Public Lab Research note


IR/Visible light camera dreams & parts

by mathew | March 12, 2013 02:31 12 Mar 02:31 | #6294 | #6294

I've been trying to envision what an Visible Light/IR camera would look like as a product. the sketch above is of a weatherproof tupperware-like case with a battery holder and a means of wiring external sensors.

Cases

I can't find any outdoor-specific cases for the Pi, and none I'd like to mount a camera to. thats why I envisioned my own. The bottom part is easy, as its made out of acrylic and could be glued up by hand. the top part I envision as polypropylene, like tupperware, because then it could have hinges. A part like that would need to be molded. For now I'll stick to tupperware, the cases on the raspberry pi site weren't intriguing enough.

Camera strategies:

Camera Pairs

Currently, Public Lab IR/Visible light cameras are created with a pair of cameras.

Advantages This is awesome for photographing things that are at enough of a distance to ignore the different lens positions, but not all that great up close. Its also as fast as the lens, and webcams are cheap. Can do video.

Disadvantages Aligning two cameras while prototyping can be tough. Up close objects like bugs and animals--which are great to look at in Near IR-- are hard to photograph with two lenses. keeping the settings and focus of two cameras in synch can be hard.

Raspberry Pi foundation has demoed two of their upcoming camera boards running in 720p video

Single Camera with filter switch

I'm going to pursue a different strategy than camera pairs, using a solenoid-driven filter switcher like those found in security cameras. Here is one for $50, half way down the page. I'm seeking a cheaper one, they're available in quantity for $2 or less The term I've found to search is "Dual-filter swing solenoid."

Advantages potentially cheaper than using two cameras, no alignment problems, can take close-up images. Will make a genuine mechanical clicking sound when used.

Disadvantages less than half as fast- has to take two images every time. Unlikely to be able to do video. Introduces another circuit into the control system.

Raspberry Pi as a camera

The Rasberry Pi is a powerful board for the price, but it leaves several things to be desired as a camera platform.

The Pi has no clock-- it therefore can't keep track of time when turned off, which is a pain because photos can end up improperly time-stamped. Without a clock the Pi also can't work for time lapse in the field without a really, really hefty battery.

the Pi is particularly finicky about power and needs a steady power supply. A regulator is needed to attach a battery. lack of a regulated power supply can make attaching sensors difficult.

The Pi doesn't have any power protection on its GPIO, so plugging in sensors is potentially dangerous.

There are a few extension boards to deal with these problems. 2-Watt Elements Raspy Juice is promising because it combines power regulation and a real-time clock with an ATmega 168 (like the older arduinos), solving all these issues. But its $42 for some otherwise quite cheap components.

here are two other RTCs.

The Slice of Pi is the cheapest ad-on for extended I/O capabilities but doesn't add power management or an RTC, which are what we really need for a camera.

A full list of RaspPi expansion boards is here.


10 Comments

a couple of ideas; you could use a beam splitter to combine two cameras. good ones are expensive but a cheap acrylic one might be good enough. You might also consider using a piece of nitinol (muscle) wire to move a filter. That would probably be my first choice. If you can come up with a lever to minimize travel that would make it easy to do it in the time required.

Reply to this comment...


i've been trying to find cheap beamsplitters to try this... any suggestions or URLs?

Is this a question? Click here to post it to the Questions page.

Reply to this comment...


i will take a look. what is the lens diameter? if you have not had the chance to play with muscle wire it is really cool stuff. I would be happy to send you a piece.

After giving it a bit more thought, have you considered a 50% mirror at 45 and two cameras at 90? That would give you the instantaneous exposure you are after with no intervention required.

I'm not sure about the IR transmittance of the cheap beam splitter but we can check it out. It would probably be cheaper to buy two prisms and glue them together.

Is this a question? Click here to post it to the Questions page.

Reply to this comment...


Well, one reason to beamsplit and recombine on one camera is to save the cost of a camera... it'd be really awesome to have an entire camera for well under $100 retail, which means parts have to come in well under $50 (or nobody will distribute it). So although the 50% mirror is a good idea, if we could just add one more mirror, we can halve the cost of webcams.

50% mirrors, well placed, could displace a beamsplitter, but at what cost? Someone suggested 50% metallized mylar film, which is a good idea, but harder to construct. I guess we'd like to do a proof of concept with a real beamsplitter before trying to DIY the beamsplitter itself... one step at a time! :-)

Is this a question? Click here to post it to the Questions page.

Reply to this comment...


$20 too much for the splitter? This actually looks like pretty good quality and they have quite a few, maybe they would come down in quantity. Lots cheaper than Edmund's for sure.

http://www.ebay.com/itm/beam-splitter-cube-prism-/250783311279?pt=LH_DefaultDomain_0&hash=item3a63d9a5af

As a small token of my appreciation for your obvious commitment to PLOTS and as a sign of mine I just bought you one for proof of concept. When it gets here in several weeks I will let you know and arrange to send it to you.

You might also consider teleprompter glass. A 1' square piece of 60/40 beamsplitter glass should cost less than $100 and if we can cut it up successfully it could yield over 100 pieces. Wouldn't that be cool.

Is this a question? Click here to post it to the Questions page.

Reply to this comment...


Huh, that is not too bad! Wow, and thank you so much!

That would be amazing... $1 each... eager to hear how the ebay one looks... was it acrylic or glass?

Is this a question? Click here to post it to the Questions page.

Reply to this comment...


Also it seems that the seller would surely do a bulk deal... they seem to be the manufacturer...

Reply to this comment...


As far as I can tell they are glass. The cheapest I could find one onshore was at Edmund Optics or Newport for about $150.

it appears the guy has not sold any for over a year, so he might sell them for a lot less.

I'll check with some of my old contacts and see if i can get somebody to donate a piece of teleprompter glass. I fear it will be tempered for safety, making cutting it up very difficult, but I am not sure about that, nor am I sure the optical quality is adequate. We will see in time.

Reply to this comment...


it had not occurred to me before that with the beam splitter, a mirror, and a pair of cheap electronically synchronized shutters you could get full rate video with alternating frames of IR/VIS that could be reconstructed to display some pretty amazing data. You had probably considered that in your dream.

Reply to this comment...


your filter switcher looks even less expensive than using a CD player. nice idea. The spinning wheel would permit faster filter switching of multiple filters, but not for $2. Your solution would also use less power and be lighter.

BTW there is something wrong with the Chinese beamsplitter. Stay away from it.

Reply to this comment...


Login to comment.