EEVblog #668 - Mailbag + FLIR TG165

[mikeselectricstuff]

I think when Flir say automatic shutter they may be referring to the Lepton's automatic nonuniform correction, which relies on the fact that as a handheld unit, the scene is moving, so the long-term avarage of local pixels will be the same, and can be used to do nonuniformity correction.
I don't see a reason why it would need a shutter with the thermopile there as well, especailly as AFIR there was no indication of the min/max temps of the visible temp scale, so the Lepton is used for sighting only.

There's only one way to be sure - take it apaaaaart!

[Unixon]

It would be much nicer if they used different color lasers for guide beams.

[bktemp]

I think when Flir say automatic shutter they may be referring to the Lepton's automatic nonuniform correction,

Have a look at the comparison:
http://www.eevblog.com/files/FLIR-TG165/TG165_Comparison_FINAL.pdf
TG165: Automatic Shutter (calibration)
FLIR ONE: Manual Shutter

+1 for the teardown!

[max_torque]

I'd love to see that core memory bank side by side with a de-capped 16kB eeprom or similar!  Just to show how technology and manufacturing processes have moved on in 40 years!

[Monittosan]

Flir One resolution 80x60 px.
New boy on the block at half the price, resolution 206x156 px.

Looks like the market is hotting up.

Further details here.

how longs that micro usb going to last? 

[R_Gtx]

Flir One resolution 80x60 px.
New boy on the block at half the price, resolution 206x156 px.

Looks like the market is hotting up.

Further details here.

how longs that micro usb going to last? 

Does it really matter, how long that micro usb is going to last? The pertinent fact is that Freescale are now manufacturing a cheap thermal imaging sensor. How long before they release a dev. board? Think of the possible applications in security, robotics, etc.

[Fraser]

For anyone wondering what lives inside the FLUKE VT02, I did a teardown a while back.

It is very different (inferior) technology to that used in the TG165. A low resolution pyro-electric staring array, called the Redeye 6A captures the images and these are interpolated up to higher resolution. IIRC the Redeye 6A is a 31x31 pixel array that is used in a 15x15 pixel format in the VT02 and 30x30 format in the VT04. The Redeye 6A uses technology similar to the original BST arrays and requires a shutter (chopper) wheel to interrupt the image.

All becomes clear when you see the teardown that I did.

https://www.eevblog.com/forum/testgear/fluke-vt02-thermal-camera-visual-thermometer-teardown/msg467641/#msg467641

The VTxx series are FIXED FOCUS on both visible and thermal cameras, Depth of focus is similar to the E4. The wheel that looks like a focus control is just the lens cover that rotates to protect the lenses and releases the trigger lock..

The TG165 is a different beast BUT I have issues with the temperature being measured by a pretty much bog standard pyroelectric IR thermometer....very old school and a technique used in older fire fighting cameras, but they had a much smaller pyrometer FOV. The VT02 and VT04 read the radiometric data from the Redeye 6A so you measure what you see without parallax error issues and close-in working on PCB's is possible. The use of a single pixel pyrometer in the TG165 is fraught with measurement error risk as is the case with the conventional IR thermometer technology. With the FOV having a direct effect on the target area size with distance, it is essential to achieve 100% or greater illumination of the pyrometers view to avoid errors creeping in. This is why thermal cameras are so effective....they show you on the screen exactly which cluster of pixels are making the measurement. For me, the TG165 is a disappointment as surely the microbolometer could have provided adequate measurement accuracy ? Or maybe not. Is this the LEPTON cores Achilles heel ? Poor measurement accuracy 

As Dave stated, the TG165 is aimed at a specific market and is well designed for that. However I would expect many to buy the superior E4 instead ? The standard resolution is the same, MSX is included and the E4 is a superior camera in most (all?) respects.

As for the VT02 and VT04, I suspect continued sales will only be due to buyers wanting FLUKE branded equipment !  My two VT02 prototypes were bought only out of curiosity 

[SeanB]

I'd love to see that core memory bank side by side with a de-capped 16kB eeprom or similar!  Just to show how technology and manufacturing processes have moved on in 40 years!

Will these do, taken right now?

First is a DLR3416 LED matrix display, made week 48 of 1994, ex a B&O VCR in the process of being turned into scrap metal, and there is a lot of steel in this thing, a lot more than the Sony Umatic recorder I did recently, but it all is there as ballast weight.

Second is a Intel 8751 erasable processor, made 1980, in a ceramic package

Third is another i8751H-88, in a hi top cerdip package, also made in 1980.

Who was saying the 8051 is dead?

As to the FLIR I will guess that when it is torn apart you will find the boards are identical to the FLIR one Mike took apart. Now we know why they used that massive amount of overkill, they probably were developing this, and were either held up in the case or the display side for a while, and decided to make the iPhone version to get a feel of the market, using what is obviously a common core set.

Wonder how long it will take Mike to "enhance" this new core and improve it to higher resolution.

[eneuro]

As Dave stated, the TG165 is aimed at a specific market and is well designed for that. However I would expect many to buy the superior E4 instead ? The standard resolution is the same, MSX is included and the E4 is a superior camera in most (all?) respects.

When we realize what this "magic" MSX is indeed by watching this intro to this technology

in the terms of image processing tools needed to do this what they call "object details" which are in fact only object contours taken from classic visual camera and... aplied to thermal camera calibrated output with temperatures, than we start thinking why they claim any patent for this MSX, while using latest OpenCV boards with Nvidia CUDA support for high performance image processing in real time it is quite easy achieve similar results to this MSX, so forget about thisTG165 for $500 if someone do not need this laser temperature measurements which looks very tricky with those 2 laser points and consider quite cheap USB OTG interface Seek thermal $200 for Android OSes

with image processing and visualisation made by this very nice $300 The NVIDIA® SHIELD™ tablet 8" HD 1920x1080 screen and Android with CUDA support for OpenCV 

This kind of image processing MSX does with visual image to extract edges I was doing more than 10 years ago at university 
Nothing special to extract & add contures/edges to another image (thermal) and some kind of user interface (not open source so no chance to change stupid behaviour) to be able mark a few spots and display its temperature when we have two scene images: visual light camera image & thermal image.
Probably, they call it Multi-Spectral Dynamic Imaging (MSX™) while from marketing point of view the more complicated names are used in product description not easy to decode by normal people, customer might think he buys something special so this has to be very expensive. Additionally Flir use classic trick "pending patent" terms to make this product more atractive 
Yeah, not everyone can write such image processing software to achieve similar results without 2 diffrent types of cameras (visual & IR) in one device but it is not anything which is worth invention and patent claims 
Multi-Spectral Dynamic Imaging Technology from FLIR Systems

"Unlike traditional thermal fusion that inserts a thermal image into a visible light picture, FLIR’s new MSX technology embosses digital camera detail onto thermal video and stills. Using MSX technology on FLIR T450SC and T650SC IR cameras eliminates the need for a separate digital image in your reporting."

So, when we use additional digital image eg. from  built in 5M into The NVIDIA® SHIELD™ tablet, than we can create output image which will contain whatever we want from those two visual & IR images, so when Flir will tell me that I can not output something similar to this what they call MSX because of they patented it and are inventors, I will tell them    while I was capable of doing such things more than 10 years ago and now using OpenCv it can be done much easier and faster in realtime 
The only problem is that it might be difficult to find the same focus using 2 diffrent cameras, so it might be not such easy to make those two images from visual camera countours & thermal camera nice aligned, so it might require some calibration, but it looks like that this
The NVIDIA® SHIELD™ tablet should easy outperform any Flir MSX hardware while it can use CUDA accelerated OpenCV for real time image processing.
That is true it can be a little bit tricky to do but advantage is we can have open source version of Flir MSX and maybe even send those realtime theraml images with scene countours like MSX to another big monitor wireless while The NVIDIA® SHIELD™ tablet was created for gamers who often wants see their action on big TV screens 
I hope I will be able test The NVIDIA® SHIELD™ tablet  & Seek thermal under Android within a few months after finishich current projects.

Anyway finally Flir's MSX myth busted 

[Fraser]

@eneuro,

Myth busted ?

Whilst I understand your views on the MSX name and capability, IMHO, you are being a little unfair on FLIR. Historically thermal imaging has suffered from issues surrounding target context. That is to say, with the lower resolution images, the users found it difficult to identify the exact source of a 'hot spot' as the surrounding detail was inadequate. The initial solution used by the manufacturers was to add a visible light camera that supplied a detailed visible light image that could be merged with the thermal image using 'fusion' techniques.

The down side of image fusion is that the visible light image can confuse the thermal image for the user as colours overlap. This was partially resolved by making the visible light image of variable transparency and so more or less prominent on the thermal image. This is how thermal camera OEM's have approached this 'problem' until FLIR released their take on the idea.

FLIR realised that the important information in the visible light image is not the colours and fine details, but rather the edges and outlines that provide context to the whole thermal/visible light combination. It is a little unfair to criticise FLIR for doing something different to the other OEM's that benefits the user, and them wishing to protect 'MSX' as a thermal imaging enhancement. Whilst the technique may have been in use elsewhere, or known to you a decade ago, was it being used as an image context enhancement in a self contained thermal camera instrument ? FLIR did well to introduce this form of visual image combination with the thermal image. As you will see from others comments, it does what it claims, and very well.

Aurora

[EEVblog]

Out of curiosity, I'm wondering how often the TG165 will recalibrate, or if it will only do it once every time the unit is turned on, or perhaps even less often than that.

The image freezes for less than second occasionally. Certainly no periodicity to it, can go for minutes without doing it.
I don't know why it does it, because as I said, it does not take any measurement from the Lepton sensor.

[EEVblog]

so forget about thisTG165 for $500 if someone do not need this laser temperature measurements which looks very tricky with those 2 laser points and consider quite cheap USB OTG interface Seek thermal $200 for Android OSes

You are not comparing apples with apples!
The TG165 is designed for a specific industrial target market of users who want a rugged trivial to use spot measurement tool. The Flir ONE and the Seek thermal add-on are nothing of the sort.

[EEVblog]

Wonder how long it will take Mike to "enhance" this new core and improve it to higher resolution.

You can't. The Lepton sensor is an 80x60 array. There is no upgraded resolution model.  I greatly doubt any extra resolution is hidden unused inside.

[eneuro]

The TG165 is designed for a specific industrial target market of users who want a rugged trivial to use spot measurement tool.

But we can't see what we measure using TG165 or other thermal cameras without edges/contoures apllied 

Now lets compare output image we can easy get with lets call it whatever we like for example OSX (Open Source eXtreme) made just a while ago using... mentioned in my post above two images aka MSX:
visual camera:

after performing old shool gradient to detect edges available more than 10 years ago in basic image processing tools (I used GIMP in this example) and when we add it (not oryginal image but its edges like in Flir MSX) with 30% blend mask to edges while for simplicity it is easy to do in GIMP we get not so bad output thermal image with scene objects edges with our let scall it OSX , so we can easy recognize its basic shape and see what is hot and where ambient temperatures are - we can add whatever we want based on how good thermal image is

while in case of any other thermal device  without our OSX, so only thermal image based we get something like in FLIR TG165 where there is no way we have no idea what we are looking for while we need another oryginal digital image of this thing


Of course this is not real thermal image of this AC mosfets switch concept (it was made by hand in GIMP) but shows how easy is to get quite good thermal image with object edges marked in output image using archaic image processing methods, so that is why I called it Flir's MSX myth buster, while we get significant improvement in what we see, and still it is not blend of oryginal visual image and thermal, but... we simply added edges to another thermal image as one could do many years ago If had two images of the same scene.
We need in our OSX improove blending of edges gray image with thermal image to get not such strong edges, etc. but even without this this object thermal image with our just developed OSX technology looks not so bad as without OSX or Flir's MSX 

I do not want compare FLIR TG165 to any other thermal cameras, just wanted to myth bust this Flir's MSX in dead simple image processing example.
While main FLIR's profits came from military contracts, they have technology much better than this what they shows for sure, Flir's thermal cameras hardware is very good too and probbaly are much better than cheaper Seek thermal, but for some customers Flir TG165 can be simply disapointment while it lacks this simple edges overlay thermal image enhancements like MSX in higher Flir's models


 

[ConKbot]

I think when Flir say automatic shutter they may be referring to the Lepton's automatic nonuniform correction, which relies on the fact that as a handheld unit, the scene is moving, so the long-term avarage of local pixels will be the same, and can be used to do nonuniformity correction.
I don't see a reason why it would need a shutter with the thermopile there as well, especailly as AFIR there was no indication of the min/max temps of the visible temp scale, so the Lepton is used for sighting only.

There's only one way to be sure - take it apaaaaart!

Wonder if FLIR is going to be shipping you , or if they were less than impressed by your shenanigens with the flirone 

[EEVblog]

But we can't see what we measure using TG165 or other thermal cameras without edges/contoures apllied 

Most of the industry has survived just fine without that for decades.
If you need it, buy that capability.

[EEVblog]

I don't see a reason why it would need a shutter with the thermopile there as well, especailly as AFIR there was no indication of the min/max temps of the visible temp scale, so the Lepton is used for sighting only.

Correct. Yet some people still don't quite seem to get this?
It is very deliberately used as a visual aid only to an existing spot thermal camera design.
This product doesn't give a rats bum about what the actual temperature from the Lepton sensor is, it just needs that thermal difference mapping.
Flir would be crazy to build any form of calibration shutter into it, unless they planned on having another model with that feature at a later date. But that would be encroach upon the E series.

[nitro2k01]

Correct. Yet some people still don't quite seem to get this?
It is very deliberately used as a visual aid only to an existing spot thermal camera design.
This product doesn't give a rats bum about what the actual temperature from the Lepton sensor is, it just needs that thermal difference mapping.
Flir would be crazy to build any form of calibration shutter into it, unless they planned on having another model with that feature at a later date. But that would be encroach upon the E series.

You know, in principle I would agree, however for the time being I'm going with evidence to the contrary (image hiccups and mentions of a calibration shutter in the marketing material.) Perhaps written material is misleading and the firmware has a bug that causes periodical hiccups in the image. I might be wrong. Who knows. All the more reason to do a teardown as soon as possible.
Or hold it up to your ear and listen while turning it on. Not sure if these shutters make a noise that is actually loud enough to be audible, but you could use the E8 as a reference.

[Skimask]

On the eRIC module...  I had the privilege of meeting and hanging out with Mike Meakin (CTO, LPRS) a number of times in Fairford back in '03 or so during an extended trip.  I needed a 10Mhz resonator for my Warp-13a to upgrade it.  Surfed around, found his name and found him right down the street from RAF Fairford.  Lost touch.  Amazing where people pop up now and again...

[kxenos]

I burst into tears. Nice deed Dave, bravo!

[EEVblog]

I needed a 10Mhz resonator for my Warp-13a to upgrade it.

I still have mine!

[EEVblog]

You know, in principle I would agree, however for the time being I'm going with evidence to the contrary (image hiccups and mentions of a calibration shutter in the marketing material.) Perhaps written material is misleading and the firmware has a bug that causes periodical hiccups in the image. I might be wrong. Who knows.

I know for sure

[mikeselectricstuff]

I think when Flir say automatic shutter they may be referring to the Lepton's automatic nonuniform correction, which relies on the fact that as a handheld unit, the scene is moving, so the long-term avarage of local pixels will be the same, and can be used to do nonuniformity correction.
I don't see a reason why it would need a shutter with the thermopile there as well, especailly as AFIR there was no indication of the min/max temps of the visible temp scale, so the Lepton is used for sighting only.

There's only one way to be sure - take it apaaaaart!

Wonder if FLIR is going to be shipping you , or if they were less than impressed by your shenanigens with the flirone 

Flir have no problem at all with my F1 stuff (message from the very top). I may even be getting a few Leptons from them....

[eneuro]

Is it possible to change percentage of this edge/countour blending in Flir's with MSX support from 0% to 100%, as well as are there any quick buttons to disable adding edges detection to final image, so we can track see thermal image overview with edges enabled, than find point where we need more detailed temperature information, so disable edges and have clear thermal image only?

If you need it, buy that capability.

No need to buy and pay for Flir's MSX logo, while former computer image processing specialists can write software which better fits ones needs. I won't pay for closed source MSX, but I could consider buying thermal camera with builtin second digital and light path divided so each of them could see the same image, than two USB plugs and if we were able to capture in the same video from visual & IR light than OpenCV can do the rest easy using latest Nvidia CUDO support for high performance when needed and no need for any Flir's MSX.

Are there any patent numbers on those Flir thermal cameras with  MSX support or still it is patent pending marketing spam?
It is common practice that this patent pending is later changed to patent number...

[ConKbot]

I think when Flir say automatic shutter they may be referring to the Lepton's automatic nonuniform correction, which relies on the fact that as a handheld unit, the scene is moving, so the long-term avarage of local pixels will be the same, and can be used to do nonuniformity correction.
I don't see a reason why it would need a shutter with the thermopile there as well, especailly as AFIR there was no indication of the min/max temps of the visible temp scale, so the Lepton is used for sighting only.

There's only one way to be sure - take it apaaaaart!

Wonder if FLIR is going to be shipping you , or if they were less than impressed by your shenanigens with the flirone 

Flir have no problem at all with my F1 stuff (message from the very top). I may even be getting a few Leptons from them....

  Good to see a company thats not turning all Golem from LoTR about their precious hardware and trying to tell you what you can and cant do with it.  (Like every other consumer hardware company... phones, TVs, gaming consoles, etc...  Tell me how HDCP provides me a benefit, or encrypted bootloaders and other shenanigans that (try to ) prevent people from putting new OS on xboxes as media centers and the like... )

Hopfully you can find some artist/organization to bankroll a big installation with a few dozen/hundreds of the thermal imagers.