By using this site, you agree to our Privacy Policy and our Terms of Use. Close
noslodecoy on 30 November 2010
Edit Thread
WereKitten said:
noslodecoy said:

Unless I'm severely mistaken, but outside of glass and water I didn't think there are many materials that absorb infrared light at any substantial rate.  I know this guy has got to be pretty smart man, but it almost seems like he's mistaking IR with UV.  Denim and leather are very UV absorbent.

I even did a quick test with a TV remote and three pairs of denim: dark blackish, dark blue and normal lightish blue.  I pointed the remote at the denim pants and the IR beam seemed to reflect perfectly off the denim pants and turned the TV on and off.  Perhaps someone can do more tests with Kryelos's software.  Maybe that's something I'll try out.

That isn't to say there aren't several problems, and I would expect him to overstate them.  There are "holes" in the IR depth perception, but I think those are more algorithmic.

Infrared allows a quite big range of wavelengths, and just like different colors of the visible spectrum are reflected in extremely variable amounts by different materials, so do the different infrared wavelengths.

I seem to remember that Kinect's laser works in the near IR, somewhere around the 900nm wavelenght. At that w.l., for example, water actually absobs practically the same as red visible light, certainly nothing like the difference between its absobtion of blue and red extremes of the visible spectrum.

As for denim, since this IR is very near to visible red, it could be that some dark dyes (that is, dark across the visible spctrum) are also very dark in the near IR. You could not have found one of those dyes in your test, and remember that a TV remote is something like a big pulsated flash of light on a quite wide IR spectrum. Also, IR sensors in TVs and other appliances are generally not picky: they read a big window of frequencies and lock in on the modulation of the signal AFIAK.

Kinect is more problematic as it projects a dot pattern in a small window of frequencies (it uses a laser, after all) and each dot is hopefully readable by the IR sensor, which has not a great resolution/framerate to start with.

Anyway, there's plenty of hackers playing with Kinect's hardware atm, so it won't take long before we are flooded with clips of people trying weird clothes and showing the 3d sensor output, so we'll know for certain if some materials are really problematic.

Wouldn't most consumer infrared fall in the near-infrared spectrum?  This is why we can "see" it with night vision, at least the cheap consumer night vision which blasts NIR light like a flashlight and then uses a sensor, much like Kinect's sensor, to pick up an image.

Also, the IR in a remote just modulates at a frequency, so it just flashes on and off and should operate at a similar wavelength and "appear" as any other infrared light in it's spectrum around 900nm, right?

If you are in the dark, with a flashlight, and you shine it on a black t-shirt, you would still see the light reflecting off the shirt.

You are certainly right about the hackers and that is one of the reasons I think this guy has it wrong.  With everything I've seen, I would think this would have been out by now, especially with such a common material.