How are "cold sparks" possible? At raves, nightclubs, and parties, sometimes there are "cold fireworks" machines that spew out a dense bunch of really bright sparks. The odd thing is that these machines are apparently safe to use indoors and, even more surprisingly, have sparks that are only at a temperature of about 62F (see company statement here: https://www.ansomproductions.com/cold-spark-machines.html) so they (apparently) do not burn your hand if you touch the sparks.

I find this all very surprising, sparks usually get their color from the blackbody radiation spectrum in addition to chemical emission lines. That usually requires that the sparks are very hot. How could it be that these sparks are "cold" and considered totally safe for indoor use?
 A: Let's have a look at the thermal-camera image from the company's website, and have a conversation about how cameras work.

The first thing to note is that $\rm 62\,ºF = 17\,ºC$ is a pretty reasonable temperature for an office building or warehouse with very aggressive air-conditioning or with feeble heat in the wintertime.  That's almost certainly the temperature of the wall behind the model. It is unlikely to also be the temperature of the spark shower.
Here's a question: how big are these sparks? In the visible-light photo on the left, most of them are hand-sized or longer: call it $\rm 10\,cm$ for a nice round number.  But that is clearly not the physical size of the physical sparks, which are zirconium-titanium granules.  The company says a $\rm 200\,g$ package runs the machine for ten minutes, so the granules are leaving the device a one gram every three seconds. That $\rm200\,g$ package has a powder volume of about $\rm 40\,mL$.  If you don't have a good intuition for this, go into your kitchen, use a big spoon to measure out $\rm 40\,mL$ of sugar, and then try to pour it out over ten minutes.  (For extra credit, do your slow pour onto a flame; the sugar crystals will ignite and spark.)
The reason the sparks are long in the visible-light photograph is that they are moving.  A photograph doesn't capture an "instant," though that's frequently a useful approximation; a photograph captures an average of all the light that comes into the camera the entire time the shutter is open.  (Or, for an all-digital camera, the entire time the CCD sensor is active.)  If we make a reasonable guess about the height of the spark shower in the demo, physics tells us how fast the sparks are going; I figure the visible-light image was an exposure between 1/30 and 1/60 of a second.
The sparks show up in the visible-light image, even though they are moving, because they are bright.  But if they were dimmer, or if they were moving faster, their light would be spread across more of the image and become harder to distinguish from the background. For an extreme example, consider the image below:

This is a ten-minute exposure of a busy city street, taken by Daguerre in 1838 (source).  However, only one person on the street stood still enough for long enough to for his image to stand out from the background. (He seems to be having his boots shined.)  Everyone else on the street has disappeared into the noise in the image of the stationary buildings and trees.
In general, an infrared sensor is (a) slower and (b) blurrier than a visible-light camera sensor.  The total mass of burning Zr-Ti is, at any instant, probably about 100 milligrams.  This is two million times less mass than the human model in the image, with body temperature around $\rm 37\,ºC$.
A thermal-imaging camera that is designed and calibrated for slow-moving solid objects like people or steaks or car engines, with one of those objects actually in its image, might very well show a stream of high-temperature particles as a faint bright streak against the background.
An amusing homework problem would be to make reasonable guesses about the total surface area of the burning granules and the infrared passband for the thermal imager, and try to predict the relative blackbody brightnesses of the spark shower and the human model's arm.  Blackbody physics says that the hotter object is brighter in all wavelength bands per unit surface area, but nearly all of the volume of the spark shower is empty space.


For a similar example, compare this thermal image of a candle (source) to a visible-light image of a candle (source).  The brightest visible part of the flame is above the wick, where soot particles are emitting blackbody radiation.  But the brightest infrared part of the flame is the hot part of the solid wick (which you sometimes see glowing blackbody-red), followed by the area of the wax pool that is hot enough to vaporize.  The visibly-bright part of the flame is a distant third in the thermal image, partially because the flame is "optically thin": you can't see through a flame because it's bright, but it doesn't really cast a shadow.
The fact that this company would design a burning-metal spark machine, point a thermal imaging camera through the spark shower at the wall, and not notice that they can change the temperature of their spark shower by adjusting their thermostat — that's an argument for regulation.
A: Anything is "possible" if you lie about it.
While they may well be safer and easier to use than regular pyrotechnics, the manufacturer's website claims that they use titianium and zirconium granules and that they're therefore safe because those materials aren't flammable. That is morally and probably legally a lie.
Titanium and zirconium do not chemically luminesce. They do burn in the presence of air. (At about 3000 and 4000 degrees C respectively.) Titanium and zirconium granules are not safe. They both will explode if they get wet while burning. For instance, if your super safe spark machine tips over and a fire starts and some well-meaning guest tries to put it out with water.
Are the sparks 62 degrees Fahrenheit? No, they're burning titanium and/or zirconium, so they're 3000 to 4000 degrees Celsius. 62 degrees Fahrenheit is probably the temperature of the wall behind the sparks in the manufacturer's thermal image - with an incidental increase from the sparks, which despite their high individual temperature, have a very low heat per solid angle. See Rob's answer for an investigation of what's going on there.
Will they burn your hand and clothing with incidental contact? I would assume that they would not, for the same reason that ordinary welding sparks are of little danger to skin and cloth, and you can sweep your hand through a candle flame without getting burned. The sparks have low heat capacity and the contact time is very short. A lot of granules with a long exposure time, however, are definitely a fire hazard.
Titanium at least will continue to burn in an atmosphere of pure CO2. One or more Class D  (dry powder) fire extinguishers should be on hand if you're going to use combustible metal pyrotechnics. Also, don't take my word for it, I'm a random commenter on the internet, obtain and read the MSDS.
Below: a screenshot from an MSDS for Zirconium powder (not the specific product sold with the linked machine).

A: Well, I have several of these machines. I can easily put my hand (or even my head) into the stream of sparks for 5 or more minutes without feeling any significant heat. I also put a sheet of notebook paper into the spark fountain for several minutes and there was no burning of the paper whatsoever. The machine does give out a lot of "dust", but nothing that makes it out of the machine is hot. The machine does heat up to an internal temperature of 550 to 600 degrees C or so. However, by the time the sparks exit the machine, they are cool enough not to burn or ignite most materials or skin. That being said, there was a case I am familiar with where the machine was placed directly under a dried flower arrangement/decor and it did eventually ignite the foliage. At least, that was what I was told. I plan to try that experiment on my own to verify it though because I have held various kinds of paper in the spark stream for long periods of time without any burning. I am guessing that the dried plant material had a lower point of ignition (which is why it is used to make fire in the wilderness for survival, etc.).
Anyway, the bottom line is yes - the sparks are hot when they are ignited. And, no, they are not very hot (relatively) when they exit the machine. My personal experience with the machines gives me full confidence to say that when used properly, they pose no practical threat of causing a fire or burning anyone.
