By Cory Doctorow at 11:09 pm Sat, Jun 19, 2010
Beautiful Photos of Infrared Trees
Reminds me of the view out the back of this old wardrobe at my grandparents.
Stunning! I love how the infrared picks up the details in the leaves. Taking pictures of bright green trees in colour, that kind of detail can be hard to pick up, but when the tree is acting like a black box, it’s very striking.
I admit, my first thought was, “Umm—why do the trees’ reflections in the cold water show up as warm‽” But of course, infrared reflects too…
The assertion that “anything alive” shows up white is not really correct. Typical infrared photography captures near IR (around 800nm), which wouldn’t give you any useful data about biological heat. You’re thinking of thermal imaging, which captures significantly longer wavelengths, around 15um or so.
What you’re seeing here is reflection of IR from sunlight by cell walls (these reflect strongly at these wavelengths, while chlorophyll is largely transparent), in combination with some filters, a false color scheme and probably some compositing with a visible light photo (I’m assuming these are shot digital with an IR filter).
Interestingly, one of the early uses of infrared film was to distinguish camouflage from real foliage (don’t expect ; it’s also used to gauge plant health. Take a look at some real film photos, they are way crazier — the pigments totally jumble and compress the color space, so it gets pretty trippy.
I couldn’t tell you why the geese are of a similar tone to the plants, really depends on the material of the feathers. You can experiment with some local waterfowl yourself, just make a pair of passive IR goggles: your eyes are somewhat sensitive to the spectrum used in this photo but that gets totally downregulated by much higher response to visible wavelengths, replace the filters in some welding goggles with a combination of lighting gels that’s non-transmissive across the visible and lets in something at ~700nm, wait a few minutes to adjust and you’re golden. You’ll probably find that various animals have totally different “coloration” than you’d expect.
This is a cool gallery (especially if you haven’t seen IR images before) but Cory, you really need to hold the commentary on subjects you’re unfamiliar with (this is a real trend here on BB, interesting link with shallow or totally incorrect editorial comments). Being a linkblog is not that embarrassing.
I read your post and your knowledge is truly impressive, but I don’t see how what Corey wrote was incorrect. I think boingboing is as popular as it is because the editors are able to consciously explain in layman’s terms complicated topics. How many people read your blog?
It’s incorrect because things that show up white do not do so because they are warm and/or alive, and not everything warm and alive shows up white. In other words, it’s completely incorrect.
I don’t agree that they shouldn’t comment on things out of their depth, though. (What could they post then? Har har.)
Corey implied correlation, not causation.
If Dick Cheney was in this shot he would be as black as coal, and the geese would be dead.
Nope. If Dick Cheney were in this shot, he would show up in full color, his shooting buddies would be in white, laying about in obvious pain and the geese would be wandering around confused about the goings-on. ;)
“anything warm and alive shows up as a ghostly white” – Eek! the birds have cold, dead necks, and warm, alive leaves grow from cold dead zombie trunks!
Quick! what does blue signify? and red reflections?
@ _qux : thanks for the nice clarification;
I might add that opening a cheap webcam (~4 screws), locating the tiny, slightly blue-tinted piece of glass in front of the sensor, easing it off and reassembling the webcam gives one possession of an visible+infrared imager. To get a pure IR imager in the 800nm range, one has to put a filter before the webcam (black to the eye, but translucent in IR, buy one, or pick up a ‘black’ slide film (or, considering the miniscule opening of a webcam, some black part of any slide of film). Both CCD and CMOS-type cameras can be hacked thusly, as they are both more sensitive to IR than visible; very new CMOS cams tend to have the filter (blue tinted glass) tightly integrated on the chip, though.
@ Cory: I second _qux’ call for less shallow comments.
That’s a tempting hack for a couple of webcams I have here. And I have a screwdriver at the ready.
What sort of changes does that have on the dusk/night sensitivity of the camera? These ones will provide fairly reasonable dusk sensitivity by dropping the framerate as the light goes and provide images long after the naked eye can’t cope.
I would assume that taking off the tinted filter would provide a little better night sensitivity (although stranger colours).
The hack will change what the camera perceives. Thusly, at lighting conditions that appear dark to the human eye, the camera might now see something, because there is some IR illumination (by pressing a button on your TV remote you activate the IR LED in the remote, giving a small IR source, for instance)
am i the only one thankful to not have infrared vision?
If I may veer of topic a lot for a second:
I disagree with _qux and loonquawl concerning Cory’s “shallow” post comments. Don’t get me wrong all the mutants are guilty of commenting outside their respective areas of expertise. They are quickly corrected by members of the BB community who ARE experts in the topic at hand. Often the post gets amended and credit is given. Win!
I would like to think the mutants do this on purpose to stimulate debate.
However when Cory types “Legos” he is just screwing with us. ;)
all digital cameras have infrared filters over the CCD, usually over the whole lens, and if you’ve ever seen photos taken in low light conditions from some cameras (my old nokia 6300 phone did this) you get a lot of monochrome green and black images in the dark. i think i still have a photo i took with that camera in the dark and sure enough you can see it. like someone said before there is a small IR response in human vision (as there also is a small UV response) and it’s part of what causes the gritty graininess of vision at night time. anyway, yes, any digital camera can have this IR filter removed (it can be identified by it’s reddish appearance) and combined with an IR LED, you can gain a kind of poor-man’s nightvision device. personally i’m surprised that this type of combination has not appeared as a budget nightvision system, standard night vision gear is upwards of aud$500 for a single one and they work usually by a set of lenses and infrared sensitive phosphors that light up when struck with infrared light, which comes from the illuminator built into the device, and i think this image is amplified somehow… anyway it’s not really that complicated.
the thing that bugs me about these images is that there is any colour at all. clearly they must be edited composites of a photo taken two in very quick succession one with the filter on the other with the filter off, or possibly even with a partial mirror system to grab two images and one filtered, and then it’s been carefully overlaid, possibly using a colour filter to eliminate all the red and green, and the black-green-white pattern of the infrared is collapsed to straight black and white.
like someone said it is not thermal vision, pretty much all sources of visible light also emit infrared, it is only when you filter out the visible from the light output or you use a device such as an LED that only emits a very small band that you get anything different. non-illuminated night vision devices can be used to spot other people using night vision devices with illuminators. very hot objects glow with visible light so it’s not any different aside from the lack of colour, and from simple examination you cannot see whether an object is simply highly IR reflective or an emitter in normal light conditions.
There are many low-cost ‘night vision’ devices using a CCD+display setup. the night vision goggles that were given as a gimmick with this FPS (Battlefield something-or-other i think) a year(+-) ago were like that, as are some cheap children’s toys. The problem lies in the source of illumination. Normal unlightened rooms&outdoors usually lack 800nm radiation just as much as visible radiation. True nightvision will source in the spectrum above 1500nm (1.5um), where much more radiation is left, thus enabling to see in ‘the dark’. For the 800nm range, you need a 800nm lighting (that does not emit in the visible range, to have fun), so a IR LED array is needed, having problems in terms of allowed intensity (eye safety) and range (at low intensity).
These aren’t infra-red or thermal. I believe the photoshopper matted the sky and water out, made the trees B&W and then made it a negative. If it was infra-red the skin of the farmer would be lighter, not darker. Also, note that the mountains in the background are treated the same way as the sky, despite being covered with trees. The photos are pretty, but the technical aspect is incorrect.
the sony f117 nightshot will flip away the IR when in nightshot mode. combine that with an IR pass filter that lets through only IR and you have the perfect combination.
the blue stuff in the picture is photoshop and not IR. IR pics often have a pinkish tint to them, or a little green. but never blue.
Sites like the source here, that scrape photos from the web and then show them without attribution really bug me. Anyone know the original source (TinEye is down for the weekend)?
That said, in response to the comment about color and that these must be composites, DSLRs that are modified for IR photography usually do show color, although it takes a bit of histogram stretching and the like to recover it. The sky typically shows up red, not blue, so my guess is that these pictures were photoshopped to swap the red and blue channels.
This picture at wikipedia shows the colors one would expect from an IR DSLR.
Although I do think these are, at least partially, real IR photos, it should be noted that IR can be faked to some degree (not 100% convincingly) in Photoshop by hand mixing the color channels when converting to black and white. You can brighten the green and yellow channels and darken the blue and aqua channels and have a fair approximation of IR photography. I moderately applied it to a photo I took a few weeks ago here: http://www.flickr.com/photos/pixorama/4698099279/sizes/l/in/set-72157624269946184/
It’s not as interesting as the real thing, but still an decent effect.
@ElStinko: Private flickr images aren’t viewable by others.
Personally, I like that BB is self-correcting because it allows comments. Better that, than no commentary for fear of being wrong occasionally.
This looks more like UV pictures to me. Trees being warmer than geese? No way! This is a picture taken with a lens that doesn’t filter out UV for sure.
Just to pile on with my understanding of a widely misunderstood subject.
Basically: “infrared” is a HUGE swath of spectrum, really really huge and there are two distinct common areas used in imaging.
This is “near infrared”, it’s basically just normal light that we can’t see. Some bugs might be able to see it and it doesn’t really have anything to do with the temperature of the object.
On the other hand ALL objects emit black body radiation dependent on their temperature. This is also infrared but it’s nowhere near, for the purposes of photography it’s something else entirely. Using different more expensive equipment you can sense this far infrared. This is also known as thermal imaging and it’s the same technology used in touchless thermometers.
It gets even more confusing because both are used for a type of night vision. A near infrared camera with IR lights can be used to record video without most people being aware of the lighting, but you are shining a light and someone else with near IR vision capability could see this. It’s not suitable for special ops, ninjas, etc so they use proper night-vision thermal-imaging goggles which can sense body heat without any additional illumination.
Near infrared looks like black and white photography with some interesting effects, thermal imaging is the predator stuff, it’s often false color.
In particular near IR photography records light that was mostly reflected and consequently has the normal shading properties of reflected light that provide a lot of shape cues. Thermal imaging is basically like looking at a fire and doesn’t give the same impression of shape (you get silhouettes obviously but reflected light gives you cues as to the standard surface properties like angle, normal, roughness, glossiness, albedo, etc.)
Anyway, apologies for a horribly rambling note.
If anyone wants to do this, there is an incredibly simple hack for the Sigma SD14 DSLR. The IR filter in the body takes about 20 seconds to remove or replace.
Then you can spend hours getting frustrated shooting through a black viewfinder and post processing :)
The near IR goggles using a few colored filters are pretty awesome too :) Much cheaper and your eyes capture detail and range that the camera can’t.
…and cheapskates might like to know that you can do the same for most webcams. This Instructable gives an example.
Thanks Beanolini! Can’t wait to try this hack!
Here’s a guy explaining how he shoots with colour near infrared film:
One big problem with these unattributed photos is that there’s no attempt to explain how they were taken, so instead we get a bunch of wild speculation in this thread.
600×400, what amazing resolution.
I somehow doubt the clouds are warm and alive. Yet they are white. Go figure.
Since we can see a blue sky, either it’s been photographed with visible light and IR, or it’s been tweaked in processing. (Not gonna say Photoshop, dammit, there is other software like gimp that does the same job)
>> either it’s been photographed with visible light and IR
That’s unlikely. Most digital cameras render the near-visible IR spectrum in shades of violet and red. If you white balance to make foliage white, the sky usually comes out red. In photoshop, it’s a quick matter to swap your red and blue color channels in the channel mixer, and, voila: IR images where the once-red sky is now blue.
Pretty images; the aesthetic effect IR brings out with life other than plants is so often ignored because of the dramatic effect it has on foliage. Good to see some critters in the mix; they *do* look ghostly.
Extraordinarily cool…the pic here just became my new desktop background.
Frank Lemire in Toronto has been shooting infrared digital landscapes for many years too and has a lot of work up at http://beyondsight.ca/
For best effect though, try to find out when he has a gallery show on since the prints are particularly stunning. It’s all very “familiar scenes rendered in a spectrally faithful yet unfamiliar and therefore seemingly unreal way”.
A computer monitor outputs in three colors – Display Red, Display Green, and Display Blue.
When you display a natural-color image, you are mapping:
Display Red = Image Red
Display Green = Image Green
Display Blue = Image Blue
Now, “Image Red” correlates to the color of the sensor on the camera, which camera manufacturers try to match to the absorption of the human eye’s cone cells on what is actually a continuous spectrum.
To portray a grayscale image on an RGB monitor, the only necessity is that:
Display Red == Display Green == Display Blue
Anything that fits this condition will appear in shades of gray. These can be mapped to any image color you want. You can make a grayscale image by mapping only the green image band to all display bands, or only the red image band to all display bands, or only the X-ray band to all the display bands, or anything along those lines. Even electron microscopists typically release grayscale images, and they’re not even working with photons. Cory incorrectly inferred that because this image has grayscale portions, they represented infrared, because grayscale is often used to display near infrared by setting all display bands to ‘Image Infrared’.
A ‘true’ grayscale image, like B&W film, is one which averages all three image bands and then sends three identical copies to R, G, and B display subpixels.
As an aside: In night vision goggles (which generally deal with thermal infrared), they may use ‘display pixels’ of a single color (white or green) since they generally only sample one image band. Thermal vision in science and engineering, on the other hand, often performs an unusual transform and turns ‘image brightness’ into display spectral color of a single brightness, where blue is cold and red is hot.
The close end of the spectral chunk we term ‘Near Infrared’ involves a region just beyond red, where the red variety of cone cells’ sensitivity fades to nothing. ‘Near Infrared’ exhibits dramatic reflectivity in healthy plant life, which is typically very dark in the ‘Red’ portion of the spectrum. Unhealthy or dead plant life is usually the reverse, and so the ratios between ‘Near Infrared’ and ‘Red’ form a central diagnostic component when sensing the health of vegetation (see NDVI).
The ‘False Color Infrared’ format, which is very common in crop surveys, is one in which the following mapping takes place:
Display Red = Image Near Infrared
Display Green = Image Red
Display Blue = Image Green
This results in a colorful(non-grayscale) image with greatly enhanced contrast between different types of vegetation, just not the colors you expect of vegetation.
In this [beautiful] image, the water and the sky appear as natural-color, if a bit oversaturated… while the rest is in grayscale. You literally can’t do this with simple spectral substitution – you have to mask out parts of the image and recolor them separately. I wouldn’t be surprised, given the shading, if some HDR-like effects had been used as well. This is likely a natural-color image that has had some heavy-duty processing applied in GIMP or Photoshop.
Disclaimer: As always, this is a ‘Good Enough’ approximation of how the world works, and avoids the more nuanced issues in retinal psycho-optics, camera/display optics, physics, remote sensing, plant biology, and photoshoppery.
It’s true, but one big problem with these unattributed photos is that there’s no attempt to explain how they were taken, so instead we get a bunch of wild speculation in this thread.
Even book publishers and book publishers need to attribute their credits properly…otherwise what’s the point?
Mail (will not be published) (required)
Submit a tip
The rules you agree to by using this website.
Who will be eaten first?
Jason Weisberger, Publisher
Ken Snider, Sysadmin