Slashdot Mirror
Making a Homemade Webcam?
Space-Bot asks: "I remember back in high-school photography the simple and very basic homemade cameras that we made that surprisingly worked fairly well. Amazing how something so great started so basic. These days we have all these high tech gadgets that do it all so quickly you never really think about any of the work behind it. Well I would like to start to understand the modern digital cameras more and I figure what better way then to make a homemade webcam of some sort. Might some of you Slashdot guru's have some ideas or experience for my project?"
Not "guru's". There's no need for an apostrophe there, you twit.
That's about as home-made as you can get, given how cheap "real" (but crap quality) USB webcams are these days.
Got time? Spend some of it coding or testing
Step 1: Buy a webcam.
Step 2: Take webcam apart.
Step 3: Make a webcam out of parts from step 2.
A webcam is a lot more electronics than optics, so your high school photography class won't help much here. A lot of the stuff that goes into a webcam is going to be surface mount only and very tiny. I understand you want to learn about them, but you might be better off buying one and taking it apart and studying that.
...like this guy. You can't get much closer to making your own webcam than this. It's not like you can print a CMOS Image Sensor from your bubblejet.
There was an improper usage of an apostrophy, and Everybody was sure that Somebody would point it out.
Anybody could have done it, but Nobody did it.
Somebody got angry about that because it was Everybody's job.
Everybody thought that Anybody could do it, but Nobody realized that Everybody wouldn't do it.
It ended up that Everybody blamed Somebody when Nobody did what Anybody could have done.
09F91102 no, 455FE104 nope, F190A1E8 uh-uh, 7A5F8A09 that's not it, C87294CE no. Ah! 452F6E403CDF10714E41DFAA257D313F.
It'd be quite difficult to make a digital web cam. Sure, basic photography is simple. The reason? All you need to do is capture the light, and you've got a nice chemical compound that does that fine. You don't have to delve into the actual process of getting it to capture (yea you have to develop it, but the chemical compound on the film does the capturing).
With anything digital, you have to use a matrix of photo-sensitive sensors, process, and send them out to the computer. Which means you either need to buy a CMOS board, or that other kind of photographic digital thing. Figure out how it interfaces, connect a USB interfacing chip onto it (I think they're pretty cheap, Buffer->USB->Program, you handle the arrows and the Program, everything else you'd practically have to buy. I guess you could create the USB interfacing yourself, but that would be tedious, and not important. Using the serial or parallel ports would be easier if you're going to do it yourself, btw.)
Anyway, what another poster said. Go buy a cheap rebate webcam, take it apart, play with the parts some, and put it back together. I'm pretty sure there's nothing that's going to be hurt by light or by touching in a webcam (though not positive, IANADigiPhotgrapher).
This post is getting kinda long, but I wanna share this. I had this idea on a way to make a cheap, possibly portable, digital camera....well, not film camera at least. I'd take three photodiodes (diodes that block when there's light, and don't block when there's not), put the three primary color filters over them, have the light coming in through a slit, and hitting two mirrors, then going to the photodiodes. When you hit the button to take the photo, it rotates the first mirror horizontally, back and forth, as fast as possible, and the second mirror slowly scans down. The output from the photodiodes would directly going to a cassette tape. Later, I could read the cassette tape on my computer, and write a program to analyze it and extract the picture. I thought it was neat because the parts were cheap, but highly impractical. Especially considering it'd take about a second to take the picture with standard photodiodes (~25ns per reading, IIRC). Anything longer than 1/15th of a second *requires* a tripod...imagine the shaking going on with the motors as well.
Anyway, yea, happy learning and stuff.
A normal memory chip is actually light sensitive, in a nasty gray-scale sort of way. :) and write consistent values out (all ones or zeroes) then display what you read back in.
So, take an old memory chip, like a 1-meg or so. Carefully split the top off of it (might take a half-dozen tries to get one with pins still intact after).
The one I saw was plugged directly into a memory card. These days you'd probably have to rig up a parallel port interface.
Then all you do is put a lens over it for focus (watch out for the sun!
I know what I have learned back in grade school will apply to this project.
step 1. get shoe box
step 2. get needle
step 3. get charged coupled device CCD
step 4. make small hole in box
step 5. put CCD in box.
Step 6. Connect shoe box to PC
Step 6. aw crap, go to Circuit city.
...::----::...
I am in no way affiliated with this sig.
Take the linear CCD array from it, add some mechanics, and with some luck you can get nice (in the range of megapixels) images.
These guys did it already: here and here
Better than a webcam, and pretty good for understanding how digital imaging works.
A home-made webcam is so far different from the type of pinhole camera that was likely made in school that it's not even funny.
It's one thing to make a cardboard box which uses some fairly standard physics to project an image on a chemical-coated piece of paper which can then be processed. Everything is big, can be made and handled by hand, etc.
With a webcam, it's not exactly like you can whip up a CCD, various other ICs, the code to run it, etc. Almost none of this can be done by hand, and it requires a extremely high level of knowledge to do it all. In fact, it's very unlikely that any one person has ever possessed all of the knowledge to make such a device.
This is like saying "well, since it's not hard to make a simple steam driven piston type engine in a metal shop, why doesn't anyone piece together an electronically controlled fuel injection engine?
It's possible to achieve interesting things just by removing a webcam's built-in lens/filter assembly, and replacing them with lenses and filters from 35mm camera. See Lundycam for examples. You can build an extreme telephoto camera in this way for very little money.
You can also change the webcam's behaviour (improving low-light performance, for example) in software by using something like the Java Media Framework.
If your interest is purely academic, you might check out the CCD Camera Cookbook Webpage. The CCD Camera Cookbook is a book covering the design of two CCD cameras for Astrophotography. The resolution of these cameras is not high, and they do not come out being cheap. I am currently reading the book and will probably build the TC245 camera as a prelude to trying to design my own higher resolution CCD camera for Astrophotography. I think the book alone would be a good start in an attempt to understand CCDs.
At least I wanted to make the same comment... ;-) And remember 'Funny' does not give one karma points...
Paul B.
This site has quite a few links to people's NBTV projects and software: Narrow-bandwidth Television Association
Oxford Dictionaries Online
Everyone keeps talking about how you can't make your own CCD chip. I don't know that that's what the poster intends. There's a lot more to creating a web cam than simply the CCD chip. First of all, you can buy CCD chips from a number of sources. You'd then need the associated logic.
Actually, a number of Astronomy hobbiests are into doing just this sort of thing because astronomy quality cams are quite expensive. A number of people have used regular web cams for astronomical work, usually with long-exposure modifications to the cams, with a great deal of success.
A team of French hobbiests created this Genesis cam from scratch. It's very impressive and better quality than most of the hobbiest level cameras you can buy since it's based on a very high-resolution and very light-sensitive CCD chip.
But if you want to create just a basic web cam, there are much cheaper CCD chips. The datasheets will probably give you enough of an idea for how to get started with a project.
The CCD Camera Cookbook provides an excellent overview of the construction of an astronomical camera from scratch. Amateur Telescope Makers (ATMs) do this all the time to obtain high performance cameras with greater sensitivity and dynamic range than conventional webcams and Digital Cameras. Such designs not only incorporate superior ADCs but often have such features as peltier coolers.
I can still rememmber when the first reply to a problem involving hardware was, 'yes we can build it!' Now the bulk of supposed 'hackers' reply that you have to go out and buy whatever you need.
Lots of people are saying you can't build the chip yourself. That's not exactly true.
Go through CMP and you can get say the AMS C35B4C3, a 0.35um 4 metal, 2 poly CMOS process, for 650 Euro/mm2. I'm sure lots of people will cringe at the 0.35um, saying that it is ancient. Well, maybe in digital terms, but it is quite nice for analogue/mixed chips imo. 0.8um is still around (290 euro/mm2)!
Alternatively, if you are part of an Educational Institution or Research Laboratory, how about the ST Microlelectronics 0.18um CMOS process for 990 euro/mm2?
Now get hold of a copy of Electric some spice or other and learn how to design design electronic circuits. geda may also be of interest.
That last step might take a while.
Design your chip, submit it to CMP, wait three or four months and you'll get it back. Now go on to do what the other comments are talking about with pin hole cameras etc.
Let's do a rough price breakdown. Suppose you want VGA (640x480) in grey scale. Let's also suppose you can get your pixel element down to 5um*5um (which would be quite small imho). This gives:
Width: 640*5um + 2*400um = 4mm
Height: 480*5um + 2*400um = 3.2mm
The 400um gaps are for the pads on each side. This doesn't include any other electronics, so let's just say it is 4mm*4mm = 16mm2.
You need packaging as well and are probably limited to JLCC packages because it needs to be exposed to the light. Let's assume a JLCC68 package. You get 20 chips back and each package costs 48 euro.
So, 16*650 + 20*48 = 11360 euro. Put another way, 568 euro per chip. Don't forget to add VAT if you pay it. For the UK, this means 9343 or 476/chip.
Now consider that 16mm2 is still a small chip (and colour would be at least 3 times larger). If you have access to a webcam and can get inside it to look at the light sensitive area, measure it and figure out how much it would cost!
Cheers,
Roger
Do you have any better hostages?
The sensor array will be a bit tedious to construct (especially if you want more than a trivial number of pixels), the response time may be slow (ISTR some photoresistors haveing recovery times in the multi-second range), and you may need to spend some real cash for peripheral equipment (if you are going to build the thing using a microcontroller, you will need something to burn the MCU's program with, which will run you at least US$100). On the upside, you can build a true greyscale device (if you use a ADC to sample the pixel photodiode/photoresistor pixels).
The resulting camera will be bulky, slow, and have absolutely terrible resolution (we're talking 1 Kpixel, tops), but, if you have a spare month or two, it sounds like a fun project.
1. Get 100, equal webcams
2. Put in an array of 10x10
3. Feed input to several computers (USB)
4. Apply respective parallel image processing, including mosaic techniques that will get rid of overlap
5. Feed resulting data to a single computer
6. Play with the resulting ~30 Mpixel image.
Don't forget to point it to the sky. You may arrange things so you have complete sky coverage, then track aircraft and meteors. Adjust software accordingly.