns 90 degrees. And if you look right here, I have a razor blade. And what this is doing, it is positioned exactly at the focal point of that mirror. I have that razor blade edge, it's able to move in and out of the focal plane. And so what I do is I cut off those arrows of light, those beams of light that shifted a little bit, and I cut 'em off, I block 'em. As I move the knife edge out, so I'm not cutting off any light right now, so you basically have a really expensive mirror, it's not doing anything for you. But as I take this razor blade, and push it in, you start to see turbulence in the air, and as I take it more and more, yeah, there you go, get your hand in there. Whoa, your Schlieren setup is so much better than mine. So take a look at this. So this is in a Shadowgraph system. Wow. So it's Schlieren's little brother, okay? It's not a Schlieren system, but it's very close, a Shadowgraph. Shadowgraph is not as sensitive, okay? But it's easier to build. Look, we have a flashlight and a white wall, and we're already seeing a shadowgraph there. So why does this work? What's happening is we have light, it's not as focused as I would like, but the more focus, the better. The light's coming up this way, it's passing through a hot flame of a different density and refractive index, and we're basically casting a shadow of that plume on the wall, which is why it's called a Shadowgraph. These can still be used to visualize flow patterns in detail. Just by looking at the footage of guns fired in a Shadowgraph, we can actually figure out the speed of the bullet and how loud the bang was. And there's our bullet whizzing off, it is supersonic because it has a very small bow shock. You see it there? Yep. Very, very small. So it's roughly what Mach 1.05 maybe, just above sonic. Compare that to the infamous AK47 assault rifle. There's two things that we can qualitatively observe here. One is, if you compare these side to side, the Smith & Wesson versus the AK47, the darkness of this line is different. This is a quieter firearm, it's not as loud, the report is not as loud. Because it's less pronounced. It's less pronounced, yep. The hot gases released when the gun is fired expand outwards, creating a shockwave. Across this shock, the pressure, temperature, and density of the air changes rapidly. And the bigger that change, the darker this shows up in the Shadowgraph. So you can get a sense of how loud it is by how dark that shock appears. And then here we see our bullet whizzing off at about, oh gosh, Mach two, two and a half. It's like two, three degrees, right? You could actually measure that and calculate the Mach number. Any object moving through air creates pressure waves that travel out at the speed of sound. Since this bullet is moving faster than the speed of sound, the pressure waves it creates all get compressed together into a single conical shock wave. It's like a tiny sonic boom. And you can use this to determine the speed of the bullet. The sine of the angle is the ratio of the speed of sound to the speed of the object. There is a different way to visualize flow, which is with a laser sheet. It illuminates fine particles. So we can see the plumes of gun powder residue created when a gun is fired. We call this laser light sheet. And the reason is, okay, so now our light's there, I can barely see it because I have laser safety glasses on. Yep, I can only see it in the camera. Yeah, so what's happening here is we have a laser beam, and we send it, we steer it through a cylindrical glass rod. All it is, is a piece of glass that's a cylinder. And what that does is it spreads that beam into a two dimensional wall of laser light. Okay? And so the way this works is whenever particles or theatrical fog cross through this sheet, they light up because of the laser. There's a lot of effort now going into understanding gunshot residue. So gunshot residue is the kind of burned and unburned propellant that occurs whenever a firearm is discharged, right? So what we're looking at here is using laser lights and high speed cameras, is we're looking at the plume of gunshot residue that's generated after a firing event. This is looking at ventilation inside an actual gun range. But look at all this. This is all gunshot residue that you can't see with the naked eye, but is there, so you can imagine if you're inside a home, this is five shots in a few seconds. Enormous amounts of material. A lot of this is landing on the hands of the shooter, but unfortunately a lot of it is also being dispersed into the local environment. It goes way farther than I expected. So the question is, can we differentiate between something that's actually been done by the shooter and something, say, by a bystander that walked in two minutes later? So this is what we're actively trying to pursue right now. So these are tools and techniques that we use here at NIST that are qualitative or sometimes semi quantitative, but they really help illustrate what's happening. And then the public security part of this deals with trace contraband detection. We're all generating heat, right? So we all have this warm plume of air. You wanna see it? I'll show you. It's kind of crazy. So you seeing it there? So that is called the human thermal plume. And guess what's in that plume? Water? Parts of me. Skin cells. No way. Yeah. So we're shedding skin cells at a pretty creepy rate. I forget what the number is, but it's a lot. Thousands an hour ish. So we are shedding these skin cells constantly, and it's actually not necessarily a bad thing from a trace detection standpoint. What do you mean? Well, whenever we are talking about trace detection, whether it's drugs or explosives, it's all based on this fundamental principle, it's called Locard's Exchange Principle. It's in forensic science. It's that every contact leaves a trace. And what that means is when you're done in this lab, there will be parts of you left behind in this lab, whether you like it or not, right? 'Cause we're all shedding skin cells and taking things, and taking stuff with us. So if I'm a bomb maker, if I'm involved in manufacturing, you know, illegal fentanyl in my basement, I am ultimately contaminating myself with the bulk material. And this contamination comes in the form of very, very small particles. And the chemical detection systems that we have available now are so sensitive, they can detect a single particle of say, an explosive residue. Imagine you have a bomb in a package, right? And you're at some kind of screening facility, right? You're at a facility that maybe is screening things coming in from overseas, maybe you've got a sniffer, you've got a commercially available vapor detector, and you've got 10 seconds to sample this package. Where do you point your device, right? These are the kinds of experiments that help us understand where you should point your device. We call this burping. We're burping the package. So we have some acetone in here. So wherever you burp it, where you squeeze it, you have acetone vapor coming out of all the gaps, right? So they're coming out in the gaps, so that gap gives you the highest chance of actually detecting what's inside, right? You don't want to go over here where there's no gaps. You don't want to go up there where there's no gap. You want to go in these corners here, where stuff is actually kind of permeating through and outside of the box. This is some recent work we're doing. We are basically simulating illicit drug manufacturing. So here's some talc powder, right? No big deal. I'm gonna shake it up a little bit. When I open this and squeeze, I can't see anything, but are you seeing anything in the camera? Yes. So again, to the naked eye, we'd never be able to see this. But because the laser is doing the illuminating for us, we're able to see this. Let's do a little experiment where I am taking illegal substance A and pouring it into substance B. First thing I'll do, just watch when I take the lid off of this. Take it from here. There is a little bit coming off, yeah. A little bit? Oh yeah. It's still going. So just imagine, you're in a basement, and you're building stuff like this, and you don't realize that there's all this contamination spreading and landing on other surfaces in your room. And I can't talk about the numbers, but they're pretty startling. Like there's a lot. There's a lot. It spreads everywhere. There's a lot, yeah. This thing's going to be full of smoke in here. You ready? What I really love about the laser sheet technique is it allows you to visualize the turbulence that's in the air. So this is a way we can see how the air flow in the room is actually being tracked. Oh, that's nice. Yep. If you look at the graph in the top right, that is actually real time counts of the particles that are being generated around this person. So that plot represents basically inhalation exposure of the materials that he's working with. That's crazy, if you think about, 'cause if that's fentanyl, and that person doesn't have a mask on, they're gone. We add fluorescent powder to this kind of stuff, so we can actually visualize where the contamination goes. The other cool part of this is that we can use a quantitative method of sampling surfaces in the house. We basically use swabbing. And it's the same idea when you go to the airport. Have you ever gone there and they ask for your hands? Right? And they take a little swab and they wipe your hands. What they're doing is they're looking for these trace amounts of explosive particles that could be on you if you were involved in the manufacture of an explosive device. But that's a drone. That's looking at a drone. And the question is, 'cause you know, you get this cool prop wash, and these four propellers that are all interacting with each other. So the question is, can we use the fluid dynamics of a drone to do the sampling for us? And here's what I mean by that. If there is a suspected, you know, manufacturing facility of something, say it's a methamphetamine or fentanyl, you know, it's very, very expensive to get Hazmat crews involved. Have them come, they have to gown up, they have to go in, to a potential really dangerous situation. What if we had a drone? We just flew the drone in, right? It buzzes around and it's got some special kind of collector on the belly of the drone and we're using the prop wash of the drone to stir up particles off of a surface, and then somehow inhale them, collect them. The drone comes back to base, you run your chemical analysis, you say, okay, the house is clean, or no, we found stuff. Okay, now it's time to pull out the Hazmat crew. You know? So the big picture idea that I want you to keep in the back of your head the whole time is this idea of public safety and security. So that's really what happens in this lab. But underneath that kind of umbrella term, think mask research. So when the Covid pandemic hit, we kind of switched gears in here to try to address some of the issues related to masks. This thing will actually breathe as a human does, that human happens to be me. So I measured my own kind of breathing rate, and then built a system, kind of engineered a pneumatic system that replicates that, but it also has a fog machine, a fog generator. So this thing basically just looks like it took a drag off of a cigarette and exhaled. What's kind of cool with this is it looks like there's nothing coming through, right? But there is millions and millions of particles trying to make it through. Some do, 'cause it's not 100%. So I wrote an image processing code. So you take this information, you plug it into a code, and it analyzes, based on pixels and it's counting white pixels, and sure enough, if you use an N95, you put an N95 on here, you run it through an image processing code, guess what percentage of pixels light up white? 5%? 5%. Okay. Now I just want you to breathe naturally. So when you're inhaling, it gets dark. And when you're exhaling, it gets lighter, because the air is being warmed by your lungs and then coming back out. Like it's interesting how fast you can see the effect of the breathing in, like just makes it go dark so quickly. But that shows me that, like this now doesn't change color at all. So this is being used to seal versus filter. I notice the color change is a lot more dramatic. With this, you know, it was in and out. I feel like that's a thinner mask, right? Yeah. So the heat transfer through, it is gonna be greater. I feel like there's a lot coming out over the top. There is, yeah, there's a lot. If you just breathe naturally, you'll see a lot coming out there. The thing with masks, right, is like, initially they were like, masks don't work, don't wear a mask. And then they were like, masks do work, and wear a cloth mask, and then they were like, no, no cloth masks don't work, like, I don't know, there was a lot of confusion around masks. What we learned early on was that the communication of, you know, mask effectiveness could have been improved, right? And so that's kind of why I did what we did with the Schlieren. The truth is, the average American is not gonna sit down and read a scientific journal article. But they will sit down and watch a 90 second video of me coughing with a mask on and off. And that's what we did. It's kind of a unique relationship we have with other federal agencies here. So you have a three letter agency that has a specific need for something, right? In the security arena. What they'll often do is they'll come here to NIST, and they'll say, hey, we're really interested in sampling people's shoes for explosives for whatever reason, right? So what we do here, and in this lab in particular, is we figure out what are some good ways to sample shoes for explosives? What are some not so good ways, right? What do the measurements need to look like to evaluate a shoe sampling system that doesn't even exist yet? And then what do the standards need to look like that support those measurements, right? And so we figure all this stuff out in the lab, package it up nice and neat, give it back to the sponsoring agency, and then they take it to industry. And industry already has a leg up on the development of these kinds of systems, because we did a lot of the heavy lifting here in the lab. Because with the success of, you know, doing the Covid related visualization, we're realizing now that the application space for this kind of technology is huge. Indoor air quality. You can imagine a bigger mirror where we could look at two people interacting with each other, and what that transfer looks like from one person to the other. The take home message is flow visualization is a critical tool that we use here at NIST to really help understand what's happening, right? It's one thing to do quantitative analysis on various surfaces, but now we can see, we can actually see what's happening, and where these particles are generated. Thanks to Lutron for sponsoring this portion of this video. Perfect lighting isn't just important for visualizing air flow, it's also essential for making your house feel welcoming, cozy and safe. And with the sponsor of this video, Caseta by Lutron, you can easily set up your own perfect lighting. Caseta provides premium smart lighting control, including switches, remotes and plug in smart dimmers, which allow you to customize your home lighting from anywhere in the house. What I love about smart switches is that one switch can control many regular bulbs. So you can effectively make all your bulbs smart just by replacing the switch. And installation is easy. First, make sure you turn off power to the switch, and then disconnect the existing wires, and connect them to Caseta's Smart Switch. If you need any help, they're just a click or a call away. And then with the Smart Hub, you can turn your lights on and off just using your phone. Or you can use another device, like Alexa or Google Assistant. Caseta works with more smart home brands than any other smart lighting control system. Plus their switches don't rely on Wi-Fi, so they always stay connected, and you always have control over your lights. I personally like setting timers in my house, for example the lights in my office turn on by themselves in the evening. And this gives me peace of mind that my family will always come home to a well lit house. And the best part of it all is if I'm already in bed, I can just check which lights are still on, and turn them off without having to run around the house. Learn more about Caseta at casetawireless.com/veritasium. Thanks to Lutron for sponsoring that part of my video.