Hacker News new | ask | show | jobs
DIY thermal vision shoots 360° panoramas (kurokesu.com)
89 points by SauliusLukse 3351 days ago
11 comments

An alternative is the 8x8 Grid-EYE from Panasonic, costs around £30. The Melexis equivalent is a 16x4 which costs about the same. Note you'll need to fill in a fairly lengthy export/release form to get hold of the Grid-EYE sensors. I haven't had any problems ordering them from Farnell, just tell them what your project is.

Both are fairly easy to talk to over I2C. You can make a passable wireless (low-res) thermal camera using something like a Particle Photon.

The MLX90640 isn't out until September unfortunately.

NB: Actually Farnell have the Grid-EYE for only £16, but the full range ( -20 to 100) model is backordered. You can still get the 0 to 80 version easily.

I really like Melexis uses SO-19 packaging instead of the Panasonic Grid-eye SMD.

Here are the octopart links for both:

AMG8851 ( https://octopart.com/amg8851-panasonic-23284568 )

MLX90621 ( https://octopart.com/mlx90621esf-bad-000-sp-melexis-52364204 )

Or Flir Lepton. A bit more pricey (~$200), but image quality (wrt. noise) and resolution (80x60) are much higher.
I used the Melexis MLX90260/MLX90261 back in December 2014 for some sensors research at work, and they were not very helpful and the documentation was appallingly ambiguous for important parts of the programming; plus, it offloads all the mathematics (not insignificant amounts of floating point and statistics) to your processor (not a good deal for embedded work). The output, once processed, though, was very accurate and provided for additional processing to make more use of the information than similar offerings.

The Panasonic Grid-Eye offering was much easier to use, and did not require additional processing power just to make use of the information.

I've used their single point IR sensor before, just had a look at the documentation for the 16x4 - holy hell that's a lot of steps just to get a calibrated sensor reading! Do they provide a C library for it, or are you expected to write and test your own?
MLX90640 would be nice, but only for speed improvements. I would like to add additional lenses to improve resolution, but they are not cheap.
Nice project! This reminds me of a beautiful series of blog posts of a guy who built a single pixel pinhole thermal camera:

First post: http://orbides.org/photobot.php?lng=eng&mode=making1

Sample image: http://orbides.org/photobot.php?lng=eng&mode=pic

Awesome project! Have to pimp it :)
HN hug of death strikes again.

Cached version:

http://webcache.googleusercontent.com/search?q=cache:VQlpQWh...

I already contacted the site owner. He is doing what he can right now ;-)
Stress test! Seems like my web page needs some optimizations :)
I built a similar, although much less sophisticated prototype at a hackathon a few months ago: https://github.com/fauria/thermoman-360

In case someone finds it useful, I published a Seeed Studio recipe: http://www.seeed.cc/Grove-MLX90614-p-1157.html and adapted an Arduino library: https://github.com/fauria/bildr-mlx90614 for the MLX90614.

Did Endesa / Opinno provided you with the hardware? Congrats for the prize, btw !
Thank you very much! No, I brought my own stuff, they provided the space, food, etc but no HW.
When I read IR in the post, I thought, "well that's not thermal, is it?" and the thought was incorrect. FLIR's page [1] has a more understandable to me summary than StackExchange [2]. Also interesting is how lensing works with radiation in the IR spectrum [3].

Thermal cameras cannot use regular glass lenses, as glass will reflect thermal radiation rather than allowing the radiation to pass through the lenses. Commonly used materials for thermal lenses are Germanium (Ge), Chalcogenide glass, Zinc Selenide (ZnSe) and Zinc Sulfide (ZnS).

1. http://www.flir.com/cvs/americas/en/view/?id=30052 2. https://physics.stackexchange.com/questions/6869/what-is-the... 3. https://www.axis.com/fi/en/learning/web-articles/technical-g...

This guy did a hackier- and probably even cheaper - DIY camera a few month ago http://www.niklasroy.com/project/195/DIY_thermal_imaging
This is really cool, given how expensive thermal cams can be. The main sensor used by the author (Melexis MLX90614) costs only ~14$ at my local retailer, you can probably get it a bit cheaper if you order online.
Flir and Seek have pretty good usb thermal cameras that attach to your smartphone for about $250 if you want something ready made
"Flir and Seek" sounds like the modern high-tech version of "Hide and Seek"... played using thermal cameras, of course.
ready or not, here I scan
Not as much DIY, but for $260, Sparkfun has a nice looking 80 x 60 FLiR thermal imager: https://www.sparkfun.com/products/13233
Good camera candidate to build FAST thermal panorama rig!
Holy moly! Two axis stepper motor controller for $67?! (You should check out the store connected with that site.) This reminds me I really need to finish my single-scanline CdS cell camera project...
Yeah it's a puny Arduino based one though. Look at the heat sinks. Don't waste your money - unless you really need the space the best ones are the ones that look like this:

http://www.leadshine.com/series.aspx?type=products&category=...

100 times better and they're not even any more expensive - $31 on aliexpress.

Well, Leadshine is mostly acceptable (but not Chinese fake!).

Please sum all components (custom PCB, Trinamic controllers, arduino, connectors, enclosure, shipping components, assembling,...) add some taxes and you will get almost the same price.

Wouldn't something like this be sufficient in many cases? https://www.pololu.com/product/1182
It would depend on the amperage of the steppers involved, but sure - though I'm partial to the DRV8825 - it's considered a better driver in the 3d printer realm.
Only for the tiniest of steppers. So basically no.

In my experience you can judge a stepper driver by the size of its heat sink.

I always wonder: how do you find the parts you need to build such things, and how do you make sure they fit together without spending enormous amounts of time in various online parts shops?
In my experience, either you get exposed to various parts at your workplace/college, or you stand on shoulders of similar projects.
3 ways.

1. Already know enough to glance at some spec sheets and build the rest in your head. Aka, lots of experince.

2. Execute a known good build configuration.

3. Combo of the 1st and 2nd points to suit your ability. Ask people with more knowledge for guidance. Maybe rebuild it once or twice.

Experience comes with practice :)
I always wanted to try to do this with RF wavelengths like 2.4 GHz.
I'd forgotten about this project, really loved this!!
Can be done, just highly directional antennas are bulky and setup needs to be much bigger.