Yet another cheap thermal imager incoming.. Seek Thermal

[jeffreythe00]

If anyone is looking for a solution to cleaning the optics of their thermal camera or any other optics for that matter. There is a company that makes a FANTASTIC cleaning solution. I use it myself to clean laser lenses and camera lenses and have not had any issues with it. Its called First Contact and it is sold here. http://www.photoniccleaning.com

This solution is delicate enough to clean laser dies which are only a few µm in size. I can confirm this.

[termal lee]

nice info there, jeffrey, but this is for "dirty amateurs of seek", i will go with Fraser's IPA+soft cloth.
nonetheles, i will remember to use ur option when i am into high powered laser devilce, which i am going to get into, pretty soon 

[jeffreythe00]

No problem termal lee!

On the note of high powered laser devices, another good solution to getting your optics dust free is to use cotton swabs (not qtips because the glue can/will deposit on your optics. Not good) and lab grade methanol.

99.9% Methanol is much cheaper than First Contact and works very well but it is hydrophobic so not ideal if you store it for long periods of time.

[termal lee]

verry well noted. thanks jeffrey 

[lod]

Just a note to recent readers that I have summarised much of this thread in a publicly editable wiki
https://github.com/lod/seek-thermal-documentation/wiki

I am still routinely reading this thread for updates in addition to direct contributions.

[tomas123]

Thank you for your wiki.

I know something about the Flir sensor values and I'm reading here by the way that seek make a linear interpolation.
https://github.com/lod/seek-thermal-documentation/wiki/Temperature-Values

kelvin(x) = -18.763 + 0.0378683*x Kelvin

Flir has a nice document:
http://flir.custhelp.com/ci/fattach/get/1667/
on page 2 they say, that the response of "A/D Counts" are linear to Radiance [Joules/m^3/sec/steradian] with an offset.

This make also sense for Seeks microbolometer.
I know, that the function temperature=f(radiance) is similar to linear  in our small temperature range but not exact.

I wrote here some details
http://u88.n24.queensu.ca/exiftool/forum/index.php/topic,4898.msg23944.html#msg23944

The Radiance formula you find in Planck's Law
http://en.wikipedia.org/wiki/Planck%27s_law

e^(h*c/(k*L*T)-1 = (2*h*c^2/L^5)/I
Planck’s Constant h = 6.626068x10-34 joule sec 
Boltzman’s Constant k =1.38066x10-23 joule deg-1
Speed of light in vacuum c =2.997925x10+8 m/s
T = object temperature in Kelvins
L = wavelength in m
I = radiance in Joules/m^3/sec/steradian
e Euler Number

solve it to Temperature formula
T= B / ln(R/I +1)
with ln => Natural logarithm
by substitute with
R = 2*h*c^2/L^5 
B = h*c/(k*L)

as explained above is Radiance "I" linear to  Flir A/D Counts "S"
we can write
I = R2*(S+O)
with
S = 16 Bit A/D Counts
R2 = constant factor
O = offset

and with R = R1/R2
T= B / ln(R/I +1)
goes to
T = B / ln(R1/(R2(S+O))+1)

[tomas123]

kelvin(x) = -18.763 + 0.0378683*x Kelvin

that the response of "A/D Counts" are linear to Radiance  with an offset.

one day later and no reaction 

for better understanding I made a excel sheet

the Flir E4 has a spectral range from 7.5–13 um (the peak 10.6 um) http://support.flir.com/DsDownload/Assets/63901-0101_en_50.pdf
 
I wrote a table for the planck law at 10600 nm wave length over the temperature range
There you can see, that the spectral radiance in [W / m^2] (per nm wave length) is not exact linear



source: http://www.ces.karlsruhe.de/culm/culm/culm2/th_physik/simulation/planck.xls

[frenky]

Sorry for late response but when I saw your first post I was like.... emmm ok....  And the second post didn't help a lot.

If I understand you correctly you have found a formula "kelvin(x) = -18.763 + 0.0378683*x Kelvin" that converts seeks raw pixel data into temperature?

My big dilemma now is: Should I put another n*10 hours of development into custom SeekThemal software, to make it better or just sell it and forget about it.
(And buy I3-ThemalExpert when it becomes available in a few days/weeks...)

I took a shot at custom software a few months ago with stepper motors and super-resolution, but that was too hardware dependant and to clumsy to use when I wanted to make a thermal image of a pcb.

If I decide to try again I'll go from the scratch:
- direct seeks module at thermal plane with the same temperature
- record a value of each pixel at 20*C, 50*C and 100*C after 1minute, 5minutes, 10minutes, 1hour
- map all bad pixels (dead and "patent" pixels and calculate best values)
- just ignore frames of data taken with shutter closed
- perhaps attach a small Peltier element on the back of the seek to prevent huge thermal drift?

That should be enough to make a software that would produce really clean 206x156 images...

[tomas123]

If I understand you correctly you have found a formula "kelvin(x) = -18.763 + 0.0378683*x Kelvin" that converts seeks raw pixel data into temperature?

No, this is not my idea.

After reading this page
https://github.com/lod/seek-thermal-documentation/wiki/Temperature-Values

kelvin(x) = -18.763 + 0.0378683*x Kelvin

... I thought, that seek users make a linear interpolation between this poor reference points.

It's possible, that seek get A/D values from microbolometer like Flir, where RAW=functions(Radiance).
Only a suggest...
To illustrate this non linear relation, I wrote the Excel sheet.

I'm not a seek user. I only poked about in this forum.

[brobbuilder]

Someone should hack together a Lytro-style superdepth camera with Seek or Lepton modules. I'm looking at you, mikeselectricstuff...

[encryptededdy]

Someone should hack together a Lytro-style superdepth camera with Seek or Lepton modules. I'm looking at you, mikeselectricstuff...

To do that wouldn't you need a camera with a relatively shallow depth of field?

Neither the Seek nor Lepton fit that bill due to small sensors and high f/stop lenses.

[encryptededdy]

http://inteliscopes.com/thermal-2/

Hmm... a questionable application for the Seek XR.

I wonder how well the Seek will work as a rifle scope considering it's low frame rate, constant shutter actuations and questionable sensitivity.

Especially after looking at Mike's teardown, I can't see that shutter assembly holding up to recoil very well.

[Uho]

Pig so will look on the screen from a distance of 8m. Where are they going to hunt? On a farm in California?

[arsenix]

Somewhere in the fine print: "For hunting domestic pigs only from distances of no more than 10 feet."

Great CAD renderings though...

[frenky]

I decided to write a new version of Win software for Seek.

First step, know your enemy. 

I'll take a raw reading of all pixels in 5sec intervals for 1 hour while pointing to uniform temp plate. This will show me the how big error each pixel has and thermal drift of each pixel though time.

I'm currently using a cork plate for uniform temperature reference (at room temp). Do any of you have a better suggestion?


Next step will be to make some kind of heating plate to take measurements of pixel values at 100*C.
What would be best approach for this. Perhaps a heat bed of 3d printer? Big peltier element?
I need something that will produce really uniform temperature plane.

Tnx for all suggestions.

[frenky]

Drifting of 8 random pixels in 1 hour. On some pixels temperature readings go up and on some readings go down.
As you can see thermal pixels need at least 20min to output stable readings.



Explanation of values:
Dark red line is a pixel which starts with value 7537 and ends with 6818. That is 719 less.

[encryptededdy]

Drifting of 8 random pixels in 1 hour. On some pixels temperature readings go up and on some readings go down.
As you can see thermal pixels need at least 20min to output stable readings.



Explanation of values:
Dark red line is a pixel which starts with value 7537 and ends with 6818. That is 719 less.

I guess that explains the excessive shutter actuations that we see on the Seek.

[frenky]

There is more weird stuff going on...

Some pixels shift raw values after each shutter calibration. So pixel value is fluctuating like this: 8500,6700,8500,6700...

[encryptededdy]

Seek posted this on their Facebook page. It appears to be from the Reveal.

[frenky]

It's not bad quality and ratio is correct. But after fake marketing images fiasco I don't trust them any more...
It could be that they are using special "lab" version of Reveal which produces better images than production version.

[Fraser]

The marketing video of the Reveal is appallingly misleading to potential buyers. This latest image is more like I would expect. Plenty of noise evident.

[frenky]

Live graph of pixels raw values.
When pointing at 19*C uniform temp plate pixels have values from 6000 (up slope of the chart) to 9000 (down slope).
Middle of the hill is at 7500 so I guess that 7500=19*C.

[daves]

Seek is piece of crap - imho.

[encryptededdy]

Seek is piece of crap - imho.

Piece of crap when compared to their advertising images? Yes.
Piece of crap when compared to other 200x150 thermal cameras? Yes.

But Piece of crap compared to other imagers in the same price bracket? I don't think the Seek is too bad in that case.

[termal lee]

i am with encryptededdy, seek is best of its price range.
if we compare flir one to seek xr, u will find that seek pwns flir one in all aspects. plus it is even cheaper.

thermApp is the best for value, if u need a better thermal cam. it is the perfect solution. but hey, if u only need a thermal cam for general purpose with low price, seek is the best choice.