Author Topic: Another smartphone thermal imager from i3  (Read 149716 times)

0 Members and 4 Guests are viewing this topic.

Offline Conure

  • Contributor
  • Posts: 39
  • Country: se
Re: Another smartphone thermal imager from i3
« Reply #250 on: March 28, 2019, 06:59:28 pm »
A few days ago I got my TE-Q1. I find the pictures surprisingly noisy. Is this how it's supposed to look? I disabled the "enhance image" box. The phone is a galaxy s6.




« Last Edit: March 28, 2019, 07:20:17 pm by Conure »
 

Offline Vipitis

  • Frequent Contributor
  • **
  • Posts: 876
  • Country: de
  • aspiring thermal photography enthusiast
Re: Another smartphone thermal imager from i3
« Reply #251 on: March 28, 2019, 07:18:14 pm »
I would say this is okay noise. Your scenes are not much of a temp Delta, and the majority of the image is very low T delta. It can be improved by capturing the "raw" data and using a third party program to render it out.

 
The following users thanked this post: wraper

Offline frenky

  • Supporter
  • ****
  • Posts: 1003
  • Country: si
    • Frenki.net
Re: Another smartphone thermal imager from i3
« Reply #252 on: March 28, 2019, 08:18:34 pm »
You should do a calibration against flat surface to remove some noise...
 
The following users thanked this post: Conure

Offline Conure

  • Contributor
  • Posts: 39
  • Country: se
Re: Another smartphone thermal imager from i3
« Reply #253 on: March 29, 2019, 05:15:48 pm »
You should do a calibration against flat surface to remove some noise...
Thanks it seems to reduce the noise a little when viewing but capturing pictures appear about the same, worse than what you see when you just look around.

Another thing, I took 2 close up pics of my leg from slightly different angles with the temperature span locked. As you can see the camera picture shows different color in each pic. Why is that? Should not my leg appear the same color from any angle if I lock the temperature span?
 

Offline Hyper_Spectral

  • Regular Contributor
  • *
  • Posts: 86
  • Country: us
Re: Another smartphone thermal imager from i3
« Reply #254 on: March 29, 2019, 06:06:59 pm »
Some of the other guys can elaborate on this, but you're changing the environment the sensor is viewing through to read the temperatures so yes it will change even with something minor like that.

 

Offline Vipitis

  • Frequent Contributor
  • **
  • Posts: 876
  • Country: de
  • aspiring thermal photography enthusiast
Re: Another smartphone thermal imager from i3
« Reply #255 on: March 29, 2019, 08:18:41 pm »
The sensors own temperature has a lot of influence. And after turning it on, you can wait up to 30 minutes to get the very best. The first 30 seconds can be a temp difference of 3° on some sensors.
 

Offline bugi

  • Regular Contributor
  • *
  • Posts: 249
  • Country: fi
  • Hobbyist using the ultra slow and unsure method
Re: Another smartphone thermal imager from i3
« Reply #256 on: April 04, 2019, 02:36:03 pm »
I have decompiled apk from android app to find out why images have so strong jpg artefacts. Well JPEG compression is hard coded to 90.  :palm:
Code: [Select]
source.compress(CompressFormat.JPEG, 90, outputStream2);
A bit late to the party, but afaik, that 90 parameter actually means it is pretty darn high quality.. assuming I interpreted the correct API doc the correct way:
https://developer.android.com/reference/android/graphics/Bitmap.html#compress(android.graphics.Bitmap.CompressFormat, int, java.io.OutputStream)

"Hint to the compressor, 0-100. 0 meaning compress for small size, 100 meaning compress for max quality."

If the parameter scales the same way as in typical image manipulation programs, 90 would be very high quality. In normal images/photos, I usually start to see issues only below 40-50 or so. The noise in thermal images might give it a hard time, though, but then again, the main content is often less demanding.  So, 90 should be good enough for viewable image.

And yet, I'd still use "raw" (when available) for processing further (as in any photography/imagery/DSP), simply because 95% of the software processing inside devices is often relatively speaking "garbage" (and that includes even things like DSLRs by Canon/Nikon, let alone small companies with less software/DSP expertise). i3 seems like a company which should know what they are doing with software even if they'd be small, but I guess they are just like rest, managing to code something that is only "enough to pass the course".
 

Offline nikoum

  • Contributor
  • Posts: 25
  • Country: gr
Re: Another smartphone thermal imager from i3
« Reply #257 on: April 22, 2019, 07:32:03 am »
Has anyone of the TE-Q1 plus owners ever used the Alarm function from the temperature sub-menu. Well according to the manual
page 42, when you choose the alarm function and you set a temperature for example 38c, then whenever an object with higher  than the set temperature (lets say 50c) is detected, then the object should appear in red and at the same time an alarm sound should be heard.

Well i have tried many times this alarm function but nothing happends. No red colour no alarm sound. All other choises from the temperature menu like point, min-max, line, rectangle, circle and emisivity work fine.
Can anybody confirm that this function works as described in the manual, which would mean that my camera has a problem, or it is not working on other cameras as well, and it might be a problem with the application ?
Thanks in advance   
 

Offline bugi

  • Regular Contributor
  • *
  • Posts: 249
  • Country: fi
  • Hobbyist using the ultra slow and unsure method
Re: Another smartphone thermal imager from i3
« Reply #258 on: April 22, 2019, 08:35:00 am »
Nope, couldn't get anything to happen with alarm. Also, the adjustment slider for the alarm temperature works in a horrible way.
 

Offline nikoum

  • Contributor
  • Posts: 25
  • Country: gr
Re: Another smartphone thermal imager from i3
« Reply #259 on: April 22, 2019, 08:39:08 am »
Thanks for replying. So if we can get some more confirmations from oter owners that nothing happends with the alarm, then we are almost sure it is the application fault. Waiting for more responses
« Last Edit: April 22, 2019, 11:31:59 am by nikoum »
 

Offline Vipitis

  • Frequent Contributor
  • **
  • Posts: 876
  • Country: de
  • aspiring thermal photography enthusiast
Re: Another smartphone thermal imager from i3
« Reply #260 on: April 22, 2019, 11:45:09 am »
It's almost certainly the application. The dongle cameras don't do very much other than give an image. The host does all the fancy stuff like palette, gain, temperatures as well as any kind of iso therms of alarms.
 

Offline bugi

  • Regular Contributor
  • *
  • Posts: 249
  • Country: fi
  • Hobbyist using the ultra slow and unsure method
Re: Another smartphone thermal imager from i3
« Reply #261 on: April 23, 2019, 07:52:23 pm »
Got the alarm function to work. It was just too nice one. After choosing it, and having the desired temperature set, draw a rectangle in the view; the highlight will work on that area and alarm can be heard.
 

Offline nikoum

  • Contributor
  • Posts: 25
  • Country: gr
Re: Another smartphone thermal imager from i3
« Reply #262 on: April 24, 2019, 05:27:50 am »
Bravo @bugi, i think this alarm function is a very nice feature. I also got it to work as you said by drawing a rectangle to cover the hole screen and it worked but with no sound. I tried a note4 phone and everything worked including sound but the phone is a bit slow in frame rate.
I just wish that somebody in this forum would come up with an improved android app for the te-q1 plus, like they have done for therm-app and xtherm cameras . I myself am willing to pay for such an app and i think there are many owners of te-q1 in this forum that would not mind paying for an improved app like ThermAppPlus or Thermviewer. I am thinking of starting a new thread on this. what do you think ?
 

Offline Ultrapurple

  • Super Contributor
  • ***
  • Posts: 1027
  • Country: gb
  • Just zis guy, you know?
    • Therm-App Users on Flickr
Re: Another smartphone thermal imager from i3
« Reply #263 on: April 24, 2019, 08:14:31 am »
Given the improvement that ThermViewer and ThermApp Plus give over the stock app I would expect something comparable for the TE series would bring similar benefits.

On the face of it, adapting ThermViewer would appear to be a good answer (AFAIK ThermApp Plus is highly dependant on the Opgal SDK but ThermViewer may be less so) but it rather depends on whether Jinhua can get hold of the TE SDK and, possibly, a camera to play with. The man's a genius - he  adapted ThermViewer for the 640x480 Pro without ever having seen one.
Rubber bands bridge the gap between WD40 and duct tape.
 

Offline nikoum

  • Contributor
  • Posts: 25
  • Country: gr
Re: Another smartphone thermal imager from i3
« Reply #264 on: April 24, 2019, 12:33:34 pm »
@Ultrapurple, i fully agree with your comment. Please see the new topic i just opened
https://www.eevblog.com/forum/thermal-imaging/express-your-interest-for-te-q1-improved-android-app-(possibly-thermviewer)/
 

Offline billyt

  • Contributor
  • Posts: 24
  • Country: al
Re: Another smartphone thermal imager from i3
« Reply #265 on: April 26, 2019, 05:29:10 pm »
A few days ago I got my TE-Q1. I find the pictures surprisingly noisy. Is this how it's supposed to look? I disabled the "enhance image" box. The phone is a galaxy s6.

I get noisy images too and also when just vieing through the phone screen it is more blury than i have seen from other cameras.    I was expecting better viewing results so i am thinking of selling it and try xtherm t3s with the thermviewer app. Any comments or sugestions
 

Offline bugi

  • Regular Contributor
  • *
  • Posts: 249
  • Country: fi
  • Hobbyist using the ultra slow and unsure method
Re: Another smartphone thermal imager from i3
« Reply #266 on: April 26, 2019, 07:47:22 pm »
A few days ago I got my TE-Q1. I find the pictures surprisingly noisy. Is this how it's supposed to look? I disabled the "enhance image" box. The phone is a galaxy s6.

I get noisy images too and also when just vieing through the phone screen it is more blury than i have seen from other cameras.    I was expecting better viewing results so i am thinking of selling it and try xtherm t3s with the thermviewer app. Any comments or sugestions
Maybe you already know these, but just in case...

After seen only the nice marketing cherry picks and the nice gallery photos here, and only a few "practical shots", I first thought a bit like that about my camera, too. But, a little bit of playing revealed, at least with/for my camera, that...

1) The depth of focus is (with these kind of lenses/camera) very narrow and the focus distance it came set with wasn't at the ranges I was trying at first. Thus, very blurry images. A bit of screwing the lens and sharpness got just fine (i.e. picking my eyebrows and individual hair on arms from 30cm away, as long as I can keep that distance stable).
2) The view on the phone is "digitally" zoomed, sort of. The native view is only 384x288 pixels, enlarged into something like 1440x1080 (x3.75 scaling, depends of course on the particular phone/display), so obviously it can not be as sharp as normal videos etc. (Although this one you probably knew, as you're comparing to other IR cameras as seen on phone).
3) Noise levels looked high at first, when looking at the initial "test subjects", e.g. a wall, or hand from close distance. As the AGC was on, it automatically expanded a very narrow thermal range to large brightness differences... and along with that sensitivity boost, it amplified to visible noise, too. After pointing to a view with larger temperature differences, the noise seemed to be closer to the level I expected. (I later played around with disabling or locking the AGC, but with weird results... I don't currently fully trust the AGC toggle.)
4) And finally, I realized that the "swap/recycle camera" icon is actually for offset re-calibration (temporary), and with that the remaining noise got reduced a lot. Though this one needs to be done quite often for best results. (EDIT: in the other thread you mentioned you had done calibration, so ignore this one.)

Granted, results still look like a phone camera trying to work in a dim room, and the TE-Q1 could have less noise, too, but this isn't a 1000-2000€ camera, and the official app apparently doesn't even try to do software based noise reduction (EDIT: it might do some NR after all) - which is actually better that way. (Best case would of course be to give a choice for how much and what kind of noise reduction to apply, but phones have limited processing power and such processing can also be done in post-processing, especially if more or less raw data is available.)

Of course, your camera could be faulty. It would help to see the "best" images you can get, when at optimal focus, with targets with good contrast and temperature differences (say, something like a warm piece of electronics (modem/router/switch, insides of PC), hot water), and just after been calibrated.
« Last Edit: May 01, 2019, 12:52:24 pm by bugi »
 
The following users thanked this post: nikoum

Offline mahony

  • Regular Contributor
  • *
  • Posts: 156
  • Country: de
Re: Another smartphone thermal imager from i3
« Reply #267 on: April 28, 2019, 06:41:59 pm »
As I cannot attach any files to PNs and this might be of interest to some others:
Attached is a more recent Android SDK for the TE-Q1 from 2018. Probably not the most recent version but maybe this helps some of you …

I think I have already seens a version with M1 support but not sure where I put it.  :palm:

 

Offline bugi

  • Regular Contributor
  • *
  • Posts: 249
  • Country: fi
  • Hobbyist using the ultra slow and unsure method
Re: Another smartphone thermal imager from i3
« Reply #268 on: April 28, 2019, 08:01:31 pm »
As I cannot attach any files to PNs and this might be of interest to some others:
Attached is a more recent Android SDK for the TE-Q1 from 2018. Probably not the most recent version but maybe this helps some of you …

I think I have already seens a version with M1 support but not sure where I put it.  :palm:
Seems nikoum already linked that same SDK version in this thread https://www.eevblog.com/forum/thermal-imaging/express-your-interst-for-te-q1-improved-android-app-(possibly-thermviewer)/msg2365212/#msg2365212

Considering how much one should or should not copy directly from the SDK (if rewriting that level stuff), even older versions provide the most important things (how to communicate via USB, and how to interpret the incoming data). The newer versions have one new nice bit of info, how to initiate re-reading the flash. Also some obvious improvements like how it calculates FPA temperature estimate, splitting things to multiple threads, ... They might have something else nice, too, but I haven't had the time to fully read through and decipher the newest version; most of what I've read so far are the same as in the code couple years ago (perhaps few unused members removed, and few unused members added :P)

If using the SDK directly as is, at some point they added at least one significant optimization; option to do certain calculations with a method that uses precalculated tables. But still including lots of silly stuff, and if their app uses the same calcs/SDK within it, the app still eating battery like crazy seems to indicate there would be room for improvement (though I don't know if the USB device current is added for the app's consumption statistics).

As for the M1 support, the Android SDK for Q1 ver 181027 doesn't seem to know about M1, either, unless it is indeed the same sensor, just windowed somehow; they seem to now provide the mysterious "recog" values through the API, instead of ignoring them inside the SDK.  The Windows SDK does seem to have some knowledge about the different cameras, but I've only taken a short peek at that with left eye.
 
The following users thanked this post: nikoum

Offline billyt

  • Contributor
  • Posts: 24
  • Country: al
Re: Another smartphone thermal imager from i3
« Reply #269 on: April 30, 2019, 01:48:45 am »

[/quote]
Maybe you already know these, but just in case.........

.........results still look like a phone camera trying to work in a dim room, and the TE-Q1 could have less noise, too, but this isn't a 1000-2000€ camera, and the official app apparently doesn't even try to do software based noise reduction - which is actually better that way.
[/quote]

Bugi thank you very much for your suggestions which i followed but as you and others have noted the stock android app has very limited abilities. You have described it beautifully in the above quote. I have read some of your posts here and in the other relevant threads about te q1 and you seem very knowlegable and capable.
So why dont you (even with the help of other capable members) undertake the project of modyfying the stock app with improvements such as the ones you already planned and also noise reduction, clearer view, etc. Judging from the interest shown in the other thread about improved app for te q1, i would say you will find quite a few supporters (myself included) if you come up with a modified app.

 

Offline bugi

  • Regular Contributor
  • *
  • Posts: 249
  • Country: fi
  • Hobbyist using the ultra slow and unsure method
Re: Another smartphone thermal imager from i3
« Reply #270 on: April 30, 2019, 08:34:44 am »
Quote
.........results still look like a phone camera trying to work in a dim room, and the TE-Q1 could have less noise, too, but this isn't a 1000-2000€ camera, and the official app apparently doesn't even try to do software based noise reduction - which is actually better that way.
So why dont you (even with the help of other capable members) undertake the project of modyfying the stock app with improvements such as the ones you already planned and also noise reduction, clearer view, etc.
Some already have sort of undertaken it, although not modifying original, but writing completely new one - copyrights and stuff.
But as these kind of projects are not our paid work, things progress only when we get enough time to do it and have no higher priority and/or more interesting stuff to do. For example, in my case, I'll have enough time to do more work on the "driver" side perhaps next summer, and I estimate that I won't be able to do much more than that "driver" side. (Which will aim at doing about the same as the current official code, just with less CPU use and hopefully few things in a slightly better way.) The UI etc. other parts of a full app would be many times larger project, so I'll probably just slap the SDK example for that and fine-tune it for my development needs.  Or perhaps co-operate with others who are doing full app if my driver could be useful for their better done rest of the app, or something.

Note how I mentioned that "... is actually better that way" - the app not trying to do noise reduction. (EDIT: the app might actually do some noise reduction, after all.) The thing is, mobile phones, even the flagships, have relatively limited processing power to do these things (especially when not using the DSP/GPU parts to help), so the result of such on-phone noise reduction is poor quality output and/or reduced frame rate. Note how many even good phones are already having troubles with the app in its current state, while not even wasting CPU power on noise reduction.

Also, most common (and light-weight) noise reduction algorithms are always compromises; they reduce noise, but also some actual information from the image (usually small details and edge contrast). Thus, it is best to consider the phone+app as the viewfinder, record whatever the camera gives, then post-process e.g. on a PC with a lot more processing power and time, better display to look at the results, ability to adjust processing settings and redo it if needed, etc.

I'm not saying that an app must not do any noise reduction; it could be an optional thing, or a specialized algorithm that works only on IR camera specific issues (I'm planning to try one or two tricks like that), but in general, the benefits are small.

My #1 change towards having less noise would be recording binary "raw" frames (with or without calibration), instead of current either 8-bit already visualized output or CSV. (CSV?!? whyyyy.. CSV was meant as human-readable and/or easy to import to spreadsheets, neither of which are very relevant for IR images.)  The raw data would allow better post-processing, and binary format is much more efficient than CSV for writing and reading.  As a compromise format, some sort of grayscale image format might be able to transfer high resolution "raw" values efficiently while still being viewable with standard programs, but I haven't studied this possibility yet.
« Last Edit: May 01, 2019, 12:51:07 pm by bugi »
 

Offline Conure

  • Contributor
  • Posts: 39
  • Country: se
Re: Another smartphone thermal imager from i3
« Reply #271 on: May 01, 2019, 11:56:36 am »
TE-Q1
Can anyone explain why the colors are changing here? The temperature ranges are locked yet the colors change a lot depending on whether the masonry heater is in the picture or not.
« Last Edit: May 01, 2019, 11:58:07 am by Conure »
 

Offline bugi

  • Regular Contributor
  • *
  • Posts: 249
  • Country: fi
  • Hobbyist using the ultra slow and unsure method
Re: Another smartphone thermal imager from i3
« Reply #272 on: May 01, 2019, 12:43:43 pm »
TE-Q1
Can anyone explain why the colors are changing here? The temperature ranges are locked yet the colors change a lot depending on whether the masonry heater is in the picture or not.
I noticed the same merely by viewing a wall and moving my hand in and out of the view. The wall got dark when hand came into view. That is why I wrote somewhere here that I don't quite trust that AGC toggle or the range locks. I have been reading the SDK codes, but I haven't been able to dig into the AGC stuff deep enough... and the app could do things differently than the SDK.

EDIT: After a bit more digging into the frame processing code, the SDK seems to always do AGC effect. There is a part in an if about AGC disabled, but it seems to only do some temperature-based limiting, and does not affect the scaling. It also counts how many pixels were clipped on each of min and max side, but then does not use those counts.  If the app uses that SDK, that would explain...
« Last Edit: May 02, 2019, 10:07:16 pm by bugi »
 

Offline bugi

  • Regular Contributor
  • *
  • Posts: 249
  • Country: fi
  • Hobbyist using the ultra slow and unsure method
Re: Another smartphone thermal imager from i3
« Reply #273 on: May 01, 2019, 12:50:16 pm »
My earlier "the official app apparently doesn't even try to do software based noise reduction" may have been incorrect. Older SDK versions didn't do it at all, and reading the newer SDK code, I first thought certain pieces of code were part of edge enhancement, but after analyzing the code better, it seems to be actually doing some effort of noise filtering.  Note, that is for SDK, the app might still not do it, or could do it in a different way.
 

Offline Conure

  • Contributor
  • Posts: 39
  • Country: se
Re: Another smartphone thermal imager from i3
« Reply #274 on: May 03, 2019, 07:53:24 pm »
TE-Q1
Can anyone explain why the colors are changing here? The temperature ranges are locked yet the colors change a lot depending on whether the masonry heater is in the picture or not.
I noticed the same merely by viewing a wall and moving my hand in and out of the view. The wall got dark when hand came into view.
Indeed.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf