Yet another cheap thermal imager incoming.. Seek Thermal

[schlafli]

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.

I keep seeing this opinion in this thread and it's simply wrong. It reminds me very much of Nvidia vs AMD discussions; "I bought X therefore it is now the best!"

Arguably one of the most important things in a thermal imaging camera is the sensitivity (NETD); the ability to distinguish between small temperature differences. This is where your image contrast comes from. If you have a poor NETD you have poor contrast in scenes with small temperature differences.

So let's compare the seek to the F1G2.

Flir One: 0.1C quoted in the specs. According to the datasheet, the lepton 3 NETD is <50mk, 20mk typical.
Seek: No-one knows because they don't report it.

Early on seek used to mentioned something like 1-2 degrees sensitivity. Even if we're generous and assume that's farenheit and use the lower end of the range, we're still looking at 0.5C sensitivity. The fact that they don't report is is quite worrying. Literally every other thermal camera you can buy has a quoted NETD.

So what does this all mean?

Is the seek thermal good value for money: sure.
Does it "pwn the flir one in all aspects": hell no.

I'll leave you with an image that you'd never be able to capture with the seek. It's a picture of a table where a cat was sitting about 15 minutes ago. The temperature difference between the paw print hotspot and the area between the paws is about 0.2C:



I've attached the fully radiometric image so you can see for yourself. Simply download flir tools and load the image.

[Redshift1340]

Seek is piece of crap - imho.

For a thermal imager priced at $250-300, regardless of its issues, it is not a piece of crap. This is not a product that can or should be used professionally, it isn't meant to be and definitely isn't priced to be.  They have some hardware and image quality issues to work out as well as software problems, but for their first product, it's not bad.  I would have bought one but I waited for the FLIR One gen2

[all_repair]

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.

+1, when you plan to use it as a general tool, one important feature is it cannot be too painful when it is lost or damaged. 

[miguelvp]

What I like about the Seek Thermal camera is the temperature range, in that aspect the Flir mobile falls short on both the higher and the lower range (120C to -20C for the Flir) so just 20C below and above freezing and boiling water.

As for sensitivity, my observation using the Seek SDK was 0.045C on the raw data per pixel.

https://www.eevblog.com/forum/testgear/yet-another-cheap-thermal-imager-incoming/msg714678/#msg714678

Granted that their software is still not mature enough and their algorithms to compensate for thermal drift and neighboring values is way too green.

But the sensor does have potential.

This is the before processing image of my hand, the full scene is from 24.52C to 32.85C



I do think with the right software the Seek sensor would be superior to the Lepton core. But of course they are not even close in the software department.

Edit: On the SDK, the floating point values incremented by 0.045C but it wouldn't go to the next float value until 4 values of the raw data, so I guess the raw data has the capability of sensing 0.01125C increments.

[Redshift1340]

One thing that stopped me from buying the Seek is the discrepancy between the image quality of different units.  Some images and videos taken by people look very good and almost noise free, while others are horrid and have tons of noise even with a wide temp range.  Seems like it's may be quality problems with manufacturing. For instance compare this photo posted by a user on their website with the ones daves posted.
https://scontent.cdninstagram.com/hphotos-xaf1/t51.2885-15/s640x640/sh0.08/e35/12142431_520619004782584_2144804393_n.jpg

https://scontent.cdninstagram.com/hphotos-xaf1/t51.2885-15/s640x640/sh0.08/e35/12139870_1694730324092320_578038826_n.jpg

[miguelvp]

Dave (edit: err, daves) needs to upgrade the firmware

Nah, these are old (last year) un-retouched with the very first firmware they released.





show me a lepton image of something with that temperature range (Edit: no pun intended re: Range)

[encryptededdy]

Seek NETD should be ~100mK.

http://support.thermal.com/hc/communities/public/questions/203781215-Temperature-resolution

The camera can detect as small as 0.1 degree differences, but the app can only report numerical readings in whole degree increments.

[encryptededdy]

Fully scaled



Measuring with FLIR Tools the entire span of that image is 0.9C

[Fraser]

There will always be a division between the FLIR and SEEK owners. This is not unusual and some of the comments are fun to read. There is little doubt that SEEK Thermal have achieved their objective of making thermal cameras available to the masses. Is it a fine, well engineered product ? In some areas, yes, in others no.

The Good

Nicely designed case. Very compact and professional design in metal.
First true budget thermal camera with decent resolution
Software appears compatible with many platforms
Impact of the SEEK camera on the market may have influenced the price of competing brands. Competition is good

The Bad

The micro bolometer appears to be very unstable and noisy
The software is still crude in terms of image processing.
SEEK thermal are slow to correct the software image processing issues, preferring to release new camera models instead.

The damn right Ugly

The physical design of the cameras internals is tragic. Placing the ultra sensitive microbolometer die directly on the same PCB substraight as the other components is bizarre and asking for trouble.
The enclosed lens tube is a crude and inelegant solution.
Lens mounting adhesive generated vapours that coated the rear of the lens with a layer of contamination.
Poor microbolometer stability requires the FFC shutter to operate at a hilarious repetition rate. How anyone could believe that such a high FFC event rate is acceptable confounds me.

Do I hate the SEEK Camera ? ......... Nope, it managed to produce a usable thermal image for an amazing retail price. There is a saying that applies though....... You cannot make a silk purse out of a Sows ear. IMHO this applies to the current SEEK cameras. I know some users have resorted to alternative hosts and better software, but they still aren't able to slow the FFC event rate and tame the instability in the microbolometer.

I am hopeful that SEEK will further develop their core and produce a more stable product. Sadly only SEEK know what their future developments will be and whether they will address the ugly elements of the current core.

I see little point in arguing which budget camera is best, SEEK or F1G2, there will be good and bad applicable to each and ultimately the purchaser needs to consider the pro and cons. I owned a SEEK and fell out of love with it. I bought the FLIR F1G2 and remain impressed with it. When you look at the size of the LEPTON 3 core, it is a marvel of miniaturisation. It DOES have to compromise in some areas. It does have a limited temperature range. The potential buyer needs to consider their uses for such a camera. Are cheap thermal cameras actually accurate below 0C ?....... You may be surprised when you test them. Is 120C maximum a major limitation ? Not in my world, but others have different needs.

I would like both the SEEK and FLIR One products to thrive. There is plenty of room for two product lines in the market. FLIR are definitely committing to the Consumer market now. They previously kept to the Professional and ProSumer markets and they are having teething troubles meeting market demand for the F1G2.

I hope most/all will agree that consumers are very lucky to have the opportunity to buy any sort of half decent thermal imaging camera for around $250.

A final thought..... If a particular camera meets the specific needs and desires of its owner, who gives a damn if it has some 'issues' due to essential cost reductions. Have camera, be happy 😀

Fraser

[Redshift1340]

I hope most/all will agree that consumers are very lucky to have the opportunity to buy any sort of half decent thermal imaging camera for around $250.

A final thought..... If a particular camera meets the specific needs and desires of its owner, who gives a damn if it has some 'issues' due to essential cost reductions. Have camera, be happy 😀

Fraser

Amen Fraser

[frenky]

There seems to be a bug in software if you rotate phone.
The labels with HI & LO temp are ok on live image but when you save image the labels are on wrong position:

Eggs in a pan:

[daves]

Yes, I admit when you play with Lepton, for example you adapt it to be IP thermal camera, then is pretty impressive for $200.
If you want it as a toy, ok. But for the rest E4, I mean upgraded E4, is much better, even with 4 times higher price.

I saw some articles about lepton hooked to raspi and this makes sense to me.

[Fraser]

FLIR are targeting a different customer base with the LEPTON based products and have a dedicated Consumer products division to deal with the differing needs of such a market. It is very different to the Producer and Pro markets in terms of product, marketing and support. I was very pleased when they released the LEPTON 2 as a discrete component. Hopefully the LEPTON 3 will follow.

Why am I talking about the LEPTON in a SEEK thread ? Well SEEK Thermal appear to be focussed on the consumer market with their present product range. Sadly there appears no interest in further development of the package into something that could be sold like the LEPTON, as a core for incorporation into 3rd party products. This is understandable as the SEEK design was not intended to be a 'core' for third party use. For developers wanting to produce a novel or specialist thermal imaging platform, at an affordable price to consumers, the LEPTON core appears to be the only runner in the race. I want to see thermal imaging used in many different scenarios that aid the consumer. Car night vision target avoidance being a favourite of mine. Such product development can only occur if the affordable parts needed for such exist. Competition is good for all and I would love SEEK Thermal to develop their own version of the LEPTON core that may be easily interfaced to different hosts, such as an RPi or other development platform. With such building blocks, great advances could be made in the application of thermal imaging to all manner of scenarios and challenges. Thermal drone Quadcopters with thermal vision have already shown their worth in non military applications, namely farm crop surveying for detecting blights that may be selectively treated instead of a blanket treatment of a whole field.

I am a card carrying member of the thermal imaging fan club and I hope that the consumer grade  technology continues to improve, and its size remains at least as small as the LEPTOn 3 for ease of integration and low weight. Cheap lenses need to improve and that remains a challenge for the
industry.

It is also a fact that FLIR did not provide a MSX overlay disable option for one reason...... The MSX provided additional image detail and context that makes it compete well against the SEEK cameras higher resolution. FLIR do not want users or potential purchasers to see the raw 160x120 thermal image only. They believe buyers may buy the camera with the higher resolution instead. I raised this with FLIR and the decision did cause a lot of soul searching at the development meetings. In the consumer market that contains many non technical users, marketing is king. First impressions count...... Hence the nice F1G2 design and presentation box! FLIR know that MSX gets in the way of macro working and other technical uses, but the product is consumer centric and not presumer. Sadly a simple MSX disablement option for the user is not on the table for future software releases.

A final note, and sadly not a very happy one. ......... Without quoting anyone, I am very aware that a large company with many products at differing levels of capability and associated cost, may be very careful about the capabilities that are provided by their cheaper products. It is not good business to provide a consumer grade product that can perform as well as your more expensive prosumer or professional products. You effectively undermine sales of your own top end offerings.

As many know, I have a large selection of thermal cameras, ranging from the consumer grade up to the $56K professional models from FLIR and NEC. It is apparent that the capabilities of these cameras have been deliberately set to fit into their various markets and to not compete too much with each other. Even the professional cameras have options that are set in software and not hardware. Some call this hobbling, I call it careful control of capability to meet its target market at a competitive price. Most thermal cameras contain electronics that are far more capable than that being offered to the user. Hardware such as lenses and software set the units limits. The E4 is an example of this, but let us not forget that even in its Virgin state, the E4 was a reasonably priced camera at the time of release. Owners benefitted from work to release its full potential, and apart from the lens, it has been made a most excellent thermal camera for consumer use. The fact that FLIR locked down it's abilities is just a fact of life in this industry. My $56K PM series cameras are all fitted with the optional high temperature filter (1500C) as standard. It is just not enabled as an option in the firmware yet. That option cost a small fortune as an upgrade and is just the setting of a software flag. That's life.

Well I am coming to the end of this epistle now. I leave you with this thought...... How wonderful it would be if a manufacturer of consumer thermal cameras decided to build and release the best thermal camera or core that they were capable of, without concerns about undermining their higher priced, higher spec cameras...... Because they do not offer such anyway ! SEEK could be such a company. They do not have the internal marketing politics of FLIR, NEC, TESTO or FLUKE

That's all folks

Fraser

[frenky]

Now I know why values are jumping up & down for 2000 or more...

If pixel drifts too much firmware corrects its value (after next shutter calibration) so that pixel value better matches the correct one:


But sometimes the correction factor is too big and the pixel is outside of desired range even after correction. This is when pixel value starts to oscillate between too big and too low value:

[frenky]

Next thing to do is to take a look at calibration (ID 1) and pre-calibration (ID 6) frames to see if I can predict when and for how much the pixel value will jump...

[frenky]

I think there is still some hope for Seek (with custom software)...

If you leave it to warm up for 1 hour, the pixels really start to stabilise:


And after applying correction value to each pixel, it starts to look really good:


I know it not really convenient to wait 1 hour for seek to warm up, but if you're trying to fix a problem around the house it might be worth the wait.

[Fraser]

A fellow forum member has done a F1G2 vs SEEK thermal differential sensitivity test. I was surprised at his results.

https://www.eevblog.com/forum/testgear/seek-vs-flir-one-g2-images/?topicseen

Fraser

[joe-c]

Hi Guys,

I also still working on the PC program for the Seek camera. i tried also other sources, but actually it is only a trial and error ...
I've made a few attempts to read another byte pattern frames, but either I could see no difference, or the camera stopped the work and had to be re-connected.

I hope that finally the SDK released soon.
Actually I only need 3 functions.
-Shutter Close
-Shutter Open
-Raw Frame (and I mean really RAW Detector Frame)
This means that the camera could move ever to a clean image. See the "extra" or "external" Calibrations from  users here.
Then you can also decide when a NUC is made ... I certainly would not do every 5 seconds the adjustment.

When the image capture of the Seek finally working as it should, I will implant it in my thermal viewer.
From then it is really helpful for me also.
Then it makes sense to think about hardware modifications times ... such as motor Focus, but also the whole board to pack a Tec in order to operate temperature stabilized. From the camera, a lot could be done ...

Come on Seek... give me access to the Raw detector values and the shutter. This should be not a big implementation for you, but a great base for me and other.   

[all_repair]

@joe-c, and others that are hacking with Seek thermal

On another thread that Fraser linked comparing the sensitivity of Seek and Flir 1, it appears that there is a sensitivity issue.  As the Seek sensor has a larger span, do you think by having a smaller span presentation just on the temperature-of-interest solve the problem? 

In the course of using the Seek compact on site yesterday, I found the MSX and the ability to have a sliding spanning presentation useful.   Seek is inhibited in implementing the MSX due to patent problem, I think it can be an add-on projects for people like you all.

Noise is less of the problem as a general tool for me.  Normally, I use my thermal imager to first get a general view. If needed (but normally not needed), the accuracy of the temperature can be achieved in many other ways after the problem area is identified.

[@encryptededdy, thank you for detailed explanation below.  Saving a separate post to say thank you so to keep SNR of this thead higher]

[Nik]

Found this video on youtube:

Seek Thermal Compact Review (New Improved Version)
https://youtu.be/NIY4irMIVsA?t=4m15s

[encryptededdy]

On another thread that Fraser linked comparing the sensitivity of Seek and Flir 1, it appears that there is a sensitivity issue.  As the Seek sensor has a larger span, do you think by having a smaller span presentation just on the temperature-of-interest solve the problem? 

The reason the minimum span is large is to reduce the appearance of noise. See this post I made from before.

I assume by a wide minimum span of the Seek your referring to the temperature differential pixel to pixel? In other words the ability to detect a sense a 76 degree item in a 75 degree room? I thought I remember reading that the Seek has a 1 degree "Sensitivity" I was calling it and the FlirOne has a .1 degree sensitivity. I don't know all the lingo so if we're talking about the same thing I've learned something new with span. Is the span of the seek controlled via software or is it something hardware limited?

As you know, a thermal imager normally maps the colour at the top of the LUT (eg. White) to the hottest object in the scene, while mapping the colour at the bottom (eg. Black) to the coldest object. However, as the difference between the hottest and coldest decreases, more image noise is visible (as you're stretching a smaller signal difference to fit a large colour span).

The seek has a wide minimal span. This means that the colour mapping will react as you expect down to a certain point (eg. 5 degrees difference). After that, the hottest is no longer mapped to white - it's mapped to slightly brighter grey and cold is mapped to slightly darker grey. This is to hide noise.

We can manually use photoshop to change the span of your dog photo to make sure the full pallet of black -> white is used, thereby simulating what would happen without a minimal span/a narrower minimal span.



Holy noise - You can clearly see why seek have set a minimal span limit.

So yeah, it's technically a software limit, however the software limit exists because the hardware cannot generate good images at low temperature spans.

[miguelvp]

Well, My a-hole son stole my seek camera. So he is totally cut off until eternity.

But I did order the Seek Compact XR

So good news, other than he actually got $160 for it without the protective case.
Bad news, he sold it with my USB adapter, so I'll have to get another one to hook it up to my PC.

Also good news, Seek did release the Android SDK as well.

By looking at that video I think I actually made the right choice with the XR instead of getting a non-XR to replace it.

BTW Nik, if you put the url directly without the shortcuts it will embed the video in the thread:

[frenky]

Just a correction. There is no public SDK jet. Some of you were just lucky enough to get beta release.
http://www.thermal.com/developers

Sent from my LG-D855 using Tapatalk

[miguelvp]

True, what frenky says.

[frenky]

I really don't understand why they don't release the SDK already....

They had no problem with releasing barely functional thermal camera to consumer market.
So why complicate things with release of SDK which will be used only by developers?

Those of us that are working with Microsoft technologies are already used of buggy software...