Author Topic: Flir E4 Thermal imaging camera teardown (Read 3789756 times)

Solare · « **Reply #7075 on:** May 31, 2015, 03:05:46 pm »

@High Voltage,

most likely, in this way mine picture will be even better, than yours

you see, I have done lots of modifications to try for compensate for low image sensor quality of my camera, so I so to say... did some overkill

For you, there is no need for any additional improvements.

But if you really want for fair comparison, you should not use nice cases with high temperature differences, but should compare hard cases i.e., minimum temperature span, 2C, isothermal environment (room with no heaters, air leaks or air conditioner, ideally wooden walls, closed for at least 1 day for allow temperature to set)? lowest possible temperature gradients of the surroundings. Then and only then, you can qualify noise levels and camera performance.

To measure, charge your camera, allow for at least 20 minutes to warm up the unit, THEN enter the room and make multiple shots with Black and White palette enabled.

Solare · « **Reply #7076 on:** May 31, 2015, 08:20:56 pm »

@tomas123, gemrough

nice, very nice!

version numbers? HW-FW-SW?

More reports, please! The picture we are getting becomes curiousier and curiousier (c) Alice

Solare · « **Reply #7077 on:** May 31, 2015, 08:24:53 pm »

@All users of Janekivi Advanced Update Menu with 2.3.0 FW (original),

Can you actually change MSX parallax setting while in EDIT mode?
No..no... of course you can

Rephrasing...

Can you change MSX parallax setting AND HAVE THIS NEW SETTING APPLIED and in effect? Have you tried this literally?

Thank you!

OrBy · « **Reply #7078 on:** May 31, 2015, 08:34:33 pm »

Not that I am aware of - I never knew that was even an option on any version. I have changed the MSX offset in post using BFIC however.

Solare · « **Reply #7079 on:** May 31, 2015, 09:50:52 pm »

@All users of Janekivi Advanced Update Menu with 2.3.0 FW (original)

I failed to make any use of the alarm bound to Delta, which is set between any objects!

It allows to set the 'above' and 'below' and any value, but 'above' never triggers, while 'below' triggers always!

Do you have issues like that?

Thank you!

Solare · « **Reply #7080 on:** June 01, 2015, 12:42:13 am »

@Bud

I have found this link http://wikisend.com/download/544016/test.avi you've sent some time ago, but I assume the link is broken, because I couldn't make it to download a file.

Could you please, describe what is 'programmable button' and 'custom popup menu'?

Thank you!

Solare · « **Reply #7081 on:** June 01, 2015, 01:39:39 am »

@All

The size of encrypted conf.cfc file is exactly 6.608 bytes.

If I add something to the file, its size increases.

Can anyone tell me if the size of this file allowed to be changed?

Or this will brick it?

Thank you!

Bud · « **Reply #7082 on:** June 01, 2015, 03:54:29 am »

Quote from: Solare on June 01, 2015, 12:42:13 am

Could you please, describe what is 'programmable button' and 'custom popup menu'?

Check the link in my reply #7042, i believe it was the same video. It goes through programmable button feature and i think at the end it also shows a small popup menu, which is just a shortcut to some features otherwise accessible via main menu.

alank2 · « **Reply #7083 on:** June 01, 2015, 01:32:42 pm »

Solare, what numbers have you specifically tweaked in which files to improve your sensor quality?

Bud · « **Reply #7084 on:** June 01, 2015, 01:56:44 pm »

Quote from: Solare on June 01, 2015, 01:39:39 am

The size of encrypted conf.cfc file is exactly 6.608 ...
Can anyone tell me if the size of this file allowed to be changed?

There is a file length indicator written in the trailer of the file. Therefore when patching encrypted file it is important to not change the file size. But if you know how to decrypt and reencrypt the file, you can add/remove content in plane text and encryption procedure will update the file length field as required.

Solare · « **Reply #7085 on:** June 01, 2015, 02:47:42 pm »

@alank2,

There is no way 'to improve the sensor quality'. The sensor's quality is what you get when you buy it. This is lottery, no less.

Also, since actual digitizing depth of the E4 is 14 bit and full ADC span is from -40 to 282 which gives 20 mK magnitude for the LSB, these 'modifications' are only of help, when own noise levels of the sensor is above 20 mk. Your sensor is well below this value, so there will not be any LSB jitter hence, either any lower bit noise.

Furthermore, when applying modifications I did with my sensor (26.8 mK own noise), one trades speed for quality. Overall dynamics and visual when panning the E4 across the view, degrades therefore. Finally, the modifications involve calibration files changes which most user wouldn't want, at least without firm understanding of physics, which is not trivial and cannot be just 'copied' from one device to another, like t. ex, menu modifications.

So there is no 'magic way', sorry. What you have now cannot be improved.

alank2 · « **Reply #7086 on:** June 01, 2015, 02:50:14 pm »

I get what you are saying Solare, I'm just curious what changes you made and what were the effects. I've been following this thread, but didn't see/get the specifics...

Solare · « **Reply #7087 on:** June 01, 2015, 02:54:05 pm »

@alank2,

I think I have posted my progress in recent posts, so just take your time and read back for details,

but for this to work, you need to have poor sensor, with noise well above the LSB level. Otherwise, you just degrade the visual appearance and dynamics.

Solare · « **Reply #7088 on:** June 01, 2015, 02:59:11 pm »

@Bud,

have you actually received my PM?

Bud · « **Reply #7089 on:** June 01, 2015, 03:43:45 pm »

Yes i will reply after work

mimmus78 · « **Reply #7090 on:** June 02, 2015, 10:09:49 pm »

We will soon have cheaper IR cameras?

http://m.phys.org/_news352479907.html

Fraser · « **Reply #7091 on:** June 02, 2015, 10:20:05 pm »

An interesting article.

It has previously been suggested on this forum that a DLP Projector DMD type mirror array may offer such potential. At the time we did not think it would work due to the materials used and the self heating of the DMD array.

This research seems to take the DLP projector DMD and redesigns it for use at thermal cameras wavelengths. Let us hope that this leads to high resolution thermal imaging at affordable prices.

It is interesting to note that the basic process of image collection is not new. The Inframetrics Thermasnap uses a Hewlett Packard designed mechanical image capture engine that moves a 120 pixel linear thermal detector array across the projected image field of the thermal lens, so producing an image built up of single pixel wide vertical strips (120 IIRC = 120x120 pixel image), albeit at a comparatively slow speed of 2 seconds per image !

http://www.flir.com/legacy/view/?id=51641

I own one of these Thermasnap cameras and must say the images that it produces are pretty good considering the mechanical nature of the image capture engine.

This new development removes the need to move the detector array and is somewhat faster at capturing pictures

Aurora

DMD = Digital Micromirror Device

http://en.wikipedia.org/wiki/Digital_micromirror_device

frenky · « **Reply #7092 on:** June 03, 2015, 06:23:52 am »

Some docs:
http://arxiv.org/pdf/1503.04267v1.pdf
http://www.cv-foundation.org/openaccess/content_cvpr_2015/papers/Chen_FPA-CS_Focal_Plane_2015_CVPR_paper.pdf

Solare · « **Reply #7093 on:** June 03, 2015, 03:19:12 pm »

@All

SOS SOS SOS!

Okay people, I ask for your urgent help and sympathy!

I was just working with snapshots organizing (removing unnecessary files), when my device halted... after reboot, I saw lots of spatial color noise (much like on un-calibrated sensor), and it did not disappear after calibration, or multiple reboots, or even the battery pull offs.

Have anyone of you experienced the like issue? If so, what could be the problem and how you corrected this?

I personally believe that maybe somehow FS gets damaged so system cannot get access to calibration data of the sensor and hence, uses the default values. Because all other features work perfectly.

Really, people, what do we know about how FS mounts and what it does if it gets corrupted? Is there any 'disk check' or partition recovery procedure we could perform in such the case?

Situations like halts (out memory, appcore crash, reboot without saving etc., error during editing a snapshot, may appear routeenly when using device. So, sooner or later, FS gets corrupted and what then?

Thank you for your support!

UPDATE:

Situation resolved before I have time to panic. But the issue is not gone I think, so the question I addressed to you still there!

I was working with the Charger connected, so the device became quite hot with closed shutter, 53.5C according the web-service temperature log (assuming it had opportunity to add the log!)

How overheating may have resulted in this malfunction?

I just let it to stay still for ten minutes, and the first reboot did well with calibration etc.

Your honorable opinion will be?

Bud · « **Reply #7094 on:** June 03, 2015, 03:36:13 pm »

Pull the battery out and leave it overnight. If that wont help then can try reflashing the firmware. Do you still have access to the file system?

Solare · « **Reply #7095 on:** June 03, 2015, 04:19:56 pm »

Thank you, Bud!

I think I have added the update?

While it was hot, no matter how I rebooted or pulled the battery out, id didn't help.

When I let the device to cool for 10 minutes, the usual boot even without removing battery before that, fixed the problem.

The question is, why this happened, and what the E4 system will do if the FS system gets corrupted due to some appcore crash in improper time?

I know there are utilities for WinCE 6.0 and above, which check the file system, fat etc., and allow for fixing the errors found.
But how do I run this application as there is no application install service enabled? Even without installation, how do I run it remotely?

P.S>

Patience is Value!

But it is so hard to wait till the weekend!

Solare · « **Reply #7096 on:** June 03, 2015, 04:23:24 pm »

Yes, I still have access to the FS by RNDIS mode.

I never had re-flashed the device. How do I do that? Will the re-flashing return the device info factory default mode and replace (possibly corrupted) disk image, FAT etc.? I assume I will have to re-hack from the very beginning then!

Bud · « **Reply #7097 on:** June 03, 2015, 04:50:02 pm »

Regarding running applications on the camera you can do it via i believe START telnet command, providing the processor can run the code. Information on the processor type and version on E4 was somewhere in this thread. To navigate applications you typically need a VNC server to run on the camera and VNC client on your PC. You need remote desktop because the camera does not have full keyboard. There was a link published in this thread on where to get VNC that works. I confirm it works, i tried it. The other application that worked was a Registry Editor. I used that VNC to look through E4 Registry, it was fun.

Solare · « **Reply #7098 on:** June 03, 2015, 06:23:39 pm »

@Bud,

Are you certain this info was actually published in this Thread, or not removed afterwards?

I did some 30 min search for keywords like "remote access, vnc, server etc." and found nothing of use.

OrBy · « **Reply #7099 on:** June 03, 2015, 06:50:07 pm »

FYI 2.3.0 firmware has locked down the ability to remotely install wince apps. You can drop in and run exe's like VNC still I think but cabs for the reg editor and file explorer cannot be loaded easily anymore.


EEVblog Main Site	EEVblog on Youtube	EEVblog on Twitter	EEVblog on Facebook	EEVblog on Odysee

Poll

Has the hackabiliy of the E4 made you buy one :

Author Topic: Flir E4 Thermal imaging camera teardown (Read 3789756 times)

Share me