How to do (Near-) Infrared Photography - DIY Optics

Infrared photography is just simply otherworldly. What people probably like around it, that it's real but at the same time it just looks like it's photoshopped. Anyway, it's a very special effect which is hard to create simply with image processing.

IR photography is about capturing an image by using only the near-infrared light part of the spectrum. Our eyes can see up to ca. 700nm (red) wavelength, but CCD or CMOS camera sensors are capable of detecting light up to 1100nm. It's gonna be less efficient and altogether there is less infrared light from the sun, but it is still plenty to take photos. There are for example security cameras which have "night vision" mode. This usually means that the scene is illuminated using strong 800nm or 940nm infrared LEDs which the camera can but humans can't see.

Eye vs. Si sensitivity (Source: maxmax.com)

Security camera example with IR LEDs:

Similar effect when you check your TV remote's front window with your phone camera and see a flashing red/purple light inside. The remote communicates with the TV using a flashing LED.

It works with some phone cameras better than with others. That's because the manufacturer uses a filter which gets rid of the IR light. IR not being visible makes the photo look strange (otherworldly!).

Ever noticed on old photographs that the eyes look kind of scary? They had no built in IR filter back then..

Camera for IR photography
To do IR photography one needs to increase the amount of IR light hitting the sensor compared to visible light. This can be done in a numerous ways:
1) IR illumination with LEDs
2) Visible light blocking filter with a normal RGB camera
3) Using a RGB camera which doesn't have a IR blocking filter
4) Using a monochrome camera which doesn't have a IR blocking filter with a visible blocking filter.

1) is used in security cameras in the night. When there is daylight, it is too powerful to see any effect from the LEDs. As a comparison: an IR LED array (50 LEDs) has about 0.05W power and from the sun 1000W power reaches every square meter of the ground. --> This method works only in the dark.

2) If you use a visible light blocking filter with your everyday camera, the contrast will be good enough to do infrared photos in the daylight. The sunlight has plenty of near-infrared light and your camera sensor's sensitivity is good enough even if they built in a IR blocking filter. These IR blocking filters will still let through a few percent of IR light. --> Works, but all the photos will be red and needs strong daylight.

3) The IR blocking filter can be removed from your camera to increase the amount of IR light reaching your sensor. These filters (windows) are placed directly in front of the sensor and glued-in using some adhesive. It's possible to remove it and replace this with a broadband antireflection coated window, but it's a risky operation. Clean room environment is needed to prevent contamination of the sensor surface. Furthermore, you will still have to post-process your photos because the camera doesn't know you removed the filter. --> Works really good, but it needs expertise and professional equipment to replace the window. Photos still need post-processing.

3B) Sometimes not just the IR blocking filter but also the Bayer array filter is removed. Every pixel of the RGB camera sensor has a red, green, blue filter. This is now the sensor can distinguish the colours. Removing this is very risky and will also remove the micro lenses which sit on top of every pixel. There are a couple of companies how do conversions like this: maxmax.com or kolarivision are the most popular.

4) The ultimate solution is that the camera sensor comes from the manufacturer without Bayer layer (so it's a monochrome camera) and without the IR blocking filter. These kind of cameras are rare in the consumer market and for that also very expensive. Leica Monochome M costs nearly 10,000 dollars. However, there are alternatives:


Cost effective monochrome infrared photography

Solution A
Industrial cameras are widely available as monochrome because often IR light is used for inspections on the production line. Sensor size will be nothing near full-frame and they usually need a desktop software to capture images. Furthermore, you will have to dive into the world of C-mount objective lenses to select a matching one. It's doable but will need extensive research and reading if you are not familiar with these devices. DIY-Optics.com can help you with selecting the fitting parts. Look for Optics/Laser Consulting Services in the webshop.



Solution B
There is one more consumer style market where you will find monochrome camera for IR photography: Astrophotography. Monochrome will cost more than RGB, but around 1000-1500 instead of 10,000. Sensor size is decent (to capture more light from the stars) but still not full-frame. Nearly any objective lens can be attached with different adapters. No complicated software or connection set needed like with the industrial cameras.

My gear:
After 4 years of infrared photography attempts I settled to the monochrome astrocamera solution. I don't think I will ever buy a Leica and options 1)-3) have their disadvantages..
I use a ASI294MM astrocamera, a KOWA LM50-IR-P lens and a 850nm filter.

The camera can shoot up to 46MP photos (bin0 mode) but as standard (bin1 mode) 12MP. The objective lens is 50mm effective focal length and it was designed and antireflection coated for near infrared. This lens needs a M42 to T-mount adaptor and a 20.46mm extension to have the right image flange distance. (20mm T2 extender and 2x 0.2mm spacer rings). Spacer rings of any size are available in the DIY-Optics.com webshop.
I designed a little 3D printed holder for the camera and lens because the total weight is now nearly 1kg and I didn't want the camera to hold all this weight.



The camera gets it's power from my Android smartphone through the USB port and the photos are directly saved on the phone using the ASICap App from the Play Store. Exposure time, gain, binning, formats, image analysis ... can be set up directly in the App.



The best IR scenes
There are a couple of interesting effects which an IR photographer will see.

Water on IR photos appears to be very dark. This is easy to explain. Water is blue and sometimes green, but these colours are filtered out. Red and infrared is missing anyway so nothing will appear dark/black on the photos. However, you will have to use a infrared polarizer (expensive) or do your photos with the sun behind you to reach better contrast. On the surface of the water there is a small reflection which is called Fresnel reflection, about 4% of the incident light. This reflection will get directly into your camera if the sun is in front of you. The reflected light has more of one polarisation than the other and can be filtered out using a polarization filter.



The sky has a similar effect. The sky is blue because blue light scatters on air molecules more than the other wavelength. If you filter out the blue, you will see still the scattering of the infrared but there is less infrared light in the sunlight anyway so it will appear darker than on RBG photos. To make the sky as dark as possible, you will have to have again the sun behind you. Back-scattering from the molecules is much less then forward scattering. Also the scattering will depend on polarization so a polarization filter (for IR) will also help.

The human body will appear a bit different in IR. Infrared light can penetrate the tissues deeper than visible. This will allow your camera to see deeper under the skin. The veins will be more visible and the eyes will look more transparent ("scary").

Photo example veins and eyes

Plant are probably the most visible difference in infrared. Plants are green becasue they absorb more blue and red light than green. However, they do absorb a large amount of green too, but our eyes are most responsive for the green light so we will see them green. Infrared light will be however reflected 5 times more than green. We can't see it but the camera can very well so the plant will appear very bright on the IR photos.



The high reflectance of the plants is used in agriculture. Multi and hyperspectral cameras using different environmental monitoring indices (like NDVI) can differentiate healthy and dry/sick plant very effectively.

Which infrared filter?
There are a couple of different standard infrared filters widely available. 720nm, 850nm and 920nm long pass (visible blocking) filters are the cheapest. There is no big difference between the apprearance of the image either of these you use. For me 720nm is just too close to the visible and 920nm is really the edge of the wavelength range which the lenses are designed for. 850nm is just ideal.
There are narrow band filters (like H-Alpha, Oxygen-III) which let through only a very specific wavelength. This is useful for astrophotography but makes no difference in IR landscape photography (it will only reduce the light what you get to the sensor).
The third type of interesting filter is the Silicon (Si) filter. This is a thin disk of silicon. Silicon absorbs/reflects everything below about 1050nm. It will highly reduce the amount of light reaching the camera sensor and the sensor will be very inefficient too (<3%) to convert this light into electric signal. This means, the Si filter can be used only with a very intense light source: direct imaging of the sun.



Polarisation
As mentioned earlier, you can increase contrast using a polarisation filter. Unfortunately a standard polarization filter just won't do. Polarization filters can filter one or the other polarisation of light with different efficiency at different wavelength. To make it work at IR, you will need a wire grid polarizer. These polarizers are unfortunately very expensive. A 50mm diameter filter would set you back about 1500 dollars best case. Having the sun behind you will cost you only the planning effort.

Finally, if you are interested of seeing the world in infrared, there is plenty of material on Instagram. My infrared photos and sometimes short wave infrared (1500nm-2500nm) satellite images are available here:Dan C (@behyperspectral) • Instagram photos and videos

Good luck and just contact us if you have any questions.