IR filter shenanigans
There’s a question we’ve been asked very frequently about the camera board. A number of you want to use it for night-time photography, and ask if we can remove the IR filter. Notably, London Zoo are planning to deploy the camera board and Pi in a number of camera traps in Africa, where they’ll be looking for nocturnal animals and for poachers. The problem is, we source the sensor/lens package as a sealed unit from Sunny, so we don’t have the option to remove the infra-red (IR) filter, which is sandwiched inside that unit. This causes trouble for those of you who want to take low-light pictures of wildlife, or for security or astronomy.
Over at Reading Hackspace, Gary Fletcher (also attached to London Zoo, and planning to use a Pi camera in his role at the Horniman Museum aquarium for detecting the night time spawning of corals, which he hopes to deploy in Guam in just over a month’s time) mailed us to ask about the filter. Eben didn’t hold out much hope for manual removal of the filter, but suggested that some very careful scalpel work might achieve results. If you attempt this, be aware that it’s not really a supported option, and that if you try you may break your camera board. (Gary, Barnaby and team did break the first board they attempted this with.) Also, be careful around IR sources if you’re playing with IR photography – the human eye doesn’t have a look-away or blink reflex associated with IR, and you can damage your eyes if you stare at a very bright IR light.
We think the results are worth it, though. Here’s some instruction and illustration from Reading Hackspace, with special thanks to Barnaby Shearer. First of all, here’s a howto video.
This video demonstrates IR pickup: you can watch the tip of a soldering iron change colour as it heats up.
Another demonstration, this time of night vision. The scene is illuminated with the IR from a television remote control.
Finally, here’s a demonstration of the pattern of light from a Kinect, filmed with the filter-less camera board.
The usual warnings apply, but if you do decide to try this yourself, we’d love to see the results. A huge thanks to Gary, Barnaby, and all at Reading Hackspace for being prepared to imperil a camera board, and for all the helpful video!
Edit to add: Gary has just left a comment below with more tips and instructions. Check it out if you’re planning to try this at home.
Some close-up shots in the IR filter removal tutorial would have helped a lot.
I see that closeups were filmed with a RPi camera that somebody else is holding near the “action” – can that video be posted too please ?
What he said!
I am afraid the close-ups didn’t come out as well as we hoped, but here they are: https://www.youtube.com/watch?v=e7MOb1vp5JU
Strap the raspberry camera to the stereo loupe.
I did this with good results:
The colours of this close-up movie are a good example of why you SHOULD NOT remove the IR filter unless you plan the use the camera ONLY for IR/Low Light.
Given the filter is glass, and the lens focal length is quite short, how much does removing the filter change the focus of the camera? or is it not noticable?
Gary said they didn’t notice any difference; I’ve yet to play with it myself, though!
Removing the filter itself didn’t have much effect, but you are dealing with very tight tolerances and it would be easy to add a small amount of extra space when gluing the lenses back in; this would shorten the focal length.
The lenses themselves are mounted on a screw thread, the is some glue to overcome but then you can adjust the focus by carefully turning the lenses inside the square mount.
I did a camera phone conversion a while ago. I found that if you squeeze the outside of the threaded collar at various angles with pliers, the glue cracked and I could just unscrew the lens. With the thread working, I could also adjust the focus.
Is the filter ‘glass’ and can it shatter? I was going to try removing the filter by glueing a shaft to it, then yanking so it’ll shatter
The filter is a very fragile glass sheet.
That’d be very interesting; if you do use the yank/shatter method, could you let us know whether it’s successful?
When you remove the enclosure there are 4 very fine recesses in each corner a very sharp fine scalpel can carefully prise it out. This is for most a one way conversion so it doesn’t matter if it cracks but if it shatters you may get dust in the lens or even scratch it. We’ve done a few cameras since and this part it fairly straight forward – removing the casing is the difficult part and not damaging the sensor.. More on that in a bit :)
Oh just re-read this and presumed incorrectly that you had the unit apart, I think you are talking about pulling it through the lens hole. No chance when it shatters and goes on the sensor you’ll scratch it and have a ton of little black specs – this isn’t going to work at all sorry.
Oh cripes: I’d made the same assumption. Also, Dean, be aware that dust is no respecter of gravity.
Great work. Have you determined if you will need a normal light removing filter to stop any exposure interference from the visible light spectrum?
This would probably depend on your application. We are planning to use them in total darkness so have left them unfiltered recording the extended spectrum.
sounds fine, if you are outside, moonlight can be rather bright (when there is no artificial light source), come to think of it in the middle of ‘nowhere’ even starlight is bright as I recall from African bush. Just a thought.
From lots of experimentation in the past, I found a black piece of colour photographic negative is an excellent ir pass filter.
IF you want it optional, put it on the front of the lens, instead of behind it. It works just as well there.
So a few tips…
When taking off the enclosure you will need the finest blade scalpel you can get your hands on a stanley knife won’t do it. You are not cutting you are scoring and prying.
You need to sit the camera on its side (with the ribbon cable denoting the bottom of the camera) then slowly run the blade through the TOP of the enclosure between the plastic and metal plate. If you use too much force you will go through and damage the sensor – it doesn’t take much. Essentially you need separate the units without the blade entering the enclosure, if the blade goes inside the enclosure you’ve had it. Once the top is scored, do the sides and then you should be able to prise the unit open very gently from the top being biased near a side edge.
The next factor you have to deal with is dust and dirt, if you get any on the sensor you’ll never get it all off. We had this issue on our third conversion and a compressor got all but a tiny dot off the sensor which we’ll live with but the rest were fine.
When you remove the enclosure there are 4 very fine recesses in each corner a very sharp fine scalpel can carefully prise it out. This is for most a one way conversion so it doesn’t matter if it cracks but if it shatters you may get dust in the lens or even scratch it.
We then assembled the units back together using the smallest amount of superglue possible – on the outside edges one the unit had been held in place (you may want to try the camera before gluing it?).
We didn’t need to do this step but you can then lightly score the glue tack on the outside lens ring and with a pair of pliers adjust the ring- be careful not to over tighten as you’ll crack the sensor.
Barnaby did on one camera we converted sand the enclosure down just under a mm before gluing it in place, this did allow much longer focal distances but the standard conversion was more than okay for us.
Hope this helps and someone can post up a more detailed pics/movie – ours didn’t come out so great.
Just a note that you don’t need a microscope but it is very handy and certainly was doing the first one or two conversion.
Excellent, it didnt appear as difficult as some have said :)
Nice videos of fingers ;O)
I hope A&Es are ready for some sliced fingers ;O)
Once you remove the housing, be extremely careful not to touch the the very fine bonding wires attached to the chip. They are likely to short against adjacent wires and are very difficult to move back without breaking
Does the optical filtering also serve as an optical low pass filter? If so it might be worth mentioning that a downside to removing the filter will be moires as well as inaccuracies with white balance and colour accuracy.
So why is the IR filter there in the first place?
Ah ok, funny colors…
So you can take photos in normal light. Which is what 99.9% of people are buying it for…
(I should be a bit more helpful here: if you want to learn more about IR photography and why you need a filter for normal usage, see http://en.wikipedia.org/wiki/Infrared_photography.)
This seems outside my dexterity. I would gladly pay for a low light capable camera.
Since there are a number of people interested in removing the IR filter, perhaps in the future a monochrome camera without an IR filter can be made available.
Please tell me what you think about feasibility of following options for RPi camera module:
Model C: RGBbayer-AAfilter-IRblock (normal photo/video)
Model BW: monochrome-noAAfilter-IRpass (night vision/astronomy/special)
Both models: standard size 1/3″ or 1/4″ sensor, screw for standard CCTV lenses, so anyone can buy tele/normal/wide/zoom/fisheye lens from CCTV seller? Where I live such lenses are sold for 4-9 Euro apiece, sometimes less.
These lenses are mounted in a screw so they have manual focus built-in, allowing for normal/macro work (almost a microscope). Tele lens for bird watching, fisheye for door eyehole (“judas”) or whole sky weather/meteor watching, wide/normal/zoom for everyday needs…
Optional equipment: camera module ribbon cable extension (ribbon with “board connector” at one end), possibly 30cm (12″) or 60cm (24″) – could anyone elaborate how long extension is feasible for signal not to be attenuated / drown in noise / otherwise unusable? Extension would be nice for mounting just the camera module e.g. at robot’s arm.
All these wish list items are interesting, but I suspect the Foundation will not be providing them. I think most (all?) board-level C/CS lens mount cameras do not use the CSI interface. They chose something small which is also cheap, due to huge mobile phone market.
Or maybe someone can invent a IR capable camera that doesn’t get funny colors in light?!
Not possible. If your eyes could see IR then some materials will be different colours from eyes that do not see IR. Simply something that appears black but reflects IR will not be black to an IR sensor if lit by IR. The same effect happens with colours. That is why the filtering is used.
But I’m sure you could do some software correction with a light sensor trigger. Or atleast make the IR filter easily removable.
I once did some work on a network camera. Removing the IR filter will typically degrade the color performance, since the color filters (red, green and blue) before the sensor are transparent for IR (on purpose, to get better performance in night mode) and hence IR “adds” to all three colors, making the image more “grayish”.
In our camera we actually had a custom motorized IR filter, switching between a plain glass (night mode) and IR filter (day mode).
Sunny should be sent a link to this page… they might actually be cool enough to make IR-ready modules available!
For the faint hearted, there is a new project currently running on “Kickstarter”, called “Infragram”. It is the creation of a low cost affordable 2 megapixel Infrared camera.
The creators of the project show in graphic detail, of how it works. And wait for it, also shown, is a picture of the ‘Raspberry Pi’ too.
Interestingly, they offer a filter set and a webcam version too.
Kickstarter direct link: http://www.kickstarter.com/projects/publiclab/infragram-the-infrared-photography-project?ref=category
For the video only of the project, a link to “youtube” : https://www.youtube.com/watch?v=ZVbC3UxlX8c
Having removed the IR filter I would be interested in how one can improve UV sensitivity. I have been considering getting my Canon 300D modified http://www.maxmax.com/IRCameraConversions.htm . But this could be an interesting project. IWe grow water lilies (over 70 varieties) and I positive the Tropical flowers will look distinctly different if I capture the UV light reflected from them, http://blogs.msmvps.com/mickyj/blog/2013/01/12/canon-dslr-eos-cameras-and-ultraviolet-uv-photography/
The Fujifilm IS PRO SLR camera was designed to respond to infrared (IR) and ultraviolet (UV) as well as visible light. There’s a small Flickr group dedicated to that camera at
where you can see the effect of no filter, IR-only and UV-only photography on a camera designed for the purpose.
(Very) broadly speaking you can probably expect similar results with a filterless Pi camera, although all the usual differences between a small (Pi) and large (IS-PRO) sensor will apply with regard to sensitivity, depth of field and so on.
VERY broadly, that is! ;)
Are they going to make a version without the ir filter? PLEASE DO!
Like I said, it’s not possible for us to source it without the filter; we buy the assembled lens/sensor as a package, and there are no options for us to buy a modified version: we get what we’re given in this case.
I cannot think of another company in the world that would actually tell you how to take a knife to one of its products and “cut it up.” This spirit of experimentation and “learn by doing” is not only what makes the RasPi project stand head and shoulders above similar endeavors, but it is what will lead to the greatest possible degree of success for those who follow and learn to lead us in to our technological future.
No IR filter means better sensitivey and infrared-“night shot” options but rather poor color fidelity, unless you add another IR filter in front. If anyone wants to make a “how-to” guide and needs some close-up photos of the (assembled) camera unit, I have some here you are welcome to use:https://plus.google.com/photos/109928236040342205185/albums/5759264418828082833?banner=pwa
I wonder if anyone in the community with the right tools at home or work would consider offering to make this change for a price? I would certainly like a non IR filter version for my astronomy project.
Knowing this lot, I’m pretty sure your wish will be answered.
Superglue?! Don’t use any superglue (cyanoacrylate) near optics. The fumes adhere to anything nearby & form a haze on optics. It can range from a heavy white film to slight enough to go unnoticed.
Its goes on the outside of the enclosure and if you get a white blob you’ve used wayyyy too much!
i managed to do a timelapse which is on my blog thanks to a helpful blog by designspark.
however how many units would sunny need as a minimuim order to do one without ir filter? im sure if enough people showed interest and you asked sunny explaining about how these can be marketed as night cameras and explain the educational benefits that something could be sorted out, after all over a million pi’s sold surely that says something ;) btw add me down for a camera without ir inside :)
It’s not clear to us where in the chain the IR filter gets put on – if it’s Sunny, they might be prepared to do it at the some tens of thousands point, but Omnivision (who make the sensor) might well not, because they’re used to dealing with mobile phone type numbers (i.e. tens of millions). We’ll ask about it; it’s an interesting question.
No doubt pre-modified cameras will be for sale on ebay within a week ;)
I think the warning re IR danger is incorrect. The lens inside your eye filters out almost all IR wavelengths and that is why you can’t see the output from your TV remote. Think about an electric radiant heater– it pumps out a metric ton of IR but it won’t hurt you to stare at it short of thermal damage. Your “look away” reflex will hopefully kick in before you cook your face.
The only time IR light can be a danger is from an IR laser such as a CO2 laser which can be used to cut steel. In which case, don’t look into beam with remaining eye.
Thanks so much to the Reading Hackerspace for this! I tried to do a close-up view of the process on my video microscope for those looking for more detail:
Father’s day is soon approaching and this would have been my number #1 desire…except until I found out about the IR filter issue which is going to be an issue for my intended project. Although I appreciate the solution to this, I’m hoping future versions of this camera will have a easily removable IR filter.
“warning re IR danger” is valid
Just because you can’t see it doesn’t mean it can’t hurt you.
“CO2 laser which can be used to cut steel”
No – even pen-sized green lasers are dangerous. They usually are IR lasers with a frequency-doubling crystal. I think some cheap ones have been found to miss out the IR-blocking filter, so you might get 5mW green plus 50mW IR. These are much more dangerous than they look. If they light matches or pop balloons, beware !
I’m interested in modified camera, can somebody sale me one ?
Is anyone selling just the Sunny camera modules? I have now damaged 2 – 1 during IR filter conversion, and 1 after conversion using too much effort to remove dust…
I was wondering where I could get my hand on just the little module itself, as the carrier boards and assembly are a large part of the cost. I suppose I will suck it up and just order the noir board next time, but in the meanwhile I have 2 carrier boards and no working cameras.
It’s all my fault because I can’t leave well enough alone, but I am just looking for a cheaper way to get back in the game. As best I can tell, none of the normal channels sell the camera module alone, and i don’t know if it’s an off the shelf model to begin with.