Thermal Sensor size and Flir vs Seek

[Jarrod]

@MyThermalWorld, I agree with @Jarrod, except I don't think you will actually focus "a couple inches" from the pcb unless you have removed the focus restriction from the lens assembly. 

The Seek Compact lens was exactly 2" from the CPU or GPU in those two images I posted.  The thing that is important to realize is that the wider angle lens on the regular (non-XR) Compact model has better close-focus macro capabilities than the telephoto lens on the XR.  The Compact XR rules for pulling in detail from faraway objects.  In contrast the regular focusable Compact rules for macro shots or wide angle shots of nearby objects (~1 inch through 50 feet).

[IwuzBornanerd]

I understand that, and your image is reasonably well focused at that 2", but the specification is, I think, 6" for the non-XR and is why I questioned the number.  I can get the XR acceptably focused to 2" after removing the restriction, but an official number might only be 4".  It is harder to tell perfect focus on a thermal image.

[Jarrod]

I have never seen 4" or 6" in any official specs for the Seek Compact, maybe I missed it.  All the reviews I've seen say 1-2" is the minimum focus distance and that is consistent with what I observe with my unit.

To get the attached image, I set the lens to the minimum focus distance and then moved the camera until I achieved best focus.  I think the result is acceptable although it does "showcase" the low of sensitivity of the Seek sensor.  Could the image be sharper (in terms of arc minutes of resolution per pixel) at an intermediate distance?  Probably so, most optical systems perform best at intermediate values of their various specifications.  But if you are trying to troubleshoot small electronic parts, making this compromise is probably worth the tradeoff so you can tell what you are looking at (in this case a micro-USB port and several tiny SMD components, from about 1.5" away).

One additional edit: For the $164 street price I paid, I'm pretty satisfied with the capability of this device.  I bought it mainly to image 3D printer parts, and for macro work with electronics.  I am fairly confident that, for those two uses, I would be less satisfied with a FLIR One for two reasons.  First, parallax is a real issue with MSX when you are shooting close up, turning one of its oft-cited advantages into a handicap.  Second, FLIR One cannot provide any detail about temperature for objects >120ºC and I regularly work with temps between 180º-240ºC.

[IwuzBornanerd]

I'm sorry @jarrod, I didn't mean to be contentious or turn this into a detailed technical discussion.  My only issue with anything you said was what exactly "a couple" was in your mind vs. my mind.  You stated it specifically and I accept that as well as the fact that your image is just as focused as mine.  I was merely trying to explain "where I was coming from".  As for the distances I stated, when Seek first came out with the XR they did indeed state a minimum focal distance of either 8 or 9 inches (I forget which) and when they made the wide angle version adjustable focus they stated a shorter distance for that one & as far as I recall it was 6".  I see now in their official spec's they don't make any claims beyond "adjustable".  Maybe they changed their set point.  The 4" number is what I had calculated from "the lensmakers' equation" based on distances I had measured in my old XR after removing the focus restriction.  For the Pi images I posted above I had my unmodified XR set at or a tiny bit above minimum focus and a distance of about 10" from the Pi.  One thing I thought of after my last post is that the images would appear more crisply focused without the pixel averaging I do, and which Seek does some form of.  You can see the difference in my last wall photo above compared with the first one.

And again, I agree with you about the Seek cameras--I am quite pleased with what I can do with them. 

[IwuzBornanerd]

As for quadcopter use, Seek cameras are 100% the way to go. You could strap one to a rPi Zero or another tiny linux-compatible device and use the available code to generate an analog video signal for transmitting using generic RC RX-TX sets.

Do you have a link to that "available code"?  I'm not sure I'm familiar with it.  I had to go to extremes to get 8fps out of the ordinary Seek with a Pi Zero....But I wasn't running it "headless".

[Spirit532]

Do you have a link to that "available code"?  I'm not sure I'm familiar with it.  I had to go to extremes to get 8fps out of the ordinary Seek with a Pi Zero....But I wasn't running it "headless".

This is the original Linux library the Windows software is based upon.
https://github.com/maartenvds/libseek-thermal

[Jarrod]

I see now in their official spec's they don't make any claims beyond "adjustable".  Maybe they changed their set point. 

Entirely possible, we've seen that they have even made substantial changes to the firmware and/or hardware to significantly reduce noise but have made no public mention of it.

[IwuzBornanerd]

This is the original Linux library the Windows software is based upon.
https://github.com/maartenvds/libseek-thermal

Okay, that one, thank you.  I thought maybe you knew of something more complete & "ready made" for drone streaming.  I am considering trying to stream temperature values out of the Zero to make the camera wireless with my own code, but that will be months away for me...even if there was something ready made.

[MyThermalWorld]

On the second and third links, quite frankly this is not how one would image a PCB with the newer focusable Seek Compact models.  You would hold the lens a couple inches away and focus it.  You would then pull SIGNIFICANTLY more detail out of the Seek than what your pictures show (although a narrower FOV).  See attached pics of my Pi with the latest revision Seek Compact.

@Jarrod @IwuzBornanerd I agree that the PCB comparisons where unfair towards the Seek. Worth noticing as I worte in the same post, this was the first Seek Compact camera without any focus capabilities. If the temperature range is high in the image, the Seek's resolution defenitely has it's advantage over the FLIR One. But it requires "sunny" images to bring the noise away. Also worth noticing regarding the seek comarisons, they where shot before Seek upgraded the app alowing to remove image smoothing. Removing it gave a lot more details. But still I think FLIR One's low resoltion (160x120) still gives the same details at close range compared with the Seek Compact.

I use my FLIR One to take images where I want to post analyze the images and uses Seek when the temperature is high enough and it's for the image itself an not the data.

Electric motor shot with Seek Compact (Added FLIR Iron palette to it for comarison)


Same electric motor shot with FLIR One.


I'm adding a few Seek images with quite good details for the resoluion:



Audi V6 engine


Horse. A bit noisy but still sharp


Roof. I think this stands out in details.

One thing I like better with the FLIR is the palette design, especially the iron palette. As I think Seek uses a linear path between the color points, FLIR has a more curved one that makes the red -> purple -> blue -> black transition much softer. For me the Seek iron palette is like draging the contrast leaver to far. This is my opinion on it, but of course others thinks different abot it 

Flir iron palette color graph


Image shot with Seek iron palette


Image added FLIR iron palette

I usually takes Seek images with the grayscale palettes so I can add my own palettes later.

[Jarrod]

Those all look really good, the Gen2 FLIR One is pretty impressive.  Do you remember what it sold for?  I understood that your Seek images were the original non-focusable version, and that version was not a good option for close-up imaging.

I was a bit let down when I learned that when FLIR came out with the Gen3 product, they priced the 160x120 model at $400 and dumbed down the base model to 80x60.  When I was in the market for an entry-level unit recently I only had the newer focusable Seek Compact vs. the FLIR 80x60 model to compare at around the $200 price point.

[MyThermalWorld]

The FLIR One works impressive good compared the price and relative low resolution. But it's disepointing that the 3. gen was lowered to 80x60. When everybody else squeezes higher and higher resolutions of of the same size, all FLIR did was to add software improvement of the already existing 160x120, and increesed the price. And that's the main reason why I'm not replacing my 2. gen FLIR One with the newer FLIR One Pro.

I'm hoping that Thermal Expert will release their iOS version soon, and I'll might by one.

But one thing I have noticed when comparing the Lepton 2. gen (80x60) with 3. gen Lepton (160x120) is that there is less noise in the 2. gen. I think it's due to larger pixels alowing more lwir photons into the pixel. They squeezed the 3. gen Lepton pixel size to fit all pixels.