Yet another cheap thermal imager incoming.. Seek Thermal

[Fraser]

@Seeker,

Who said it would be a potential replacement for the more expensive industrial thermal cameras ?
Not I.

To remove the emissivity setting is just weird. It is, after all, a mathematical offset against which all measurements are adjusted. Incorrect emissivity offset = incorrect reading of temperature.

This could of course indicate that SEEK are not confident in their units measurements so why bother with 'fine tuning' that may just confuse buyers.

Wait until you try to measure the temperature of a DUT that is lower emissivity though.

i.e. A DUT with emissivity of 65% requires a positive offset of 1.538 (100/65) in order to provide any sort of reasonable measurement. If the Seek is set with a fixed 97% emissivity offset you have a huge error in measurement. The unit will work fine on most dull, high emissivity surfaces though as they will likely be in the ballpark of the 96%.

Much depends upon the intended use of the SEEK. For personal building survey work it should be fine. For the more unusual DUT's the user just needs to be made aware of the traps caused by fixed emissivity. Remember, emissivity is at the very heart of thermography and is the most basic of concepts that should be understood for many uses of a thermal camera, beyond a toy,

I highly recommend that users of ANY thermal camera read up on the basic principles of thermal imaging so as to get the most out of their device. It is very different to visual photography. I can recommend the user manual for FLIR equipments and those of other thermal camera manufacturers like FLUKE and NEC AVIO, they normally contain a chapter detailing the basics.

I am under no illusions regarding the capabilities of the SEEK camera and its most likely market. It doesn't help to warn people of the pitfalls however 

Aurora

[Seeker]

No, it was just a reaction to your parting shot, Aurora...sounded belittling, as though you thought there was something amusingly crappy about it.
No idea whether emissivity settings would affect me or not...primarily for isolating domestic heat leakage (and some playing). If it wasn't so inexpensive and neat I wouldn't entertain buying it.

[Fraser]

@Seeker,

Fair enough...  I was having a pop at SEEK Thermal over the loss of emissivity setting ..... that was just my reaction to the news. I was running Software 1.3 prior to 1.74 so it was a surprise loss.

From my early comments in this thread you will see that I was/am a keen supporter of what SEEK Thermal are trying to do......affordable thermal imaging for the people. It just makes me sad and a bit scornful when they cock-up ! I honestly want them to do well. I should be patient though. They will get there in the end.

Didn't intend to offend

Aurora

[igor88]

last update totally ruined application performance

it lags as hell on my MotoX 4.4.2

camera images quality in now worser than in previous... 


...update

uninstall and install again - all is ok.. 

[Uho]

I have a Moto G. Does not slow.

[igor88]

re installed it, all is ok

[Fraser]

Performance on Moto G appears OK but I have to wonder how much computational work SEEK have loaded on the host to sort out the gradient.

For me the jury is still out on the image quality. In my ambient test I thought I saw more noise in the image but that isn't the fairest of tests. My hard test for thermal cameras is to see if they can reveal the plaster dot and dabs behind my homes plasterboard walls. Out of 29 thermal cameras, the SEEK is the only one that cannot reveal such detail. It is just too noisy and likely the span too wide. Sadly as span is reduced, noise gets worse so that does not look good for the SEEK.

BUT......

This is the cheapest thermal camera on the present market in terms of resolution against cost so I will not criticise it just because it fails a hard test. I have yet to do some real world imaging though so time will tell. Hopefully the application will continue to improve over the coming months.

Very glad that the gradient has been resolved. Now as to whether it was cured or disguised so that we can't see it....well that's another matter.

Aurora

[Fraser]

On a different topic.....host compatibility ....

Google Play used to refuse to load the app on a Moto E host. It now allows it. The app runs but the seek camera is not recognised. I suspect the E is not equipped with OTG.

Just in case anyone wondered about the E. 

Aurora

[Lurchbox]

Hey guys, can someone please link me to the newest APK? All of the apk downloader sites are giving me the old version 

Also did the new update fix the issues with the nexus 7 showing a blank screen?
I also have been messing with the seeker app slider settings and have found ranges that are very useful, will post pics soon.

Thanks

[Uho]

I downloaded using raccoon-3.3.exe.

[miguelvp]

For me the jury is still out on the image quality. In my ambient test I thought I saw more noise in the image but that isn't the fairest of tests. My hard test for thermal cameras is to see if they can reveal the plaster dot and dabs behind my homes plasterboard walls. Out of 29 thermal cameras, the SEEK is the only one that cannot reveal such detail. It is just too noisy and likely the span too wide. Sadly as span is reduced, noise gets worse so that does not look good for the SEEK.

Odd because before the update I could see cold spots in my wall where some repairs where made and that was with my wife's phone (Galaxy S4), with the PC app it was easier to spot them, I can't test it now because I wont get my camera back until next week.

[Seeker]

Yes be interesting to discover whether the gradient has been removed or cloaked. The reduced size of the app may suggest removed, but that's one of my guesses.

No guess--the Moto E definitely doesn't support OTG, I just bought one in error; had to sell it and get the G instead.

[Lurchbox]

I downloaded using raccoon-3.3.exe.

Thanks that a good tool 

[efahrenholz]

I checked the new version. I'm one of the individuals who devastated my camera in the sake of science. I might need to reinstall app. On my phone (Note 4), the image lags by almost a second behind real time. This us pretty bad. Also, the image seems to have weird sharpening speckles around edges. So they added a sharpening filter. My gradient is mostly gone, except along the bottom of the image. There is a actual warm edge. It runs the whole length. I checked against a flat wqll, and it appears as almost edge glow. Another thing, does your images have a weird splotchiness to them? Not quite smooth, but more like someone did a heavy blur an image with heavy noise. I'll get some photos up when I get home.

I think they might have a way to control the power through the software using some function calls. Perhaps they changed or tuned something to help reduce chip heating.

[Fraser]

Regarding chip heating.....

I ran my camera on s/w 1.3 for 10 minutes. Took some test pictures of a flat surface and then installed s/w 1.74. Total down time less than a minute. The image went from gradient to no gradient so either heat is not an issue or the new algorithm (?) can cope with it as my camera would still have been warm from the 10 minute warm up period and testing.

I have noticed a change in the noise pattern between 1.3 and 1,74. I am not surprised as there may be some clever filtering going on. Take a look at the ambient flat surface images I uploaded yesterday. Apart from the loss of the gradient there are other differences visible.

A part of me wonders whether the whole gradient issue was caused by the cameras image processing. The app has shrunk so maybe the faulty 'module' has been removed and replaced with a different approach, hence the additional image processing file.    It would be great if SEEK advised what the original problem was as that would gain them respect and an appreciation of what they have had to do to rectify the gradient issue.

Aurora

[Fraser]

I have just noticed something else that has changed with my move from S/W 1.3 to 1.74.....

The SEEK logo is smaller and lower contrast so that it does not dominate the image so much. This may have changed in s/w 1.4 but I never installed that version so I do not know. I have not bothered to switch it off but it shows SEEK are considering even small improvements in the user experience.

Having revisited my test images there is a definite mottling effect in 1.74 when looking at a uniform temperature surface. Yet we have seen an improvement in image definition .... no idea what would cause apparent blurring in one situation and better definition in another. As you can tell I am no expert on image filtering and enhancement.

Aurora

[YewSuck]

It has been suspected some of the gradient was caused by heat near the sensor. Maybe they just found a way to make the processor run cooler (optimising the code, underclocking, etc. who knows) thus virtually/completely eliminating the issue.

Now I have to get me one of these, damn it! 

That very well maybe some of what happened.   Maybe they sped up the shutter speed, Thus reducing the energized time making heat.  They might have lowered the wattage to the shutter coil as well.  If any of that is software controllable.  I am thinking they are offloading more of the processing to the phone.  Maybe using less features of the die's abilities and moving them to software instead of hardware.  What ever they are doing is working.  That is good.  I am sure the Guru's of this forum will let us know what is going on once they get a chance to look at things.  All in time.

-YS

[efahrenholz]

I tried the indigo palette. Usually it looks terrible. I must admit, they really did some good work on this new version. Now, let's see if they can add some new features like changing the max/min thresholds. Or is that called the temperature span?  They could add a palette bar, with  a slider to change the center of span, and pinch to scale the span. Hint hint Seek ;-). They could even plot the temperature range on it.

[eneuro]

I've been working on pan/tilt motorised head lately:

I've been working too, but I think we found easier way and more compact package to prepare Seek dongle for more serious tasks


Unfortunate, can not reveal details while idea behind this might be novel  in some thermal applications.
The only thing I can say for the moment- it is circular motion approach.

Will see what Seek changed in his new software version 
Anyway fixing gradient, by removing emmisivity setting looks like customers gets better looking images and nothing more   
It is easy "fix" change image and make it looking flat but it is much more challenge make in the same time  correct temperature measurements, so they gave people better looking images while most of them have no chance to see correct thermal object temperature readings while.. they have only this Seek device without emmisivity settings 

After winter holidays back to work 

[Fraser]

@efahrenholz

I agree, the imagery definitely appears much better than in the previous version. I see you are a fellow cat owner.....cute and always on hand to model for the camera    They are a study in animal thermal profiling in themselves. Also useful 'models' for testing automatic wildlife thermal tracking and imagery  I presently use very expensive cameras for such activities.

As I said previously, it would really interesting if SEEK were willing partake of this forum, as other manufacturers have done previously, and discuss their challenges and a high level non sensitive view of the approach they took to solve the various challenges that they have faced. I believe they could do this without giving away trade secrets yet they would also gain a community that wishes to support and improve their product. Free Beta testing as well  Now that the gradient issue has been proven resolvable, the future looks bright again fro SEEK if they embrace their customers and improve their communications a little.

I will still likely buy a FLIR One Android but at least I will be keeping my SEEK and hopefully developing its usage. Its compact size and low price is still a novelty that can prove very useful in the lab and outside. A user could mount the SEEK on a simple 'copy stand' and connect to the host using a cable. After fitting a close-up lens it would then be possible to carry out thermal profiling of anything placed on the copy stand, such as power supply PCB's etc. It could be a very compact little assembly. Just needs a mounting bracket to hold the SEEK. An easily created item with a 3D printer or even a piece of wood !

Aurora

[Uho]

It seems that the work of the shutter has not changed. Oscilloscope view shutter.

[YewSuck]

Awesome thanks for checking that theory on the shutter.   

    BTW, The Seek Thermal device works great on my Samsung Galaxy SII after a firmware update to 4.4.4 from XDA Forums.  I used the AOSB Rom.   Those people that do not have Full USB support should check there and get an aftermarket Rom.  Be warned you will void your warranty, though.

-YS

[efahrenholz]

I still can't figure out what they did to relieve the gradient issue. I, however, have a gradient. It is just in a different place now. This leads me in the direction of a mask. It's not a physical fix, but a mathematical fix. Also, I notice that my temperature gradient does not exist if the whole scene is above about 65F. I get less than one degree variation in hi low. Below that, and the scene is about wrecked. If I check the freezer or a cold sky, I have a very cold corner and the whole bottom edge glows. It looks like back light bleed on an lcd. I can't fault seek on this, I screwed my alignment, possibly damaged the sensor die as I have a chip in the corner of the glass. That might explain the very cold corner I get. Maybe I screwed up something when I shorted the bond wires. Who knows. Can anyone else develop a gradient at low temperatures? I would love to see some bread crumbs.

Anyone else notice the sharpening artifacts? I get bad noise along object edges.

Also, I have been getting very bad lag. I am using a note 4, so I'll check it again at a galaxy s4 later to confirm. I wonder if the dirty sensors take longer processing time to clean up, while the pretty sensors have quicker refresh rates.

Finally, I will be ordering a flir one gen 2. I *was* going to buy another seek (considering I basically murdered this one), but I still don't see any improvements hardware wise that give me confidence in the noise level.

[Uho]

I also have a problem with the gradient at low temperatures. In a warm room - perfect. But sending the cold balcony - a problem. Even if the imager is in a warm room. Until I found the cause of this behavior.

[efahrenholz]

I really wish the seek engineers could answer a few questions. Like why the white paint on lens housing. I sharpied the inside of mine. I wonder if that is what's messing with the gradient correction...but still why would it work perfectly fine above 65F? This all makes little sense.