Sciencemadness Discussion Board

Infrared photography, only partial success

woelen - 14-11-2014 at 00:12

Some time ago, beautiful pictures were posted of bromine and also of other chemicals, photographed at infrared wavelengths. Some examples are given bij nezza and endymion17, e.g. http://www.sciencemadness.org/talk/viewthread.php?tid=23985&...

I tried doing this myself. Recently, a nice camera is introduced for the Raspberry PI:

http://www.raspberrypi.org/pi-noir-infrared-camera-now-avail...

I purchased a Pi noIR, a Raspberry PI B+ and a suitable case:

http://www.kiwi-electronics.eu/raspberry-pi/raspberry-pi-mod...

Beware: buy the black case. The camera module has a very bright red LED and this screws up your pictures if you use it for macro purposes. The LED is switched on while the camera is capturing an image and processing the data.

This camera is fixed focus, but it fairly easily can be modified, by removing some glue, which keeps the lens in fixed position:

http://www.instructables.com/id/Raspberry-Pi-Birdbox-Camera/...

After this "mod", the focal length can easily be modified with a blunt needle, by applying force in one of the holes at the rim of the camera and carefully turning around the lens (some force is required, but when the camera is in the case, as shown above, then this is safe).

With this camera you can make infrared pictures. I also purchased a handful of IR-LEDs and an IR-filter (720 nm).

The camera is quite good, it makes very sharp images and with the change of focal length you can make amazingly good macro pictures. I tested with an IR-cut filter, which I purchased for a few euros on eBay.

However, what I notice is that my IR-pictures of bromine show that the liquid is not clear like water, but only partially transparent (it looks grey instead of colorless and objects behind the liquid are only weakly visible through it).

Now I am wondering, what wave lengths should I make pictures at? According to documents I read, the Raspberry PI NoIR camera is sensitive between appr. 380 nm and 1000 nm and with the IR-filter in front of it, it ranges from 720 nm to 1000 nm.

When I use the 720 nm IR-filter, I also really need to illuminate the scene with IR-LEDs, without them I only get very dark pictures. I do not know the specs of my LEDs, they are from some el-cheapo no-name source. They emit a lot of light, as the camera shows, but maybe the wrong wavelengths for good pictures?

What can I do to improve the pictures and get some really fine IR-pictures like I have seen over here on sciencemadness? With my setup, the IR-effect is not strong enough, I only get a mix of normal light and IR-light behavior.

[Edited on 14-11-14 by woelen]

food - 14-11-2014 at 15:31

Just a comment. With the regular pi camera module you can disable the led. As far as I know the noir is the same camera minus a filter.

Like so:

http://www.raspberrypi-spy.co.uk/2013/05/how-to-disable-the-...

I've picked up a noir as well, but not tried it out.

Little_Ghost_again - 15-11-2014 at 03:30

Not sure this is much help, IR leds on ebay differ widely in wave length, the better ones tend to be SMD size. I have many IR leds I use for electronics, if you know the best wavelength I can send you some. It might be worth using a simple variable resistor, I would use 5V and start around 680 Ohms, then start taking the current down by increasing the ohms. You may find you need to drop the current right down to around 5mA or even less.
I can try this Sunday if you like, I only have access to a Nikon D7000 or D3s, no bromine but can try and take night shots with different mA on external leds. I think in your case its going to be wave lengh though. The good news is IR leds definitely drift with mA adjustment once you get right down, I know this because its caused me no end of problems :D

phlogiston - 15-11-2014 at 03:59

A simple way to generate IR with a wider spectrum is to run a relatively high-power incandescent light bulb at a reduced voltage, e.g. using a variac.
Or perhaps you can get one of those lights made to keep certain animals warm (little chickens, turtles, etc).

However, I suspect that monochromatic light would be better for your purpose, provided you can find a wavelength at which the bromine is fully transparent.

Nice project

Little_Ghost_again - 15-11-2014 at 04:37

Just a thought.............. Some cams have a IR filter, you need to remove these little red filters on the can lens chip before it can see IR properly, if you take the lens bit off you should see a red filter, if it has one then take it off. This is how you get aq normal web cam to see in the dark

woelen - 15-11-2014 at 08:05

I think that my LEDs indeed have the wrong IR wavelength. I have seen that most LEDs have a wavelength around 850 nm and this may be too low for complete transparency. As suggested by phlogiston, I'll look for LEDs in the 900's nm range.

@Little_Ghost_again: The camera I have already is without IR-cut filter. It is a standard Raspberry PI camera, but without the IR-cut filter.

@food: Good to know how to disable the red LED on the camera. In my final version of the camera control code, I'll use GPIO32 to switch off the LED. Programming GPIOs is easy, I have done that quite a lot already on other devices (Beaglebone Black and Odroid U3) in the C-programming language.

Little_Ghost_again - 15-11-2014 at 08:50

Sorry I missed the 850nm wavelengh, do you have a old remote control unit? you can use the leds in those, look for ones with clear leds not the ones behind the dark screen, the clear ones are over 900 the ones behind the filter often are less.

Little_Ghost_again - 15-11-2014 at 08:52

The other thing I was going to mention was, this kind of technique is being used to read the dead sea scrolls etc, and for forensic work. Very interesting technique, I like day time pics taken with IR filters. Very atmospheric for woods and scenery