Flir E4 Thermal imaging camera teardown

[daves]

With 16 Bit convert (Q16) we can simple stretch the brightness from 16*256=4096 to 238*156=60928

Good job !

[eyesky]

Hi Chaps,
Sorry for this basic....... falling at the first hurdle,   but where do I find E8.cfg ?!

[mikeselectricstuff]

Good day, friends! I just discovered this awesome thread!
Please tell, is there a hack for latest Flir E4 HW model (1.1) with
latest firmware (1.2something), or I am too late to the party? 

There is no credible evidence that any version later than 1.19.8 exists. 

[mrflibble]

Hi Chaps,
Sorry for this basic....... falling at the first hurdle,   but where do I find E8.cfg ?!

See first post for latest instructions: https://www.eevblog.com/forum/testgear/flir-e4-thermal-imaging-camera-teardown/

[eyesky]

Thanks, but that's where I read about needing the e8.config........
begins.... OK, so this is it.....
Seems to not be bothered about 7zip - I just used Windows "send to compressed folder"

Open E8.cfg in a PC text editor (not anything that will mangle linefeeds etc.!)    where's that from 

Hi Chaps,
Sorry for this basic....... falling at the first hurdle,   but where do I find E8.cfg ?!

See first post for latest instructions: https://www.eevblog.com/forum/testgear/flir-e4-thermal-imaging-camera-teardown/

[mikeselectricstuff]

Thanks, but that's where I read about needing the e8.config........
begins.... OK, so this is it.....
Seems to not be bothered about 7zip - I just used Windows "send to compressed folder"

Open E8.cfg in a PC text editor (not anything that will mangle linefeeds etc.!)    where's that from 

Hi Chaps,
Sorry for this basic....... falling at the first hurdle,   but where do I find E8.cfg ?!

See first post for latest instructions: https://www.eevblog.com/forum/testgear/flir-e4-thermal-imaging-camera-teardown/

It's in the .zip attatched to the hack post linked from the start of this thread

[eyesky]

Thanks Mike! 

[Wizzard]

@ Aurora - The Cadillac / Raytheon imager is 320x240, but has a rotary mechanical chopper and is most certainly not 30Hz. It also suffers from very significant ghosting / hot+cold outlining, and the poor quality of analog NTSC video.

[Fraser]

I have seen one working and beg to differ on frame rate:

http://www.computer.org/csdl/mags/cg/1999/05/mcg1999050006.html






May be the unit you tried is suffering from its age ?

Yes it uses a shutter wheel . That is because it is a BST FPA. Chopper wheels are not, in themselves, a serious issue. they have been used for years and just create a slightly softer edge to an image. I have several fire fighting TICs that use this technology and work very well indeed. The issue can be reliability of the motor driving the wheel. The major advantage of the BST+Chopper wheel over the older micro-bolometer designs was the continuous image with no NUC shutter freezes.

IIRC the Cadillac thermal camera used a 160 'line' field x2 interlaced to create 320 horizontal 'lines', and a frame rate of 30fps was specified

If poor performance is noted, the following should be checked:

1. If the lens protector is fitted, is it in good clean condition.
2. Remove the lens protector (if fitted) and check the condition of the primary lens. Corrosion and stone impact damage will severely degrade the image. Check for a build up of dirt on the lens.
3. Listen to the unit running, does the chopper wheel motor spin up quickly and does it 'purr' like a cat ? If not it is likely in need of a service.
4. It is a good idea to service the chopper wheel motor as the bearings dry out over time. This adversely effects the image and will eventually lead to complete failure due to motor bearing failure. I have seen chopper wheels that cannot be turned due to seized bearings.
5. Be aware that a BST FPA needs a chopper wheel to maintain a change in the scene, as required by the detector elements (they only 'see' change). If a chopper wheel fails and the shutter is not obscuring the FPA, an image will still be visible but only thermal targets that move will be displayed. Early TICs allowed the user to switch off the chopper wheel to avoid image blur whilst panning a camera quickly.

Also a test here:

http://trydeal.com/nightvision/

The first video showing the install in the car looks to produce a more than acceptable image for a car mounted system.



You tube has a few videos of these cameras working. They are certainly not 9fps or other low frame rates. Image quality comes down to the condition of the camera and ability of the installer/user. Many of these cameras suffered lens corrosion and damage due to failure of the lens protectors over the years. Good ones give excellent images.

As to the quality of analogue NTSC....well considering this is a monochrome camera system running at 320x240 resolution the issues of analogue video and the vagaries of NTSC colour reproduction are pretty mute. This is not an SVGA platform and was designed for simple implementation in a car. Who wants a computer running just to show a thermal image ? Digital has its place but sometimes analogue makes better sense....i.e. the KISS principle. If you want to see the nightmare that digital brings to the party, take a look at an AUDI or BMW night vision system....I doubt you will get it running without it being fitted in the original car with the original ECU's and dashboard computer.

[iDevice]

I have taken some sample pictures for you to look at. The titles are the descriptors but here is more info.

Conc-T is the concave side facing the target

This what I have and it confirms exactly what I was writing about.

ConV-T is the convex side facing the target

Much less edge distorsion, but at the cost of a slight defocus indeed.

100mm shows the image produced using the 100mm FL lens

This is of great help to me, I know what will be my next purchase from China 

I thank you Aurora.

[Fraser]

After repeating my tests of the 50mm lens I have come to the conclusion that my orientation of the lens with the concave side facing the target is compromised by the E4 application. As such I have changed my view on which way to install the lens. I will be changing my 50mm lens to have the convex side facing the target. The defocussing isn't serious enough to concern me and the edge distortion is significantly reduced.

The 100mm lens is bi-convex and therefore has no preferred orientation  Nice and simple.

[OrBy]

here is the zoom enable instruction:

1. enable zoom in FlashFS\system\appcore.d\config.d\e8.cfg (from E4 hack): .caps.config.image.zoom.enabled bool true , don't forget to renew CRC01!
2. edit menu part in FlashBFS\system\ui.d\toolbar-config_z3.xml (from menu hack) as shown:

      <ToolBar name="zoomMenu">
         <ToolBar name="zoom_1x"/>
         <ToolBar name="zoom_2x"/>
         <ToolBar name="zoom_4x"/>
         <ToolBar name="zoom_8x"/>
      </ToolBar>

that's it

Works like a charm! Thanks 

[david7586]

Good day, friends! I just discovered this awesome thread!
Please tell, is there a hack for latest Flir E4 HW model (1.1) with
latest firmware (1.2something), or I am too late to the party? 

As others have said, you won't know until you've tried. Please do let us know how everything turns out for you.

If you don't mind me asking, where did you buy your unit from?


Sent from my iPhone using Tapatalk 2

[larky]

First try at panorama, converting to tiff, resize/sharpen and stitching went fine.
I get an error when running  the splitjpg.php

N:\flirscript>php.exe.lnk N:/flirscript/splitjpg.php -i N:/flirscript/FLIR0172.j
pg -r N:/flirscript/stitch.tiff -o N:/flirscript/pano1
.fffrebuilt FFF from 3 parts
split FFF file to segments
FLIR Record 0x20, offset 0x0200, length 0x09ac
FLIR Record 0x22, offset 0x0bac, length 0x0310
FLIR Record 0x21, offset 0x0ebc, length 0x0038
FLIR Record 0x01, offset 0x0ef4, length 0xb5bf
FLIR Record 0x2a, offset 0xc4b4, length 0x0060
FLIR Record 0x0e, offset 0xc514, length 0xa007
convert.exe: Unknown field with tag 18248 (0x4748) encountered. `TIFFReadDirecto
ry' @ warning/tiff.c/TIFFWarnings/847.
new image w/h=1098/357
write new radiometric jpg
part 0: 65534 Byte
part 1: 65534 Byte
part 2: 65534 Byte
part 3: 65534 Byte
part 4: 65534 Byte
part 5: 65534 Byte
part 6: 65534 Byte
part 7: 65534 Byte
part 8: 65534 Byte
part 9: 65534 Byte
part 10: 65534 Byte
part 11: 65534 Byte
part 12: 714 Byte

N:\flirscript>

N:\flirscript>convert.exe.lnk --version
Version: ImageMagick 6.8.8-0 Q16 x64 2013-12-21 http://www.imagemagick.org
Copyright: Copyright (C) 1999-2014 ImageMagick Studio LLC
Features: DPC Modules OpenMP
Delegates: bzlib cairo freetype jbig jng jp2 jpeg lcms lqr pangocairo png ps rsv
g tiff webp xml zlib

The resulting pano1.jpg now is the same as the FLIR0172.jpg.

[mrflibble]

I get an error when running  the splitjpg.php

What error? All I see is a warning that can be safely ignored.

convert.exe: Unknown field with tag 18248 (0x4748) encountered. `TIFFReadDirectory' @ warning/tiff.c/TIFFWarnings/847.

It still generates the correct output, right?

@tomas123:
It might be a good idea to add a -quiet option to your convert command line to prevent user confuzzlement.

[mikeselectricstuff]

here is the zoom enable instruction:

1. enable zoom in FlashFS\system\appcore.d\config.d\e8.cfg (from E4 hack): .caps.config.image.zoom.enabled bool true , don't forget to renew CRC01!
2. edit menu part in FlashBFS\system\ui.d\toolbar-config_z3.xml (from menu hack) as shown:

      <ToolBar name="zoomMenu">
         <ToolBar name="zoom_1x"/>
         <ToolBar name="zoom_2x"/>
         <ToolBar name="zoom_4x"/>
         <ToolBar name="zoom_8x"/>
      </ToolBar>

that's it

Works like a charm! Thanks 

I tried it but no obvious difference - was in a hurry so may have missed something. What FW version do you have?

[tomas123]

@mrflibble
You are right, the script it's only a privat hack and I love warnings

First try at panorama, converting to tiff, resize/sharpen and stitching went fine.
I get an error when running  the splitjpg.php
...
The resulting pano1.jpg now is the same as the FLIR0172.jpg.

The same size?
Open the large pano1.jpg with FlirTools and enjoy ...

[OrBy]

I tried it but no obvious difference - was in a hurry so may have missed something. What FW version do you have?

Recheck your steps. (edit the e8.cfg, re-crc/fif it, upload it, replace the section in toolbar-config_z3.xml, upload it, then cold boot the camera). Whole new menu comes on the main toolbar. It's only digital zoom so it works by reducing the resolution but it works none the less.

Running 1.18.7.

[Fraser]

Regarding the digital zoom. IIRC my PM695 interpolates and processes the zoomed image to improve its appearance and then presents it as a 320x240 image. I have not added zoom to my E4 yet but will definitely do so. I wonder whether the E4 will do any 'clever' image enhancement when in zoom mode. Having the ability to narrow the field of view at the expense of resolution is still a useful function to have. Even if only X2 is used (160x120 pixels), it's a bargain when compared to the cost of a X2 Germanium lens group 

The zoom may also be useful when using a close-up auxiliary lens for PCB work 

[mrflibble]

I wonder whether the E4 will do any 'clever' image enhancement when in zoom mode. Having the ability to narrow the field of view at the expense of resolution is still a useful function to have. Even if only X2 is used (160x120 pixels), it's a bargain when compared to the cost of a X2 Germanium lens group 

I didn't notice any really clever image enhancements in zoom mode. But it did look quite adequate for a digital zoom. I only ever used zoom scripted over telnet though, so no idea about the menu based zoom being discussed here.

Does the menu allow you to zoom in on something off-center? Although maybe not, since that is less of an issue during handheld operation.

Anyways, using zoom and grabbing the video stream gives a better result than grabbing the video and doing your own zoom. Presumably because by the time the video stream get to you there are all manner of dropped bits on the floor. Now if only we could get a raw 16-bit stream. Even at 9 fps that would be quite useful IMO.

[iDevice]

After repeating my tests of the 50mm lens I have come to the conclusion that my orientation of the lens with the concave side facing the target is compromised by the E4 application. As such I have changed my view on which way to install the lens. I will be changing my 50mm lens to have the convex side facing the target. The defocussing isn't serious enough to concern me and the edge distortion is significantly reduced.

I had time this morning to reverse my lens and the picture is much better now with the convex side towards target.
Unlike you, I didn't notice any degradation in sharpness, so for me it's a no brainer, it stays like that.

The 100mm lens is bi-convex and therefore has no preferred orientation  Nice and simple.

I ordered the 100mm today as well so I will report back when it arrives.

[larky]

I was stupid not to open the pano1.jpg in flir tools, even if it looks wrong in windows preview, it is bigger after all.

However, the temperature scale is all wrong when imported to flir tools, the input images are all in about 15-28°C range, the result is in the -36-289°C range with auto scale.

[Wizzard]

@ Aurora - I recall reading 15hz, but the motion on my video I took seems certainly better than 9hz I could be mistaken in any case.



I did enjoy the unit, but the lens was poor for normal work (but great for infinity focus!), and the 'glow' of hot and cold objects was equally poor. But still, extremely useful.

Do you work with these Raytheon models often? I have another unit I am at a loss for checking the pin-out on - It's similar to the Cadillac one, but has an extra pin, and all the pins vanish into the PCB... I can work with an unlabeled but traceable circuit, but this one's out of my realm.

[tomas123]

I was stupid not to open the pano1.jpg in flir tools, even if it looks wrong in windows preview, it is bigger after all.

However, the temperature scale is all wrong when imported to flir tools, the input images are all in about 15-28°C range, the result is in the -36-289°C range with auto scale.

sorry larky, my todos are not end user proved
amazing, that you have succesfull stitched a panorama in reverse byte order

as I wrote here, the Flir E4 use PNG with reversed byte order for embedded RAW images
https://www.eevblog.com/forum/testgear/flir-e4-thermal-imaging-camera-teardown/msg343791/#msg343791

I extended the tutorial for E4
read https://www.eevblog.com/forum/testgear/flir-e4-thermal-imaging-camera-teardown/msg348715/#msg348715
please test it

[larky]

I get the first step done on windows, to old exiftool on linux.
N:\flirscript>exiftool -b -RawThermalImage -if "$RawThermalImageType eq 'PNG'" FLIR*.jpg -w %f.png
and I get a collection of very ugly .png images. Great

For the next step I installed ImageMagick 6.8.8 Q16 on linux;
for i in *.png ; do convert $i gray:- | convert -size 320x240 -depth 16 -endian msb gray:- _$i.tif ; done
This leaves some me a bunch of .tif images that I can not open, but don't let that stop me;
convert *.tif -resize 480x -sharpen 0x1 a_%03d.png

convert: no decode delegate for this image format `_FLIR0172.png.tif' @ error/constitute.c/ReadImage/555.   

Something is wrong in the byte order switch thing