Yet another cheap thermal imager incoming.. Seek Thermal

[efahrenholz]

I'm thinking fur has an emissivity of around 0.8.

[Fraser]

Ok, I installed APP version 1.3 using Airdroid. Version 1.4 was declined by the American Verizon Motorola Moto G and would not install for some reason 

Once the software was sorted out the SEEK was connected and I watched as the thermal gradient gradually appeared.... no surprises there 

BUT........

I then proceeded to walk around my house 'happy snapping' the sorts of things consumers will target. The result..... I very much doubt that SEEK will be receiving complaints about the gradient from the user base at which it appears to be targeted.

My critique of the images that I obtained :

1. Thermal Gradient on left hand side : Yes
2. Noise content : Not good, needs more processing
3. Focus : Poor. Images are not as sharp as they should be at this resolution.
4. Temperature measurement accuracy : not tested yet but readings do look reasonable.
5. Resolution : Mediocre. Due to noise content and imperfect focus.

So having read the above you might think I am going to criticise the SEEK heavily. Well you would be wrong there. For $200 I do not consider this poor VFM and I am IMPRESSED with what SEEK Thermal have achieved at this retail price point. Yes some refinement in the noise processing is needed but that is expected by 'early adopters' of technology. It is still a great achievement to produce a thermal image that is even recognisable in terms of image content for less than $200. The engineering and presentation of the product appears impressive for such a low Retail price. And we must remember....$199 is the retail price that has to cover R&D, prototypes and production costs. I honestly do not know how they have achieved this and still made a reasonable profit   

The gradient issue...... OK I know that we have proof of a gradient issue in this SEEK camera. Some will recall that I previously asked whether the Gradient was truly such a big problem in normal everyday use. Well having done a rough and ready test in my house I am of the view that this gradient issue is being overblown. Sorry to those who disagree but that is my view.

When surveying a scene of flat temperature it is very true that the images do have an area of lower indicated temperature on the left hand side. This is the harshest of tests for a thermal camera as the flat temperature target reveals any inadequacy in any cameras FFC. The camera would normally automatically reduce the span in such circumstances and so increases the visibility of the gradient further.

When using the SEEK in a real world house scenario I was aware of the presence of a gradient on the left hand side only when viewing flat temperature area and even then the fact that I knew it was there enabled me to ignore it. I imaged my cats Alfie to see how the SEEK coped. Poor focus and noise masked detail of his fur that should have been visible at this resolution. I moved the camera over my other cat Oliver to see how the temperature gradient effected the images. No great drama there. The images looked more than acceptable in terms of gradient across the image.

I will upload some images in a moment but I have to say that it is my opinion as a regular user of thermal cameras costing tens of thousands of US$ that the SEEK has achieved great things considering its likely BoM cost. You will not find me calling it a toy any time soon as it does have very real uses if its limitations are accepted and allowances made. A tool that needs intelligent use, much like a $56000 thermal camera really.

I am obviously hopeful that SEEK will address the thermal gradient issue as that will further improve their product. sadly the less than ideal focus is likely to be a product of the hardware design and so will always be a limitation.

For those who disagree with my view of the SEEK, no problem at all. But that would tend to suggest that you are expecting too much for too little expenditure. I would strongly suggest that anyone who wishes to own a very refined and capable thermal camera at a very reasonable price consider the FLIR E4 and upgrade it to the higher resolution. You will not be disappointed. It gives my FLIR PM695 a run for its money in terms of  producing 'pretty' thermal images with good FFC.

As for the software. It is indeed early days and I would expect more development especially on the noise reduction front. The FLIR One shows what can be done with what, theoretically is an inferior microbolometer. The work being carried out by some very clever persons on this forum is also opening up opportunities for more varied uses of the SEEK camera. I remain very impressed by the quality of picture that is being achieved by them.

Finally (and I know the readers will be pleased that I am finishing here )

My sincere thanks to SEEK Thermal for producing such a revolutionary miniature thermal camera and selling it at an affordable retail price for many. I further wish to thank a very kind member of this forum (you know who you are) who purchased a SEEK for me and shipped it to the UK. Without his help I would still be without the camera and poorer for it.

Aurora

[Fraser]

My initial pictures taken with the SEEK camera. These were taken on a Motorola MOTO G running the version 1.3 application

The targets in these images were specifically chosen as representative of those that a consumer would take in assessing the cameras capabilities. I will add some external building images later.

Look at the pictures and see whether you can spot the thermal gradient issue. I believe you will be more concerned about the noise content and lack of sharp focus !

Enjoy 

[Fraser]

Alfie pics

[Fraser]

Oliver pics

[eneuro]

Enjoy 
...
Gradient at 10 minutes.jpg

Really nice temp difference in the corners-they show only 2 points-minimum and maximum, so many people have no idea that other corners are thermal useless images too 

Today I have more spare time to "sick" coding so I hope today will implement this gradient correction image processing and downgrading to 1Hz output frequency to get better quality thermal images when this donlge will be in tripod setup or security mode-not moving.

1.4.0.2

We'll see what they changed in software in comparision to earlier versions 

[DEHiCKA]

BTW, all images I've posted earlier are made with 1.4.0.2 on Galaxy Note 4. So, gradient issue is not solved yet.

[Fraser]

@eneuro,

Sorry I don't understand your comment "thermal useles". There are two types of thermography applicable to the SEEK camera....  Relative temperature monitoring and Radiometric measurements. The radiometric readings are adversely affected by the temperature gradient and so it could be said that the SEEK is not a good Radiometric thermal camera. Now relative temperature displays tend to be used to give the user a feel for the temperature distribution in a scene. Already knowing that the camera suffers a temperature gradient on the left side and, as you say, the corners does not prevent its use in this manner. "Thermal useless" infers that the corners provide no context to the image due to the gradient. I see no evidence of such in the context of the images I captured. The content is clearly understandable and the spot measurements useful as well.

Now if someone were to do a scientific numerical value teardown of the images I am sure errors will be evident BUT is this a reasonable act on a $200 bargain basement thermal camera ? IMHO No. Does it produce useful thermal images that help a user assess the thermal content of a scene with full knowledge that it is not perfect ? Yes I believe it meets that requirement well for the cost.

We need to be careful to compare apples with apples and not compare a low end budget thermal camera with a top of the line industrial thermal camera that costs tens of thousands of Dollars because the designer wanted a highly accurate Radiometric camera for serious use where errors can have unpleasant consequences. I do not think SEEK have ever claimed that their camera is fully Radiometric or for that matter even very accurate. 

Aurora

[slackaz]

@Aurora

I agree and disagree with what you have said. If looking at something with a high temperature then the gradient becomes less important because it affects low temperature areas in the image. The images you posted were of high temperature items such as a cat and teapot but I doubt that will be the main use of this camera for its intended audience. I highly doubt people will use this for security purposes and how many times do you really want to look at thermal of your cat or kids. I believe the main use for this camera will be DIY home inspections which look for low temperatures and not high temperatures. In these low temperature situations the gradient will cause more of a headache and result in more complaints.

Nice cat btw.

[Fraser]

And more from Aurora on product context  .........

I recall when I repaired my very first thermal camera. It was an EEV4428 that was used by military fire fighting crews on US and UK naval vessels. The 4428 was ground breaking in that it was an uncooled and portable battery powered camera. perfect for use when fighting fires in confined spaces.

The camera is analogue and used a Pyrolectric Vidicon tube that looks much the same as the old CCTV vidicon tubes that are long dead. The 4428 was released in the mid 1980's and lead to much positive comment from users as a "life saver". I bought a faulty unit and repaired its power supply. The first time it came to life I was presented with a relatively low contrast thermal image of a power supply that was within is FOV. I jumped for joy that day as I at last had a camera that could see heat rather than visible light. I wanted the camera for diagnostic work on PCB's and the less than pin sharp and low contrast image did not phase me at all. To be able to SEE heat was the important requirement. I later obtained the service manual and another 4428. Both have been returned to full operation. The pyro-vidicon tubes age and the various drive voltages need adjusting for best image focus and contrast. They both produce decent images now.

The 4428's were followed by the EEV ARGUS1 that was also Pyro-Vidicon based. I have sevral of those awaiting my attention. The first BST based Argus camera was teh ARGUS 2. I have a couple of those and the images are far superior to those of the Pyro-Vidicon based cameras. Better contrast and sharpness of image. Still only monochrome though.

I then moved onto high end industrial thermal cameras for my collection. these were what I used in my job so I was already aware of their capabilities. The ones I have purchased have often needed to be repaired so I have experience of their internal workings and design. They are almost hand built and use excellent engineering throughout. They cost i the region of $56000 as a result. Yes their images far outperform the earlier Pyro-Vidicon and BST based thermal cameras but that is not to say they are perfect. Far from it. They have thermal issues of their own and they have the usual noise issues associated with microbolometer based cameras. Professional cameras cannot get away with blurring the noise to disguise it or removing data from the image. As a result a cheap E4 may actually produc e a prettier image than a high end industrial camera.

Before people are too harsh on the SEEK camera, please remember how far we have come since teh 1980's and EEV 4428 Pyro-vidicon. What we have available to us today at an affordable price is simply amazing.

Aurora

[Uho]

Take a picture of the bottom of the door. This photo shows that it is - a tool or toy. For example, photo 64h48 pixels resolution. Made with a scanning imager.

[efahrenholz]

I wish they would at least allow adjustment to the palette mapping, not just changing colors. Like, change the stretching and contrast? How about custom palette colors? I prefer the sepia tone. Also we should be able to do a white hot/black hot change no matter the palette.

[efahrenholz]

I owned an isg talisman. That was a heavy firefighting camera. It did have a cooled 320x240 sensor, but it was full of dead pixels. even though it was higher resolution, I greatly prefer the Seek. I just wish it was higher framerate and had more image control. Itar sucks.

[Fraser]

I know the ISG Talisman series well 

I have an original Pyroelectric Vidicon Talisman and a few Talisman Wasps that use teh 320x240 BST FPA. Those cameras really do look hand made inside, and I do not mean that as a compliment 
The BST core is a very capable Raytheon unit. A pity yours had so many bad pixels. Sound like the BST FPA has had a hard life.

I attach some pics of my units. They have also had a hard life and await some TLC from me 

[Fraser]

All external doors in my house except garage access are new and UPVC so not much to see on those. Here is a picture of the internal garage access door.

I already knew it leaked around the bottom edge but it was good to see that my cover on the garage side keyhole was preventing cold air whistling through it.

Gradient is definitely worse in lower Delta T scenes, no argument. Can I still use it as a thermal camera ? Yep    Know your tools limitations and work within them.

I deliberately inverted a second picture of the VW car to show the gradients effect. The colder side of the car moves ! But then I do get the intended image of the car with heat leaking around the bonnet so is the gradient preventing me working ? Nope.

No one would claim the SEEK is anywhere near perfect but it is IMHO still a lot of thermal camera for your money.

I look forward to SEEK Thermal's software fix and hope it addresses the gradient to a level acceptable to most users. Next thing they need to address is image noise content 

Aurora

[Fraser]

SEEK saved image file names ?

Maybe its in the on line manual but has anyone worked out the damned picture numbering on the SEEK please ? When you download them they are not shown in the chronological sequence due to the numerical 'names'. Now that will be something that annoys me !

I am off to study the manual on my phone in case it is explained there.

Aurora

[miguelvp]

No idea, looks like a global unique id generator if that's the case it's going to be pretty random.

[Fraser]

Weird, why not use the simple numerical sequence found on most digital camera. I wondered if it was a time date sequence but no. Then I wondered if the title held data for the APP but that would normally be embedded as jpg Metadata. I will keep digging and maybe ask SEEK but they are not v=being too communicative at the moment.

[Fraser]

Found the answer regarding file names.

It is deliberate to enable file sharing. The file names are random to prevent file name conflicts between users of the camera who share the images with each other. I wish they had made it optional !

https://play.google.com/store/apps/details?id=com.tyriansystems.SeekThermal&hl=en

What's new:
* Corrected Thermal+ thumbnails.
* Corrected thermal image sharing issue - files are now uniquely named.
* Improved image quality.

[eneuro]

The file names are random to prevent file name conflicts between users of the camera who share the images with each other.

Common, those "sick" names looks not so bad.. but only when compared to SHA512 

$ sha512sum  sts.raw.img.000011.png
700585e4c04d94059f93516807793d077c685b9e2953b3e1d9501517f9afe166bf8670c6cb553d55f5d97f8ea95545bb1d04b70eda022ca3e85fdb0f03c34c21  sts.raw.img.000011.png

[kihon]

Hi guys

Can someone with the seek upload some pictures of a house from outside please? I want to see how good this would be to detect sources of heat loss.

Thanks

Kihon

[Fraser]

I included one picture of my house in the images on the previous page. You need a house that leaks heat though as most modern UK houses are quite well insulated ! My house looks pretty flat on all of my thermal cameras so isn't a great test piece for the SEEK.

Looking inside a house for cold spots is also part of the survey and that provides better contrast.

Aurora

[Hyperion]

I noticed a reduction in noise going from version 1.3.0.1 to 1.4.0.2, has anyone else thought the same?

[Fraser]

I will take a reference image and install 1.4 to compare noise content.

I have just received a response to my request for hardware supply assistance from SEEK for R&D. No dice, just a canned response about shipping into Europe sometime in the future.

Aurora

[efahrenholz]

I was just browsing the e4 forum and I saw some photos of the raw image output. Not great looking at all. But that's to be expected. What really is interesting though is whatever they are doing to clean up the image so dramatically. What the heck are they doing to clean the image output so well?

Also, I did some experiments, and the lens on the Seek might have intentionally placed a rubber grommet over it to decrease the depth of field. Basically, its letting in less radiation in order to improve sharpness of the image. I'm contemplating breaking off the lens part of the outer case and 3-D modeling up an adjustable focus lens holder which will glue directly to the case. Just have to remove the lens from the housing without breaking it.

EDIT: it appears that the casing has an outer piece for the lens, purely cosmetic. The front half of the case has the lens housing molded to seat it, but it not a removable piece. Bummer.