Thermal Sensor size and Flir vs Seek

[Circa]

I am new to thermal imaging and would like to get a thermal imager for personal use like energy leaks and animals. After researching what I can afford I have narrowed my choices down to two, the Seek Reveal XR or the Flir One (Gen 3) phone attachment.
The Seek Reveal has a 206 x 156 Thermal Sensor and a 36 degree FOV. 9hz/fps refresh rate
The Seek Reveal XR has a 206 x 156 Thermal Sensor and a 20 degree FOV. 15hz/fps refresh rate. (Smaller FOV is better for objects at a distance)
The Flir One (Gen 3) has an 80x60 Thermal resolution and a 50 x 38.6 Degrees FOV (37 deg?). 9hz/fps refresh rate. The Flir also has MSX. (Larger FOV is better for being up close)

Is the larger thermal sensor resolution of the Seek more important than the wider FOV and MSX of the Flir?

[Vipitis]

there is also a PRO version of the FLIR ONE that offers a 160x120 resolution, so 4 times the pixels.

The needed FoV is really depending on your use case. Flower FoV means you can see further at distance - like detecting animals at range. High FoV gives you the full view of something clsoe up, like you don't need to take a step back(which is sometimes not possible).

there are other issues with the Seen, you can just search it up in this forum and learn more.

[sstepane]

And pro version offers change of FOV as far as far as I remember.

[Vipitis]

55°x43° it is slightly more.... but the 80x60 flir one also crops in a bit in the app and wasting a few pixels.

[jpanhalt]

Re:Other issues

Such as?   I am aware of the warranty issue. 

The fine print seems to indicate that if you buy from Seek itself you might get a replacement for a DOA.   If you buy from an authorized distributor, you get repair or refurbished.   And if you buy from a dealer who is not officially authorized, then the warranty is even less solid.  Some of the authorized dealers may actually offer a better warranty (e.g., replacement with "new in box" for a DOA within a limited time).   Of course, there is no telling how many "new in box" from Seek are actually repaired or refurbished.

[Circa]

I think I understand the FOV better now but the Thermal IR Sensor size still has me confused.
Sensor size in normal cameras, generally speaking, is the larger the better.
Is this the same way with Thermal? Is my Seek Compact going to be better than a Flir One (gen 3) because it has a significantly larger thermal sensor?

[jpanhalt]

I should think it is the number of elements (microbolometers) that matters, not the size.   The smaller the element, the more can be fit into the same space. 

[Circa]

Is this the same way with Thermal? Is my Seek Compact going to be better than a Flir One (gen 3) because it has a significantly larger thermal sensor?

I should think it is the number of elements (microbolometers) that matters, not the size.   The smaller the element, the more can be fit into the same space. 

I think I need to find a Beginner Thermal forum. Can anyone explain this simply?

[Spirit532]

I should think it is the number of elements (microbolometers) that matters, not the size.   The smaller the element, the more can be fit into the same space.

And the noisier the image is.

You simply cannot compare the Seek Compact Pro(which is about 330x250 resolution, via third-party software) and the FLIR Ex and Exx series sensors.
The Compact Pro will produce an incredibly noisy, low-sensitivity(~70-100mK) image, whilst the FLIR Ex and Exx sensors will output an incredibly clean, <40mK image, not to mention the fixed-pattern noise, the "patent avoidance" pixels of the Seek(which necessitates removal of every 15th pixel), and general nonuniformity.

Though I must add that though the FLIR ONE Pro(the 160x120) model and the Seek Compact(the 206x156 model) have similarly sized sensors, the FLIR ONE Pro(or, the Lepton 3 inside) is significantly less noisy, because it does a lot of in-core processing.
If I had to choose, I'd still go for the Seek Compact however, since it has a higher resolution and allows focusing.

I think I understand the FOV better now but the Thermal IR Sensor size still has me confused.
Sensor size in normal cameras, generally speaking, is the larger the better.
Is this the same way with Thermal? Is my Seek Compact going to be better than a Flir One (gen 3) because it has a significantly larger thermal sensor?

The field of view is calculated from the focal length of the lens and physical sensor size, which is not known(not accurately, anyway), that's basic geometry.
And this is also true for thermal cameras, the larger the sensor, the better - but you need larger optics and the manufacturing process for it and the sensor are expensive.
A full-frame-sized thermal image sensor alone will cost an order of magnitude more in fab costs(not retail price!) than the entire Seek or FLIR cameras, let alone the optics required to support it.

[Circa]

I wanted to dumb things down for anyone else who finds this thread and had a similar question and wants a straight answer and not microbolometer talk.
Hope this helps someone. I am an inexperienced consumer homeowner with zero previous knowledge of Thermal imaging.

Today I received my Flir One (Gen 2) to compare with my Seek Reveal Compact.
The Flir wins my unprofessional vote.

Both were tested using a Tripod phone mount and centered on the same spot.

For the past 14 days I have been using the Seek Compact daily to try to fix some draft spots in window and door frames. I was unsuccessful and couldn't determine the specific insulated areas.
With the Flir One I was able to see the exact spot that the most intense cold was coming from. It happened to be at the door latch area.
The Flir One reduced the cold area to an about 6 inch strip where the Seek looked more like the insulation in the wall had fallen and left a gap.

One draw back that the Flir has vs the Seek is that the Flir only has center temp. However the center temp on the Flir extends to 1 decimal (63F on Seek, 63.4F on Flir).

Field of View difference in nearly identical with the Flir looking better to me but that could just be due to the MSX technology. The Seek is zoomed in slightly more while the Flir is zoomed out.
The focus ring on the Seek Compact seems to just add problems.
The Seek also clicks to calibrate A LOT more often than the Flir.
Both apps seem to function identical with Hot, Cold, and Rainbow Palette.
Battery life: The Flir One (Gen 2) has a rechargeable battery and the Seek Reveal Compact is powered by the phone.
The included water tight case with the Seek was nicer than the rubber sleeve that came with the Flir.
The Flir seemed to lag in FPS a little bit more than the Seek but only while auto calibrating. Both are 9FPS/Hz.

My conclusion: Inconclusive. The seek works all the time while the Flir requires a well lit area to use MSX and autofocus.
To answer my thread's topic about "Thermal Sensor size and Flir vs Seek":
Thermal sensor size in these two devices is irrelevant to the eye. The smaller Flir sensor performed as well as the larger Seek sensor. It's the software backing up the FLIR's poorer resolution sensor while the seek doesn't need software filters to make it work well.

Edit:
Saved image resolution was one of the things that I thought was Sensor size dependent and was something I hoped that someone would have provided when I posted this thread; so here are my two test picture's image sizes:

Seek Reveal Compact
1.2MP 1280 × 960 96.4 KB
FLIR One (Gen 2)
1.6MP 1440 × 1080 150.7 KB

The Flir again produces higher resolution and larger sized images which I was told was irrelevant.

[Spirit532]

I wanted to dumb things down for anyone else who finds this thread and had a similar question and wants a straight answer and not microbolometer talk.
Hope this helps someone. I am an inexperienced consumer homeowner with zero previous knowledge of Thermal imaging.

The reason you like it so much is because of the MSX - it adds detail, sure, and helps you visually identify the item you're pointing it at, but it provides much less thermal detail. What you see as a gradient with the FLIR One might actually be a completely different issue, like several spots.

One draw back that the Flir has vs the Seek is that the Flir only has center temp. However the center temp on the Flir extends to 1 decimal (63F on Seek, 63.4F on Flir).

Someone with no experience in thermal imaging(you) will have no use for this anyway, because you're definitely not going to look up the emissivity coefficient of each surface you're looking at and you're not going to set it up in the app for each scene. A decimal digit is nice, yes, but meaningless. Both devices are capable of producing this with the right software, and the right software is not on the phone.

My conclusion: I will be selling my Seek Reveal Compact to purchase a Flir One Gen 3. I may even go for the Flir One Pro because of how much more I liked using the Flir.
To answer my thread's topic about "Thermal Sensor size and Flir vs Seek":
Thermal sensor size in these two devices is irrelevant to the eye. The smaller Flir sensor performed as well as the larger Seek sensor.

Edit:
Saved image resolution was one of the things that I thought was Sensor size dependent and was something I hoped that someone would have provided when I posted this thread; so here are my two test picture's image sizes:

The Flir again produces higher resolution and larger sized images.

The FLIR One Gen3 is a quarter of the resolution(80x60) of the standard Gen2(160x120), and only the PRO matches it(160x120), with a slightly higher resolution on-board visible light camera.

The image size is completely irrelevant as it only depends on the software - you wouldn't like an 80x60, 160x120, or 206x156 picture saved on your phone with no upscaling.


My suggestion is - don't dumb it down for yourself. Thermal is not as easy as snapping a selfie, and it won't be for the foreseeable future.
Learn how thermography works, read the topics, understand what emissivity is.
Put a piece of white paper over the visible light camera of your FLIR One to get the real thermal image, without MSX, and you'll see that it's a lot worse than you think.
The devices you currently have are capable of achieving good things if used correctly, and I wouldn't jump the gun on buying something that just looks better.

If you're looking to spend $500 on a thermal imaging camera(which is how much the the FLIR One PRO Gen3 costs), look at the Seek Compact PRO. It's got a 320x240 resolution sensor and provides great images. Not to mention some models(in the US and EU) provide >15Hz out of the box.

See this video:


Or better yet, find one of the deals on a FLIR E4, upgrade it to an E8 with software, and have a device an order of magnitude better than most of the phone dongles. There's some listings on eBay currently sitting at ~$300 in bids.

[Circa]

You make some good points. I am going off topic a bit, but here is a comparison of the leak.
What am I doing wrong with my Seek, could it just be defective? I heard seek was a nightmare to deal with for warranty and repair.

[Spirit532]

What am I doing wrong with my Seek, could it just be defective? I heard seek was a nightmare to deal with for warranty and repair.

To me, the Seek image looks heaps better than the FLIR image, even if a bit blocky.
As a quick hack, Cover your FLIR's visible camera with a piece of white paper(not black(!!!) - it will get really noisy, you need some uniform light), and you'll see the actual resolution.

For a more permanent solution use the third-party FOne Ultimate app - the native phone app crops the sensor to 120x90(which is as stupid as it sounds), and doesn't let you turn off MSX.

To add, you can also connect the Seek cameras directly to the PC(replace auto-installed driver with WinUSB using Zadig) and get an image out using JoeC's software(pinned thread in this subforum). This will let you pull more data out of it and process it further, including doing your own flat-field calibration(stick to uniform surface and grab a reference frame) to get rid of the noise.

[Jarrod]

I've researched this (a lot - even read the entire 96-page thread on the Seek Compact that ran from 2014-present) and in a nutshell the Seek trades sensor resolution over thermal precision vs. the FLIR products in this category.

In other words, if you are trying to image and/or measure differences that are just a few degrees apart, the FLIR blows the Seek out of the water.   Circa's A/B above illustrates this pretty clearly.

If on the other hand you prefer to see larger temperature differences but at higher visual resolution, the Seek is the clear winner at a given price point.  Defeating "MSX" (which to me is a bit of a gimmick and/or intentional tool to hide the woeful resolution on the consumer-friendly FLIR products) turns the FLIR images into unintelligible blobs of color.  "Blobs" is about all the 80x60 thermal camera on the FLIR One is capable of producing.

In the end I went with the Seek despite its serious drawbacks vs. FLIR in terms of actual thermal performance/sensitivity.  I sort of regret not having gone with the Seek Compact Pro, but at this time didn't want to invest $400-$500 on something that is more a hobby/toy for me than a serious tool.

[Circa]

I did another comparison using a candle and both imagers.
Found two more short comings with the Flir:
1. MSX doesn't work in the dark so focus is terrible and the image is extra blob-y, probably how the raw thermal image is supposed to look.
2. The Flir one Gen 2 doesn't have a high temp range so it goes negative and shows a cold black spot when the temp is out of range.
My imager is going to be used at night in the woods, so finding that MSX doesn't work in the dark and the resulting poorly focused and blob-y image is becoming a problem.  Why doesn't anyone ever say things about MSX not working in the dark?!!!
We go out to help search wooded areas to help find lost dogs, so I was intending to use a thermal imager to help with the searches. Eventually I am going to mount one on my quadcopter and transmit the image back over 2.4ghz or 5.8ghz to my ground station.

I've researched this (a lot - even read the entire 96-page thread on the Seek Compact that ran from 2014-present) and in a nutshell the Seek trades sensor resolution over thermal precision vs. the FLIR products in this category.

In other words, if you are trying to image and/or measure differences that are just a few degrees apart, the FLIR blows the Seek out of the water.   Circa's A/B above illustrates this pretty clearly.

If on the other hand you prefer to see larger temperature differences but at higher visual resolution, the Seek is the clear winner at a given price point.  Defeating "MSX" (which to me is a bit of a gimmick and/or intentional tool to hide the woeful resolution on the consumer-friendly FLIR products) turns the FLIR images into unintelligible blobs of color.  "Blobs" is about all the 80x60 thermal camera on the FLIR One is capable of producing.

I like your explanation and your puppy. The first thing I did when I got my imagers was tested it on my dogs, found some very cold noses.

[Jarrod]

Yes the temp range was the one big practical issue for me that closed the deal for Seek over FLIR.  I wanted to be able to scope out things on my 3D printer and the FLIR models cap out at 120º C.

People do say things about MSX not working in the dark.  FLIR doesn't though ;-)

[Circa]

Yes the temp range was the one big practical issue for me that closed the deal for Seek over FLIR.  I wanted to be able to scope out things on my 3D printer and the FLIR models cap out at 120º C.

Are those images from a Seek Compact or a Compact XR?
I also noticed that your images resolution are only 640x480 (or 480x640 in portrait). Did you crop them or are they saving at that resolution? My Seek Compact is saving at images 1280x960. Maybe a phone thing? I'm using a phone with a 1080 display resolution and a dual core processor.

Edit: I snatched your jpeg's metadata. The iphone 5S has an 1136 x 640 pixels.
I compared the metadata to my metadata and I noticed that my Android verion is using SDK 1.9.8.11 and your iOS version is using SDK 2.1.3.2.
Does screen resolution actually effect final image resolution?

[Spirit532]

It should be pretty intuitive why MSX won't ever work in the dark - it's just a visible light camera(a crappy one at that), all it does is edge detection and overlay.
The saved JPEGs are dependent on the screen resolution. They're just fancy screenshots with most of the app cut out, really. Don't rely on them.

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. Not to mention with that, you could probably strap a regular camera close to it and roll your own MSX - from a quadcopter's perspective, the parallax error is negligible.

When used properly, Seek cameras beat FLIR's current(3rd gen) offerings by an order of magnitude.

[Jarrod]

Regarding resolution:  I'm using the Apple/iOS Seek Compact (not XR) on an iPhone 5s.  An image upscaled to 640x480 seems to be what the current version of the app puts out in this configuration.

[Vipitis]

Flir one Pro has a high gain mode for higher temperatures.
You can also use a filter that works just like an ND filter so you can readout very high temperature. ThermalCamera+ offers function to add a Tau value and calculate those temperatures correctly.

On your comparison image, it should be noted that the flir app does a form or histogram stretching instead of applying the pallette linear to get the best contrast.

[MyThermalWorld]

I bought the first generation Seek Compact and also a FLIR One second generation (160x120 resolution) and I have compared them a lot. I rarely uses the Seek as it lacks of sensitivity and post analysis.

The Seek has better resolution, but way lower sensitivity and a lot of noice compared to the Flir One.
- Seek sensitivity: 70 mKelvin
Flir One (Lepton 3. gen): 25 mKelvin

Due to this the quality of the images was much higher from the Flir One camera. Another thing that's very valuable for me is that the images can be analyzed after and that can be done with ex. Flir Tools where multiple measurements can be done and palettes can be changed. There might be some Seek analytic software, and I have requested it from Seek, but those questions are always dodged.

Links to some compares I've done:

BUT then I'm looking for a small thermal camera in size with a minimum of 320x240 resolution and a contact at Seeks core modules informs me that the sensitivity of their 320x240 sensor is increesed to the same level as the FLIR Lepton module. Does anyone have any experience with what the Seek Compact PRO can deliver? Pictures?
(I own a FLIR E75 as my primary camera, but I would like a small traveling companion and thats where the Seek Compact PRO comes in place...)

[Spirit532]

Does anyone have any experience with what the Seek Compact PRO can deliver? Pictures?

See my video above. That's dumping raw images to the PC, with my own NUC added(only gets rid of FPN, to make the image more pleasant).

P.S. - Tried hacking the E75 to the E95?

[MyThermalWorld]

Tried hacking the E75 to the E95?

Thanks. The details looks good on Seek Comact PRO. Most of the reviews are using other color palettes than grayscale that I think Seek could have done better.

The thought of hacking it has crossed my mind, but I'm very satified with the camera as it is. But I will at least wait until the warenty is over.

[Jarrod]

Links to some compares I've done:

No argument about the sensitivity issues or lack of post-analysis with the Seek Compact, but I feel like I should comment on some of these comparisons you made.

On the first link (the one of the truck) I think the Seek actually looks a bit better than the FLIR.  That scene encompasses a wide range of temperatures and this is where Seek's higher resolution pays off (grill details, tire treads).  Not night and day, but I prefer the Seek image.

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.

In your fourth link, this is where the FLIR shines - pulling out contrast in scenes that have only a few degrees of difference across the FOV.

[IwuzBornanerd]

@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.  I am attaching some samples I just shot with my late model XR of my Pi2B doing nothing.  The first is set to minimum focus distance as restricted, the second is focused with the camera at the same distance, & the 3rd is with the palette adjusted so that it  is  not limiting as hard.  I think this is one area the Seek app fails in not allowing manual adjustment of the palette application.  The 4th image is with MY palette in which if there is a 0.2 degree Fahrenheit difference you should be able to see it.

As for how the Pro performs on narrow temperature range I am attaching some samples showing my mantle wall with a 256 shade grayscale applied over at 80 shades per degree Fahrenheit--or a 3.2 degree F range.  You should be able to see a 0.1 degree F difference, or at least 0.2*F depending on how many shades of gray your eyes can distinguish.  The first image is with a fresh re-cal compensating pixel "gain".  The second is the same only with stored cal factors that were made at a different time & slightly different camera temperature.  The 3rd image is with those stored cal factors and the bias correction turned off.  Seek constantly fusses with the individual pixel bias and the image gets very noisy with a few degrees of temperature change if this is not corrected for.  I believe that the Seek app does nothing to correct for this and therefore the Seek app cannot get clean images at fine temperature resolution.  A "3rd party program" can get you better results.

Also note that I have pixel averaging on in  all those images--I average in the adjacent 8 pixels at each pixel to smooth out the noise...Okay I'll add one with it off but bias correction on & with the fresh cal factors.