Yet another cheap thermal imager incoming.. Seek Thermal

[callipso]

This pic looks GOOOOOD https://fbcdn-sphotos-g-a.akamaihd.net/hphotos-ak-xap1/v/t1.0-9/p403x403/1779340_10107440634031644_4193828371859292485_n.jpg?oh=a0832251bb664421551ca14a40bb0107&oe=54AF3705&__gda__=1420626237_3e90ad32dc1196a0b40ce406413a636c

[eneuro]

There seem t be happy customers there and pictures look pretty reasonable for $199 !

Did you saw this comment by "Louis Rigo Jr" 

How about you stop trying to sell and give iPhone users an update!?!

Worth to see these visual vs IR comparisions from Motorola X with Seek Thermal tand look closer to those output IR images there:
http://wildfiretoday.com/2014/10/27/a-thermal-infrared-camera-attachment-for-smart-phone/

Update: Raster of 8x8 pixels cleary visible and it does not look better than simply linear resize I used  from 208x156 to higher 832x624 

One  should be able to have options in this Seek Thermal app to toggle different setting depending on for what purposes we use thermal camera..

But tha's fine-this thing can be usable with additional external lens and own software written for given thermal analysis task 

[Fraser]

I decided to give SEEK a small 'prod in the ribs' and publicly ask them to comment on the thermal gradient, noise, dead pixel count and their lack of response to email on the topic. It may be a bit mean of me to out the issues on Facebook but SEEK need to improve their attitude and communications if they are to be a competitor in the thermal camera marketplace. I would have expected Ex Indigo employees to know that already. Sometimes you have to be cruel to be kind.

Fraser

[bktemp]

Update: Raster of 8x8 pixels cleary visible and it does not look better than simply linear resize I used  from 208x156 to higher 832x624 

The 8x8 blocks are probably due to jpeg compression, they are also on the visible image.

[callipso]

I decided to give SEEK a small 'prod in the ribs' and publicly ask them to comment on the thermal gradient, noise, dead pixel count and their lack of response to email on the topic. It may be a bit mean of me to out the issues on Facebook but SEEK need to improve their attitude and communications if they are to be a competitor in the thermal camera marketplace. I would have expected Ex Indigo employees to know that already. Sometimes you have to be cruel to be kind.

Fraser

I don't use facebook nor have I an account so I don't know how it works, but can they delete the stuff you post? If my expectations are correct, this will be the case...

[efahrenholz]

I'm still curious about the black pixel pattern. Is this something that is common on vanadium oxide imagers?

Also, I found an article posted on gizmag about Raytheon producing smaller chips here
 
http://www.gizmag.com/raytheon-thermal-imaging-chip/32068/

It looks like they are really designing these for detection of objects, not the quality or accuracy. Seek then sought (pun?) to license the design for use in a smartphone application, which is right along the lines of what Raytheon is targeting the sensor at. It's a good product, but never intended to be used commercially. In the article they specified it would be a flashlight replacement, and in my opinion likely much cheaper than issuing gen 3 night vision equipment to combat soldiers. Each mil-spec rated thermal monocular could be sold at half the cost of a pvs-14, given the thermal core is less than $200 in bulk, where an omni vii intensifier tube would fetch over $2500 even in the private sector. Of course, Freescale is actually the manufacturer of the die.

Perhaps anyone else has any speculations?
 
My general reason for posting all that is to clarify that seek probably can't fix, or will not waste time with, the quality of the image as it is currently good enough for detection and basic spot thermometry. Unless the sensor gets better, this is likely only going to improve slightly with software and small refinements in the assembly. Its obvious they have done heavy post processing to make the image look good enough to sell to consumers.

[eneuro]

Its obvious they have done heavy post processing to make the image look good enough to sell to consumers.

Thx for this link.
They have more customers than those thermal cameras for the moment, so all they have to do now is create very good but... their own marketing image 

The 8x8 blocks are probably due to jpeg compression

Later realized that too and missed it's not PNG when investigated this

BTW: Is it even possible make shots from Seek Thermal app in something other than jpeg compressed?
Those videos are also probably are compressed and it is not lossless I guess.

Can anyone verify this just to know what is quality of those images and  what they output and allow to write to disk? 
Is it possible to record data from this thermal cam as lossless images in their official released app?

[Seek Thermal Tech]

Seek Thermal has been following this thread with great interest.  We would like to be as transparent as possible, realizing that competition may use this as ammunition, but we believe that in the end we will be helped far more than hurt by an open and honest exchange.
 
Seek Thermal Inc. has been built from the ground up bring affordable IR sensors to the commercial market. 
 
We greatly appreciate the professional attitude and creative troubleshooting your collaborators have demonstrated.  We are actively reviewing our product to confirm your findings.  Identifying these issues early in our production cycle gives us a good opportunity to implement improvements when appropriate.  With the low cost of our camera some compromises need to be made between performance and cost.  We will be looking for cost effective improvements to address some of the issues you have identified. 
 
Epoxy invasion.  The good news is that our lens attachment process is fully automated.  Thus the process ‘should’ be well controlled and any corrective action should be effective with low variability.
 
We image test every detector visually before shipment, so the worst units will be screened out.  Our experience is that anything under the shutter will be almost perfectly removed by a Flat Field Calibration.
 
Thermal Gradient over time,  We are actively investigating possible improvements to this issue.  No resolution or definite direction yet.  Note that for ‘relative’ thermography where the spot is fixed in the center of the display, we expect thermography to retain its ‘relative’ accuracy. 
 
Dark Pixels.  No great mystery here.  Every 15th pixel is intentionally blanked to avoid a potential patent infringement.  Seek has an updated design for future product that eliminate the need for this measure.  With the effective blur length of a 12 micron pixel resolving 8-13 micron Radiation, loss of single isolated pixels does not (in itself) degrade image quality.  

Thank you again for your interest in our product, we look forward to continuing our dialogue with the community.

Best,
The Seek Thermal Team

[efahrenholz]

Whoever you are (representative), thank you for your feedback!

Hmm, so blanking every 15th pixel, this is an interesting...

[marshallh]

Great to hear, answers some questions. Keep it up Seek Thermal

[iRad]

Seek Thermal has been following this thread with great interest.  We would like to be as transparent as possible, ... 

Thank you again for your interest in our product, we look forward to continuing our dialogue with the community.

Best,
The Seek Thermal Team

Thank you Seek Thermal for your reply and answering some of the concerns of a portion of your users. I'm sure everyone looks forward to your continued participation here, as it should help much to quell speculation and assure that Seek is indeed interested in a productive dialog with its customers.

That said, there is a very large portion of your customer base wanting to know when their iOS version of the camera will begin to ship. Unfortunately, we don't even have early production units to play with yet, and feel somewhat disadvantaged and displaced when reading these threads. I understand that there was a holdup until the App was released in the Apple App Store, but that has now been remedied (in fact an update to the iOS App has already been released since then). Please give us any news you can share on Seek's iOS camera availability.

[Fraser]

@Seek thermal,

Thank you for your engagement via this forum.

As you will see from this thread, many of us have high hopes for your company and its products. I am a long term user of FLIR cameras and the SEEK offers an amazing specification for the retail price that you have set.

I truly hope that your product is a runaway success and leads to further developments and products. As you will see from my post on Facebook, the concern to date has been a very unusual thermal gradient across the image and the image noise content. Noise may be improved with software but the temperature gradient appeared to be a hardware issue.

Thanks again for engaging... we are a useful and free human resource for UAT feedback.

Keep up the good work and communications 

Fraser

[KK]

thumbs up to Seek Thermal for responding to these concerns. Here's to your success!

[callipso]

Kudos to SEEK for joining the community, I'm very glad you guys don't just turn the blind eye!

[mikeselectricstuff]

Dark Pixels.  No great mystery here.  Every 15th pixel is intentionally blanked to avoid a potential patent infringement.

I'm really curious to know what patent claim this works around - Flir (I presume it's them!) have a lot of patents to comb through though....

[eneuro]

Hello, we are not alone in this thread 

Dark Pixels.
...
With the effective blur length of a 12 micron pixel resolving 8-13 micron Radiation, loss of single isolated pixels does not (in itself) degrade image quality. 

There is close to 7% such useless black pixels, so it might affect image quality depending on application 

BTW: Maybe, It could be nice to send each EEVblog member in this thread who made some effort and wrote more than 3 posts there, brand new Seek Thermal for teardown with latest software on CD/DVD ? 
OK-software not so important, one can write his own, but Seek experimental hardware dongle from Santa Claus to European EEVblog fans banned so far from official relase could be a very nice surprise  and clear evidence that really Seek Thermal Team is there on EEVblog

I could for example test my Linux version of Seek Thermal app on real hardware, not only on USB sniffied 8bit channel RGBA PNG's 

[amyk]

Dark Pixels.  No great mystery here.  Every 15th pixel is intentionally blanked to avoid a potential patent infringement.

I'm really curious to know what patent claim this works around - Flir (I presume it's them!) have a lot of patents to comb through though....

Seconded; if anything I'd think that specific pattern of black pixels was a patented technique for something like calibration as suggested earlier, and a sensor with all functional pixels would be too obvious to patent. But then again, I'm not so surprised, given the absurdity of some of the patents out there.

[mikeselectricstuff]

Dark Pixels.  No great mystery here.  Every 15th pixel is intentionally blanked to avoid a potential patent infringement.

I'm really curious to know what patent claim this works around - Flir (I presume it's them!) have a lot of patents to comb through though....

Seconded; if anything I'd think that specific pattern of black pixels was a patented technique for something like calibration as suggested earlier, and a sensor with all functional pixels would be too obvious to patent. But then again, I'm not so surprised, given the absurdity of some of the patents out there.

Could be something really silly like "a contiguous array of 15 or more pixels"...
A few years ago I worked on a project where a feature had to be activated by a button press because there was a patent on a similar feature that was enabled all the time....
 

[callipso]

Hello, we are not alone in this thread 

Dark Pixels.
...
With the effective blur length of a 12 micron pixel resolving 8-13 micron Radiation, loss of single isolated pixels does not (in itself) degrade image quality. 

There is close to 7% such useless black pixels, so it might affect image quality depending on application 

BTW: Maybe, It could be nice to send each EEVblog member in this thread who made some effort and wrote more than 3 posts there, brand new Seek Thermal for teardown with latest software on CD/DVD ? 
OK-software not so important, one can write his own, but Seek experimental hardware dongle from Santa Claus to European EEVblog fans banned so far from official relase could be a very nice surprise  and clear evidence that really Seek Thermal Team is there on EEVblog

I could for example test my Linux version of Seek Thermal app on real hardware, not only on USB sniffied 8bit channel RGBA PNG's 

OOOH OOH OOOH yes please

[efahrenholz]

I never thought about it but yes the black pixel doesn't actually affect the image quality. The 12 micron pitch is pushing the limits of the detection wavelegth. Anything resolving on any given pixel could very well be the same as any given neighbor. Very clever Seek. It seems the solution is to find the black pixel, find its 8 pixel neighbors, and average their value, fill in pixel with that value. It would be nearly likely to be that value in real life. No need to blur the entire image. Dead pixels in theory should be eliminated in a similar fashion. Then run a sharpening algorithm on the output to enhance edge detection.

[Hyperion]

Been lurking for a while... had my seek thermal for a while. Avid viewer of Mike & Dave's channels for years now.

My unit has about 2 to 3 degrees thermal gradient (measured against a foam blank) so it's not as extreme as the one in Mikes video. I can't justify modifying it for such a small difference, especially with the risk of damaging parts under the lens housing. At most it causes minor colour variation in the display output, for example, in this image (top left corner):

• 

A larger concern that I have is what seems like a USB access issue on the Galaxy S4. There seems to be some issue when the app requests permission from the user to use the USB slot. It often ends up with garbled noisy display giving extremely high temperatures (thousands of deg C) and the shutter goes insane. The issue is resolved by unplugging the camera and attaching it again, and the problem comes and goes seemingly randomly. By luck this happened when I first connected the camera, and I thought I had a dud.

Overall I am very impressed with the unit at its $199 value. Although I had to get it shipped by a third party to Australia, so it was really closer to $290 USD in the very end. And getting the app to work outside the USA was a bit of a pain until I found some seemingly-legit apk downloading sites. If the product can be made available here for the normal price+exchange rate, I'd get a second one to play with 3rd party lens options.

The inherent issue with this forum is that it brings together a lot of experts and "tech tweakers" who will not be happy until they have a FLIR E8 equivalent device for $200 or a camera that outputs something like this (NSFW?) https://www.youtube.com/watch?v=BS46C2z5lVE. If I were to look for a "commercial quality" infrared pyrometer from a retailer here, it will be $150-250 and it will measure one spot, that is the extent of its functionality. The price point for these Seek Thermal devices has to be met and so the product will never be perfect. I still think it is an amazing product for the casual user, tradesman and tinkerer alike.

I've been playing around with some astronomy CCD / image processing programs to see what can be done with the noise. No amazing breakthroughs so far other than making smoother / poorly defined images.

Obligatory portrait pic:

• 

[efahrenholz]

In reference to the gradient issue, I propose a software solution...

I'm guessing the software side at Seek could implement a gradient calibration event, where the user can, at will, select to calibrate out the gradient. The software could prompt the user to hold a sheet of anything flat up to lens, press on screen button to confirm its there and store a 5 frame average. On each FFC, combine the outside and inside calibration together before subtraction from the scene.

Of course it seems some users report the gradient in different areas, and in different intensities. Also it seems to be present at start-up or it gets worse after a few minutes. Calling this when the user chooses to would allow someone to calibrate it out when they feel the need (for scientific reasons.) It's probably not something necessary for the average user, but for the more inclined it would be a nice feature to have. It's certainly a great way to add value without modifying the camera.

[bktemp]

It is probably more difficult to compensate. From mike's video it seems to be dependend on the temperature: If the measured temperature is at the same temperature as the lens assembly, there is almost no gradient, but looking at a much warmer or colder object, the gradient gets more pronounced.

@Hyperion
Considering the temperature span of about 80°C the gradient looks much worse than 2-3°C in your image, more like 10-20°C.


Could someone take some pictures of a uniform object with min max display at very low and high and room tempertures to see if the gradient increases with temperature difference from the lens assemby temperature?

[Yushir0]

Seek has indicated that they may have a software solution to the gradient problem and will be pushing a software update sometime in the near future.

link here: https://m.facebook.com/story.php?story_fbid=761401457265761&id=628554970550411&refid=17

[ricksastro]

It's great that Seek is participating in this forum and plan to release a fix!

One data point here.   I tried to characterize the gradient, at least in my camera, and have determined almost certainly it's thermal oriented.  I used the Hi/Low to measure the drift over time on an even surface.   At the start, it is 80F and perfectly even, with both crosses in the center of the image.   Over time, the crosses migrate and deltas increase.

1F - 15s
2F - 35s
4F - 45s
6F - 60s
7F - 120s

Interestingly, generally the "blob" area goes down in temp relative to the starting point (hi stays 81F or so, the low drops to 73 or 74).   This tells me something is heating up when the shutter is closed, causing the image calibration to lower that area's estimated temperature.

If I use B&W color schemes and center point temp for relative measurement, it's usable.   But the threshold, color palettes and hi/low are about completely unusable, looking like a big mess.

So it looks like there may be a few different cases out there.  Some are saying they see the gradient right away.   I'm definitely not.   The above table is quite repeatable.   Not sure how a SW solution would account for a time based function like that, although if I could choose a "gradient correction" calibration time like after 2 minutes I'd be OK to let it warm up to get the gradient corrected.