EEVblog Electronics Community Forum
Products => Thermal Imaging => Topic started by: Logan on December 03, 2021, 02:11:20 pm
-
Hi guys.
Since my thermal hobby actually began from night vision, I'll post it here. However mods can move this if you feel off-topic.
To avoid advertising, I'll do my best to hide the brand/model. This is a pure technology talk.
I just got a unique night vision device that is not IIT(tube)-based, but based on "black silicon" CMOS.
https://en.wikipedia.org/wiki/Black_silicon
Before this, all so-called "digital night vision" are like jokes after jokes. But just look at the pictures attached, this device is truly amazing. (Pictures are from internet because I haven't got a chance to take decent photos yet, however I can assure you these pictures are from the same model I'm talking about)
It have a sensitivity similar to Gen-2 tube(or ISO 200k), and with color. It will not be damaged by daylight, or have a limited life-time like IITs. And the resolution is 720p, similar to Gen-3.
More surprisingly, its spectrum sensitivity goes all ways into SWIR. :box:
What more would I ask for a device much less than $1k?
Oh wait, actually I do want one more function: wired video signal output. Because the EVF is not always comfortable to use, and the built-in wireless transfer never work.
One thing I just thought, is they should use a custom CFA, not the conventional bayer filter:
1. CMY color-resist would be a better choice instead of RGB, because they allow 2x light through.
2. The filter block rate should be as low as possible, just enough to let the processing stage to guess the color.
3. There should be half or more pixels just without any colored layer.
4. All color-resist should be transparent to IR.
Since they already go so far to "Black Silicon CMOS", I don't see a reason why not do this to make it better.
Finally, this is not the "X27" scam from the scammy "SPI corp" shit! :-DD
Thanks for reading.
-
The BBC has been using very low light colour cameras for wildlife imaging work recently. I wonder if those cameras are using this technology. It was described as groundbreaking colour imaging technology as it provides good clean images when most colour cameras would not even provide an image, let alone a colour balanced image !
Interesting stuff and no problem detailing it in this part of the forum. Discussions cover IR to LWIR and THz imaging here :)
Fraser
-
I have a Si*nyx Aur*ra sport. Favorite toy during power failures, boating, checking on things that go bump in the night, working in the laser lab, etc. While I feel a black and white version would be awesome, color does add some unique aspects. Near IR sensitivity which I have tested to to 1084 nm is awesome.
Optics are quite decent and well designed.
Questions not requiring an ITT-Excelis grade NOD test bench will be answered...
One caveat, if I'm going into combat, or playing paintball, a tube device designed to purpose would still be my preference. Aur*ra does have a minimal inverted video mode and a long eye relief for helmet mounting. That is based purely on the S*ort model. Not sure about the LE versions as I do not have access to them.
I will not do a tear-down, I love it too much.
Ask away...
Steve
-
Thank you Fraser, you are very kind.
I think BBC maybe using something like Canon ME20F-SH. That's even more sensitive than my device, but many times the price...
-
Wow you have the same device as me!
I already did a tear down, and broke something(whole thing still working after reparing), but I don't regret it. I believe if I don't see a device inside, it's not mine.
You want some teardown photos?
-
The Canon M20F-SH looks to be an impressive bit of professional camera kit :-+
https://www.canon.co.uk/for_home/product_finder/camcorders/multi-purpose-cameras/me20f-sh/ (https://www.canon.co.uk/for_home/product_finder/camcorders/multi-purpose-cameras/me20f-sh/)
-
Just want to know if the module in my Sp*rt looks exactly the same as the DEV kit device on their website. As the DEV Kit has a variety of direct video outputs. I have an idea from looking at the case there are "other" options either planned or discarded. A good question is what is the led or photosensor thing under the viewing screen, inactive on my model. Is there a slot for a standard GPS module?
Things of beauty:
Very flexible optics. A rifle scope or camcorder with a good zoom lens can look right into the eye piece, and the adjustable eye relief is very, very, good. Take note FLIR, they have you beat on that one in the low end.
Common battery, standard camera battery.
Decent battery life.
Color lets you correlate your daytime inherent knowledge of your area with the night scene.
Color with a little moonlight shows what the backyard Coyotes are doing, much better then gen II / Gen III greyscale (Greenscale>)
Also makes identifying cars much easier. Coyote fur blends in well with my local grass on tube based NODs.
Deals VERY well with bright sources in the field of view. Not producing serious halos..
Nicely complements my thermal imager.
Well programmed digital zoom code.
Decent weight.
Charges very quickly, keeps a charge. (Take Note FLIR, My Scout has to LIVE on a charger or external USB B pack)
Great dynamic range, much better then my Find-*R-Scope in the laser lab. Also no chance of indirect viewing if something went wrong.
Reasonably Tough.
Exporting images is very easy.
Even though it is shooting long exposures in deep darkness, algorithm used makes for a nice real time display.
Plenty of video color schemes, great ability to record stills and movies, with playback.
Downsides:
Loud Beepy noises on startup or mode change, . Makes me want to do a speakerectomy.
External function LEDs give your position away by design in Sp*rt model. Might just be they want other civilians to know your there.
Would be nice if it fades the display as you move it away, it will "beacon" you if not against your eye.
Would be nice to turn off the data strip at the bottom of the screen.
Rating for the cost: ***** Five Star.. Would definitely buy again.
Steve
-
How do these cameras compare to the "scientific grade" CMOS cameras, which seems to be more available? :-//
-
Tear-down photos!
16 photos total.
1/3 post.
-
16 photos total.
2/3 post.
Anyone knows the CPU model shown in "mobo1.png"?
-
16 photos total.
3/3 post(finish).
-
If you can see a way to get video output, please tell me!
The sensor behind the EVF window is not exist.
The GPS BGA place is on the button board, see my photos.
The notice sound can be turned off in settings.
Just put a sticker over the LED...
Turning off the boarder and use it for picture is a good idea, but:
The EVF is 1024x768(4:3), the sensor is 1-inch 1280x1024 (5:4) physically, with 1280x720(16:9) active, this is why they can use a more common 2/3 lens... Edit: You cannot, I tried the 8mm model lens within the same brand and series, dark corner appears. If you can unlock the full resolution, you're a hero.
Since you did a review, here's a simple one for my mind:
Bad:
No wired video (like HDMI) output;
External power don't work without battery inside.
Need improvement:
Have more lens (FOV) options;
Fix the Wifi function or remove it;
Fix a bug that you can't set time over 12:00 in 24-hour mode.
Maybe I ask for too much:
Custom CFA mentioned in OP;
A more "tear-down friendly" construction.
Good: Everything except above!
-
How do these cameras compare to the "scientific grade" CMOS cameras, which seems to be more available? :-//
sCMOS is actually much more expensive than this one...
Do you have a approximated ISO number of a normal sCMOS? that would be a good way to compare.
-
Wow, Thanks for the teardown. I'm impressed with how it is built. Clearly not intended for hacking, in fact probably quite secure. Their Intended market outside of the personal sales must be interesting.
Steve
-
is the lens a C or CS mount? Is it glued in?
Steve
-
i have the aurora pro and it is c-mount , or at least a c-mount screwed in just fine, the original was glued in by some filled holes around the lens , araldite type stuff but easy to see as its a whiter color
-
Check Thorlabs (https://www.thorlabs.com/newgrouppage9.cfm?objectgroup_id=11421 (https://www.thorlabs.com/newgrouppage9.cfm?objectgroup_id=11421)) for kinda affordable new ones and Andor for super fancy ones. Andor also offers expensive EMCCD cameras, and I once used one when I was in a cold atom lab.
These scientific cameras hardly give ISO specs but only Peak Quantum Efficiency (how many photons -> electron), Read Noise (how much additional noise when counting the electrons), and dynamic range.
-
I've heard they already have a agreement to supply sensors to US military, I forgot the source...
For the mount, read johnelot's reply, he's right.
-
Andor also offers expensive EMCCD cameras, and I once used one when I was in a cold atom lab.
I also saw an EMCCD camera in a lab, but never had a chance to see how does it look like when looking at a dark environment(instead of through microscopes...)
These scientific cameras hardly give ISO specs but only Peak Quantum Efficiency (how many photons -> electron), Read Noise (how much additional noise when counting the electrons), and dynamic range.
Yes that's the problem. I really want someone with both sCMOS and a "normal" camera that can set ISO, to compare them. But for years, no one done that yet.
-
There is some very interesting lens mod info and telescope builds with the aurora syonix and pro on the astro forum:
https://www.cloudynights.com/topic/681442-sionyx-aurora-for-nv-photography/?hl=%20aurora%20%20lens%20%20mod (https://www.cloudynights.com/topic/681442-sionyx-aurora-for-nv-photography/?hl=%20aurora%20%20lens%20%20mod)
And then there is this firearm enthousiast who modded it into some thermal fusion night vision device
(fiber optic periscope type)
https://www.thefirearmblog.com/blog/2020/09/25/friday-night-lights-diy-modified-aurora-digital-thermal-fusion/ (https://www.thefirearmblog.com/blog/2020/09/25/friday-night-lights-diy-modified-aurora-digital-thermal-fusion/)
[youtube] https://youtu.be/OUq2-VoDCoA (https://youtu.be/OUq2-VoDCoA) [/youtube]
-
Seems they have a paper here: https://www.spiedigitallibrary.org/conference-proceedings-of-spie/11741/117410E/Ultra-low-light-CMOS-image-sensors/10.1117/12.2588472.full?SSO=1 (https://www.spiedigitallibrary.org/conference-proceedings-of-spie/11741/117410E/Ultra-low-light-CMOS-image-sensors/10.1117/12.2588472.full?SSO=1)
~80% QE and 1.1e readout noise are pretty close, if not better, than the SoTA scientific CMOS. And to me, it looks like there isn't much headroom for further improvement in terms of sensitivity as it will get to shot noise limitation. I would like to get one if it's available in Europe.
Update: with a closer look at the website, I feel it's a bit dodgy as the spec for products is sort of vague, for example, I can not find what's the longest possible integration time. I hope I can find more info about the sensor used.
-
Thank you guys for replying.
There's something strange about this device: The "black silicon" should be several times (10x? I forgot) more sensitive on physical level, but comparing to Sony A7S which is not black silicon, this camera is about the same performance per pixel area (A7S have slightly better low light performance, with its lightly bigger pixel pitch).
I guess it's mainly because Sony have a much better processing algorithm(such as noise reduction), so for this camera, the advantage on hardware level gets buried by the worse processing...
-
Google Hyper-Had by Sony, "Hole Accumulation Diode" sensors were from say 9 years ago. I'm sure they improved upon it since then. For something like an SLR they might have used the newer "Super HAD" with lens on pixel technology and textured Silicon. The newer 5 transistor per pixel noise reduction circuit plus HAD has achieved really phenomenal results.
I used to have a picture shot with a HH sensor of a large conference room shot with scattered light from candle hid between a filing cabinet and a wall. Let a few long 100 mS exposures build up. Let it sum the exposures in the camera's DSP with whatever Kalman filter they use. Summed in the camera software, was almost as good as Gen II. night vision for a still scene if you were willing to set still for 10 seconds or so.
So I have no doubt Sony can push Silicon to its limits.
Steve
.
-
I've just acquired a used Sport and can confirm that it's an interesting toy. Sensitivity to red light, in particular, is extremely high. I suspect also to near-IR. When set to 7.5fps and black & white it's possible to see around a room lit by one power-on neon. But those YouTube videos claiming performance equalling or exceeding Gen III are at best somewhat misleading.
I plan to do some tests in due course to try and relate its sensitivity to ISO figures on a 35mm camera. But it'll have to wait until I can get to some open countryside: there's too much local IR floodlighting to do a fair test.
-
Thank you.
Google Hyper-Had by Sony, "Hole Accumulation Diode" sensors were from say 9 years ago.
If I remember correctly, early HAD series can back to 1980s...
For something like an SLR they might have used the newer "Super HAD" with lens on pixel technology and textured Silicon.
Their modern DSLRs are using CMOS instead of CCD.
Thanks.
-
Sensitivity to red light, in particular, is extremely high. I suspect also to near-IR.
Yes, I've read that the treatment for "black silicon" not only increase overall sensitivity, but also shift the spectrum response towards longer wavelengths.
-
I've seen strong response at 1084 nm on a laser emission line. Far stronger then normal silicon. I will check 1152 this weekend with an IR Helium Neon Laser. I'd be dreaming to see response at 1.35 or 1.5 um, but that would seriously make my day as a laser specialist.
Steve
-
It will be interesting to see whether the big DSLR manufacturers start using 'Black Silicon' in their cameras. I have a not-quite-top-of-the-range camera from a couple or three years ago that produces excellent results at ISO 10,000 (https://www.flickr.com/photos/ultrapurple/28063713968/) and even ISO 20,000 (https://www.flickr.com/photos/ultrapurple/38582873801/) - and that's still not at its maximum sensitivity. Getting (say) a factor-of-ten improvement would be interesting, although there are many other matters at play. For instance, my DSLR has a 35.9 x 23.9mm, 8256 × 5504 sensor, meaning a rather higher pixel density than today's SiOnyx sensors. But still, perhaps it's a portent of the future...
I know there's at least one camera, out for a few years, that claims a maximum sensitivity of something like ISO 4,000,000. Increase that by a factor of ten and you'd have trouble keeping enough light out of it!
-
I've seen strong response at 1084 nm on a laser emission line. Far stronger then normal silicon. I will check 1152 this weekend with an IR Helium Neon Laser. I'd be dreaming to see response at 1.35 or 1.5 um, but that would seriously make my day as a laser specialist.
Steve
I would love to know that too!
Please report back :-+
-
I know there's at least one camera, out for a few years, that claims a maximum sensitivity of something like ISO 4,000,000. Increase that by a factor of ten and you'd have trouble keeping enough light out of it!
That's the Canon ME20F-SH I mentioned earlier in this thread, it max out at ISO 4.5 million.
-
Impressive out a of a 1" sensor! This is an example of what Sony have been shipping in their high end security camera line (based on the A7s 35mm sensor).
https://www.youtube.com/watch?v=fi8v3hXBWNA (https://www.youtube.com/watch?v=fi8v3hXBWNA)
-
I've seen strong response at 1084 nm on a laser emission line. Far stronger then normal silicon. I will check 1152 this weekend with an IR Helium Neon Laser. I'd be dreaming to see response at 1.35 or 1.5 um, but that would seriously make my day as a laser specialist.
Steve
Such longer wavelength are another topic than just getting the highest sensitivity with silicon.
Germanium photodiodes on silicon are supposed to replace InGaAs sensors as a more affordable solution.
-
The Sony A7S line is well known among photographers as having incredible sensitivity.
https://www.youtube.com/watch?v=tAn27PrX1Ec (https://www.youtube.com/watch?v=tAn27PrX1Ec)
Of course, the latest micro-channel plates have inherently better resolution and response time (and they are much more compact and the power requirements significantly lower), but the good thing about the A7S's is you can get them at Best Buy :)
-
Impressive out a of a 1" sensor! This is an example of what Sony have been shipping in their high end security camera line (based on the A7s 35mm sensor).
https://www.youtube.com/watch?v=fi8v3hXBWNA (https://www.youtube.com/watch?v=fi8v3hXBWNA)
Interesting finding for another camera using (maybe) Sony a7s sensor.
Canon also have ML-100 and ML-105 using (maybe) Canon ME20F-SH sensor.
-
I've just acquired a used Sport and can confirm that it's an interesting toy. Sensitivity to red light, in particular, is extremely high. I suspect also to near-IR. When set to 7.5fps and black & white it's possible to see around a room lit by one power-on neon. But those YouTube videos claiming performance equalling or exceeding Gen III are at best somewhat misleading.
I plan to do some tests in due course to try and relate its sensitivity to ISO figures on a 35mm camera. But it'll have to wait until I can get to some open countryside: there's too much local IR floodlighting to do a fair test.
Apologies for digging up a slightly older thread, but that would be an interesting comparison, particularly how the response varied between a full spectrum DSLR/MILC as you get into near-IR frequencies. Full-spectrum conversion companies have said their modified cameras are sensitive up to 1100nm, the Sionyx appears to be significantly better in this regard. This is the closest I've seen to a comparison between them:
https://www.youtube.com/watch?v=sk2US8E-38o (https://www.youtube.com/watch?v=sk2US8E-38o)
-
I also want to play with it's monochrome SDK. Without color filters, sensitivity will be even better.
-
They released 2 new models: OPSIN and Nightwave(one finally with video output). Youtube tests shows they actually have further improved sensitivity.
BTW, while everyone is talking about Sony A7S with 400k ISO, no one mention Nikon D7500? It have 1.6 million ISO, even in video mode. Too bad their newer mirrorless models don't have that capability.