Thermal Imaging Gallery II

[bap2703]

The lens has been disassembled

Finally you made it 

[Ultrapurple]

Seek Reveal Pro FF.

1st image - a cold day, accidentally getting the sun in the image - notice the wiggly 'comet tail'

2nd, 3rd and 4th image - the tail remained on the sensor for a few moments, long enough to take these shots, but after a calibration flag event it was nulled out and the image was fine.

5th image onwards - selection of random images from the Seek. All 320x240, exactly as-shot.

[DaJMasta]

Not especially great quality images, but some neat ones I got a couple weeks back after a few inches of snow.  Very funny to see a sort of underlighting effect where the snow on top was dark and the bottom of things were warm.  An odd perspective in the tree branches, and free underbody lighting for all the cars parked on the road.

[globoy]

New poster here.  Psyched to find this forum.

This is image from a home-made (Lepton3.5) camera.  Black fox on rear deck with a few inches of snow on a cold day.

[Ultrapurple]

A couple of deer seen with a Therm-App Pro and 35mm lens, ThermViewer software superresolution to 1280x960. Image straight off the camera, no processing. Notice how you can see the warmth of their hoof prints.



Bonus: the second file is an animated GIF of three deer, made with the Therm-App Pro - but you'll have to download it, as the forum doesn't seem to support animated GIFs.

[Ultrapurple]

And the same deer - plus a friend - seen using a FLIR T4325ZS 320x240 PTZ security camera (standard 13mm lens).

[Ultrapurple]

If anyone fancies reprocessing the two-deer image, here's the raw image data. I'd love to see what can be done with it.

[dizzyzzid]

If anyone fancies reprocessing the two-deer image, here's the raw image data. I'd love to see what can be done with it.

Done with python and opencv2 and very simple sharpening filter.
You gain a bit of detail on the deer but the background gets a little noisy.
Which looks promising for me as I don't care about the temperature accuracy at all, spotting purposes only.
Run time is around:
818 µs ± 50.7 µs per loop (mean ± std. dev. of 7 runs, 1000 loops each)

Can I ask what the distance was between the camera to deer?

[Ultrapurple]

Thanks dizzyzzid - you've certainly brought up a lot of detail in the deer's bodies.

The distance to the deer was very roughly 30ft (10m). I was watching them from an upstairs window.

[JimM]

Here is Ultrpurple's deer image after conversion to binary data and saved as a PGM P5 file using Objective-C code. Then simply contrast-stretch by 1%x1% in ImageMagick and save as PNG. Note the image is reversed (mirrored) from the original image posted by Ultrapurple.

Edit to add: Changed contrast-stretch to 2%x0% to minimize blowout at top end. Also flipped image to correct orientation.

[frenky]

Took image from dizzyzzid upscaled it with http://waifu2x.udp.jp/ and cropped.
Soon we will be able to see a wolf in deer eyes reflection.

[Ultrapurple]

This is getting to be fun.

I took frenky's image, did a tiny bit of gamma modification and quite a lot of area-selective noise reduction on the background, then flipped the image back the right way round.



If you look carefully at that reflection in the deer's eye you'll see it's not actually a wolf: it's a farm cat who self-identifies as a wolf.

[JimM]

I know this topic is on pictures rather radiometric data., however does anyone know how to get temperatures from the raw data that Ultrapurple posted for the Therm-App images?

[frenky]

He posted file with matrix of temperatures. Image is 640x480 so his file contains 640 columns and 480 lines of data.
Each value is temperature of single pixel. First pixel has value 1944 which is probably 19.44*C.

[JimM]

frenky, Ok that kind of makes sense, the minimum value in the file is 1832 and maximum is 3303, so that's a minimum temp of 18.32 deg C and maximum 33.03 deg C. Wonder why they didn't just use floating point values. Would have been a smaller file.

Edit to add: Those temperatures may not actually be the target temperatures. See this post from Stackoverflow:

https://stackoverflow.com/questions/56695676/how-to-open-a-txt-file-of-ir-temperatures-as-an-image-in-matlab-or-other-analysi and specifically the following:

"edit, since writing this I've found that the output .txt file of temperatures is actually sensor temperatures which need to be corrected for emissivity and background temperature to obtain the target temperatures. (One way to do this is to use the software which comes free with the camera to create new output .csv files of temperatures and use those instead)."

[Ultrapurple]

Perhaps this should move to a new thread.

It is worth noting that (at least in the case of the Therm-App series) the 'temperature' of the pixels depends on the lens used. This particular Therm-App Pro was purchased with that specific 35mm lens and, as it's a very early unit, I suspect Opgal may have taken a little more care over calibration than may normally be the case.

There is no overriding reason that the pixel data should be an accurate representation of the actual scene temperature, though it may be a happy accident that it works out that way. I'll have to make an image of something like ice-water beside boiling water so we can see what happens. And if I do it with different lenses we can maybe learn more.

New thread started: Deer photos and Therm-App Pro data. Let's leave this thread for pretty pictures, as was originally intended.

[IwuzBornanerd]

Okay, back to making pretty pictures. 

While manipulating the data to make details visible is certainly useful, I like to start by getting all the thermal details with a distinct color palette.  While in some cases this produces some cool isotherms I recognize that in many cases it makes a rather "busy" image as well as noisy.  I do like the "grass" look here, though, even if it is noisy.


Someone who prefers gray scale only might instead apply the entire gray scale to the background & then do the same to the deer & then combine the 2 to get the following.  Not much more detail in the deer, but noticeably more in the background.


In similar fashion, taking the green from my distinct color palette & applying Seek's "Amber" palette to the deer yields brown deer on green grass (which I think looks pretty cool). 


As far as simply applying a single palette, I think my favorite is the standard "iron" palette:


And applying Seek's "HiLo" palette gives us deer on snow. 

[Ultrapurple]

Thanks IwuzBornanerd. I too really like the brown deer on the green grass.

<irony>
If only someone could invent some kind of device that would record the scene in front of the user and then render it in natural lightness and colour...
</irony>

[Ultrapurple]

Therm-App Pro (640×480 with superresolution to 1280x960), 35mm and 19mm lenses.

[Ultrapurple]

Just a few more from the Therm-App Pro, all straight off the camera and unedited.

[Ultrapurple]

I thought I'd busted a toe. It hurt like a really hurty thing. The bruise was obvious and it was warm in thermal imaging but the X-Ray didn't find a fracture. It settled down after a few days.

[Ultrapurple]

An old thermal imaging favourite - a really hot cup of tea.

Therm-App Pro (640x480) with superresolution to 1280x960. Lens probably 19mm.

[Ultrapurple]

I bought a tiny 5.6GHz FPV video transmitter via eBay. It seemed to be running a bit hot so I added a heatsink over the large synthesiser and smaller power amplifier chips. It was still a little on the warm side so I decided to frighten it with a thermal camera.

If the board has bits running at ~120°C with a heatsink, I can only wonder at how hot it was getting without the additional dissipating surface. Who designs these things? Or are the thermals now considered OK as long as the solder doesn't quite melt...?

Thermal image made with FLIR SC-660, standard 40mm lens plus Agema Thermovision close-up lens "20°LW" p/n 193 672.

Postscript: I added a rather larger heatsink (15x15x10mm) plus a 10x10x5mm heatsink on the other side of the board (which is also populated but with control rather than RF electronics) and the thing still runs at well over 80°C.

[downtodo]

SSD control chip 12x12 mm

HT-301

[Ultrapurple]

That last image - the ruler - is intriguing; it's interesting to see how the engraved marks reflect heat from different parts as the angle to the camera changes.