NightVision
Contents |
Motivation
In continuing with adding senses to the human body, night vision (or near infrared vision) has been available commercially even at low prices for some time. Even now there are night vision toys and night vision goggles being given away with special edition video games. On top of this there are also many examples of DIY/Homebrew night vision. There is, however, to our knowledge no open source kit/collaboration on night vision. We intend to source and package a consistent high quality set of parts as a platform to build on.
Market Survey
Professional Night Vision
Looks like pro monoculars with Gen1 tech come in as low as $150 though apparently gen1 technology has a fish eye thing going on. Gen 2 stuff starts at about $450.
You can also buy separate IR illuminators. These are most certainly Infrared lasers.
Toys
Toy brands and price ranges include Eyeclops for $50-$75, Discovery for $30, and Spyhawk for $13.
Spy Car
Wild Planet is the leader of the low price toy night vision market. They released a Night Vision Spy Car which has seen a good bit of hacking here and here. They also sell some sort of cheap night vision goggle now as well. There is new model to be released for Christmas 2010 called the TRAKR which would be good to dissect. The now unavailable Spy Car apparently utilizes a Kopin screen. Kopin quoted me at $22 each for 10-99 pieces of the 300M sold directly through them. They use a OVT OV5116 B/W QVGA CCD with a DY-301P lens and an large IR led of undetermined origin.
DIY
There are tons of homebrew/diy night vision out there. The majority of these instruct to take apart an old camcorder/camera to harvest the expensive lcd, lenses and imagers. These cameras generally remove IR as humans aren't used to seeing it and it would distort and wash out images. The homebrew guides generally then begin to look into adding more infrared light to the scene. The most common method of doing this is adding an array of IR leds. The Evil Genius series of books on Spy Gadgets suggest an interesting method of illuminating a scene by using a non focused (lens removd) IR laser. He also mentions that green lasers, ~530nm in the spectrum, aren't actually manufacturable and that they are actually composed of IR diodes.
Theory
Night vision looks to illuminate a scene with light that is not visible to the human eye. Presumably we would like to illuminate the scene in such a way as to be discrete or we would use visible light.
http://en.wikipedia.org/wiki/Infrared Breaks the infrared spectrum to between 700-30,000nm. http://electronics.howstuffworks.com/gadgets/other-gadgets/nightvision1.htm Further breaks that down and explains its the applications. .7 to about 1.3 are generally consisered near infrared and are what we are talking about when we speak of night vision. Mid range is where IR remotes tend to function. Above that is thermal infrared.
Wikipedia explains 700nm is where the visible spectrum tends to fade, though focused light up to 880nm can be seen as a faint red light. As such we want to stay above 700nm and go even higher if we do any focusing.
Wikipedia breaks down the professional night vision technology well in their article. It appears most seem to use light intensifier tubes projecting photons onto phosphor screens. For illumination they use lasers (and have warnings as such). The toy and cheap versions of night vision use IR leds to illuminate with monochrome lcds and a cmos/ccd with high lux and no IR blocking filters.
Theramal Infrared
A branch of infrared night vision is thermal imaging. (Heat Mapping) Thermal imaging would be of course amazing, but all research so far seems to indicate that currently available ccd/cmos technology is still incredibly expensive. Unless we can figure out a cheap way of imaging this seems out of reach for now.
What we can offer
A couple options so far:
- kit under 100 for sure, under 50 would be far better
- otherwise people are looking for low cost option for video, night vision open source schematic -- seems like it would be appreciated to have something which is more maleable than altering the toy stuff
- are pro items hackable? take some apart
- document ways to hack existing toys, add different ir leds, lcd screens, different applications, etc
- worst case scenario document roundup between ir/lasers, beam angles and brightness
Process
We will generally want to use a cheap low resolution camera, probably black and white though maybe color. We will want a lens to focus light onto our imaging sensor (ccd,cmos) (sans any kind of infrared blocking filters, and possibly with the addition of white blocking filters). We will want some sort of infrared illuminator to bring (invisible) light to the scene. Finally we will want a comfortable and adjustable mounting for all this, hopefully made available on thingiverse.
LED IR emitters
Frankly the options in LED illuminators is pretty poor. These diodes are generally so incredibly low in brightness (<1 lumen) we can't expect to illuminate much more than in front of our faces. The general solution to this has been to create arrays of these leds and overdrive and pulse them like crazy to squeak out as much illumination as possible. There are exeptions, but they are expensive. So far I have sourced an emitter at digikey made expressly for IR CMOS illumination, but its $82 dollars...
Testing so far has led to somewhat obvious conclusions. LED emitters are tough to get bright enough. Tighter beam areas (a 6deg beam was tested) can illuminate at least a couple feet but with a small beam area. A 60deg emitter on the other hand barely transmits a few feet. Obviously if you wanted to work in front of your face a 60deg beam would be appropriate but further research will probably be focused on multiple 6deg leds.
Lasers
So we may want to look into the previously mentioned green and or IR lasers for illuminating the scene. Sadly IR lasers are indeed expensive as there seems to be little call for them in the consumer market. Green on the other hand are incredibly common. Apparently cheap Chinese green lasers on ebay are generally to be avoided by laser enthusiasts as they give less visible light bang for their buck and are more dangerous due to a lack of IR filtering. This lack of IR filtering is actually kind of good for us though. We could take out the green aspects of the laser, or we could simply filter it afterwards.
Danger
Any light, focused, is dangerous to the human retina. Non visible spectrum light like Infrared and UV are, however, extremely dangerous to human vision. Welding goggles for instance are specifically designed to protect us from this part of the spectrum! Lasers for instance are classified generally into 3 classes. http://en.wikipedia.org/wiki/Laser#Laser_safety The idea is, by the time the light could damage your eye, you would look away and would know you were in danger during that time. Higher class lasers are so powerful by the time you knew you had damaged your retinas, they would already be damaged. IR (and UV) are however invisible to the human eye, so you would never know you were or had done damage to your or anyone else's eyes.
Filters
A red filter will remove Violet, Blue, Green and part of Yellow. It will allow through Red, Orange and some Yellow which we don't care about. A red filter would remove any of the green laser left over if we didn't wish to modify it to remove the green logic within.
Also used heavily in commercial grade night vision are visible light filters at the ccd/cmos. By removing visible light you can remove any washing out of the image and apparently create a cleaner image. Wikipedia mentions polysulfone which we should look into.
Sourcing IR Emitters
- http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail&name=475-2844-1-ND
- 120deg max dc 1amp 320mW/sr @ 1A but typically 530mW/Sr (mw/Sr is = to candela I believe)
- $12 interesting if not expensive
- http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail&name=475-1200-ND
- 6deg but 400mcd!--Wonder if smaller angle will help, many smaller angle emitters for more pinpoint viewing?
- $.69
- really nice for several foot distance at least
- http://search.digikey.com/scripts/DkSearch/dksus.dll?lang=en&site=US&WT.z_homepage_link=hp_go_button&KeyWords=SFH4740&x=0&y=0
- $82 1amp 1200mcd
- also offer the SFH4730 900mcd, different wavelength
- Obviously too expensive. Apparently directly manufactured for this purpose. Will research to see what we're up against.
- Need to find mating connector to test this.
- http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail&name=67-1001-ND
- 60deg 60 candela at 100ma
- $.27 much cheaper... but do NOTHING..sigh
- http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail&name=475-2846-ND
- 4751 $74 1amp 940nm >630 900mcd typ
- nah
- http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail&name=475-2845-ND
- 4750 $74 1amp 850nm >630 1000mcd typ
- nah
Sourcing LCD's
The idea here would probably be to go BW and cheap. I would like to prototype a nicer display to begin to build up general purpose HUD on it, but am not sure how feasible that will be along side this project with the possible necessity of color filtering. Current sourcing through general electronics vendors such as digikey has led to very few promising results which are all very over priced. Sparkfun sells a few cheap lcd's including some knock off Nokia displays. I am currently thinking of going with these cheap cellphone displays from the source. There displays available through Chinese marketplaces such as aliexpress for $5-15 dollars.
Todo
- test leds for illumination, difference in mcd and beam angle
- source cmos sensors
- source lenses
- source lcd
- prototype
- create a housing which is comfortable and adjustable