Very low light sensor

Started Jan 12, 2006 | Discussions
ArtieD
Forum MemberPosts: 65
Like?
Very low light sensor
Jan 12, 2006

Anyone hear / know anything about this company and the prospects for products using their technology? PC Mag says that they showed this sensor at the CES.

http://www10.nanotechcafe.com/nbc/articles/view_article.php?section=CorpNews&articleid=232849

http://www.pcmag.com/article2/0,1895,1908418,00.asp

Artie
--
From the Blue States of America

Roland Karlsson
Forum ProPosts: 20,637Gear list
Like?
Re: Very low light sensor
In reply to ArtieD, Jan 12, 2006

ArtieD wrote:

Anyone hear / know anything about this company and the prospects
for products using their technology? PC Mag says that they showed
this sensor at the CES.

The pcmag article says the sensor is 1000 times more sensitive than a normal sensor and that they produce 1000 electrons out of one photon.

Now - there might be some misundertstanding somewhere - but generating 1000 electrons out of one photon still gets the same photon noise.

By this technique you can of course virtually remove other noise sources. So - then you can get a super clean sensor without cooling. But - I doubt that the noise in an ordinary senor is 1000 times larger than the photon nose.

Or is it?

Roland

 Roland Karlsson's gear list:Roland Karlsson's gear list
Canon PowerShot G10 Pentax K-5 Pentax smc DA 15mm F4 ED AL Limited Pentax smc FA 50mm F1.4 Pentax smc DA 21mm F3.2 AL Limited +6 more
Reply   Reply with quote   Complain
Caer
Senior MemberPosts: 2,160
Like?
Re: Very low light sensor
In reply to Roland Karlsson, Jan 12, 2006

I wonder if that's the point - since it generates 1000 electrons per photon it basically amplifies the signal without amplifying the noise.

-- hide signature --
Reply   Reply with quote   Complain
Roland Karlsson
Forum ProPosts: 20,637Gear list
Like?
Re: Very low light sensor
In reply to Caer, Jan 12, 2006

Caer wrote:

I wonder if that's the point - since it generates 1000 electrons
per photon it basically amplifies the signal without amplifying the
noise.

Thats probably the point.

But -- it is not likely that you generate exactly 1000 electrons per photon - so you add some noise there also. But - of course - that noise is probably below the photon noise.

One other thougyt - if you generate 1000 electrons per photon you will saturate the sensor very fast. In practice it sounds like you will get a verry shallow dynamic range.

Roland

 Roland Karlsson's gear list:Roland Karlsson's gear list
Canon PowerShot G10 Pentax K-5 Pentax smc DA 15mm F4 ED AL Limited Pentax smc FA 50mm F1.4 Pentax smc DA 21mm F3.2 AL Limited +6 more
Reply   Reply with quote   Complain
Andy Moreton
Regular MemberPosts: 241
Like?
Re: Very low light sensor
In reply to Caer, Jan 12, 2006

Sounds like a modern version of the old night vision device:
http://electronics.howstuffworks.com/nightvision5.htm

At very low light levels photons will be arriving relatively infrequently (per photosite), so the images can appear to be made up of lots of small flashes of light making moving images appear very grainy. Unless a long shutter speed was used I would expect still images to be grainy too, though that would depend on how much or how little light there was.

Reply   Reply with quote   Complain
Roland Karlsson
Forum ProPosts: 20,637Gear list
Like?
Re: Very low light sensor
In reply to Andy Moreton, Jan 12, 2006

Andy Moreton wrote:

Sounds like a modern version of the old night vision device:

Yepp - thats what it is.

Nothing wrong with such a device - but it is not HiFi photography.

Roland

 Roland Karlsson's gear list:Roland Karlsson's gear list
Canon PowerShot G10 Pentax K-5 Pentax smc DA 15mm F4 ED AL Limited Pentax smc FA 50mm F1.4 Pentax smc DA 21mm F3.2 AL Limited +6 more
Reply   Reply with quote   Complain
Thomas Nilsson
Contributing MemberPosts: 529
Like?
But it is...
In reply to Caer, Jan 13, 2006

Caer wrote:

I wonder if that's the point - since it generates 1000 electrons
per photon it basically amplifies the signal without amplifying the
noise.

Well, the photon noise does get amplified. But as I understand it isn't the photon noise that makes up the majority of the noise that we normally see in digital photos. It is thermal, dark current and readout noise.

However, in a camera like that, where the photon noise would be amplified a 1000 times, while the other noise sources are not amplified, it's a completely different story.

Ciao

Thomas

Reply   Reply with quote   Complain
omr
omr
Senior MemberPosts: 1,118
Like?
I already have a '<1 lux' camera :)
In reply to ArtieD, Jan 13, 2006

Planet82 says the sensor can capture images when light level is less than 1 lux, 'or up to 0.1 lux'.

"0.1 lux" does sound impressive.

I have a camera that can capture night video at less than 1 lux:
It's the old Vista Imaging design which gained popularity as the
3com HomeConnect camera. The latest version is now sold by
IONetworks and is called the Watchport/V2.

http://www.ionetworks.com/products/environmentalmonitoring/watchportv2.jsp

"... exceptional low light sensitivity (

Years ago I used it for surveillance in a very dimly lit parking lot.

-- hide signature --

omr

Reply   Reply with quote   Complain
Piotr123
Regular MemberPosts: 395
Like?
12 December 2005
In reply to ArtieD, Jan 13, 2006
Reply   Reply with quote   Complain
DRG
DRG
Veteran MemberPosts: 5,216
Like?
Re: But it is...
In reply to Thomas Nilsson, Jan 14, 2006

Thomas Nilsson wrote:

Well, the photon noise does get amplified. But as I understand it
isn't the photon noise that makes up the majority of the noise that
we normally see in digital photos. It is thermal, dark current and
readout noise.

Actually, modern imaging sensors are now near the photon-noise limit, with other noise sources being far lower. If it were the other way around, noise would grow increasingly worse with exposure time, which it does not except for those few "hot" pixels with exceptionally high thermal/leakage noise. Noise would also improve as pixel area scaled down, since thermal noise is proportional to the pixel charge well volume (photon statistics noise goes the other way). However, as we all note, noise increases with decreasing sensor size and/or increasing pixel count. Direct measurements on DSLR sensors also shows noise quite close to values expected from photon statistics.

Basically, this implies that these new sensors simply possess the properties of very high ISO, conferred by a built-in photomultiplier, and will thus be both noisy and prone to saturation once light levels reach moderate levels.

David

Reply   Reply with quote   Complain
JKP
JKP
Regular MemberPosts: 232
Like?
Re: But it is...
In reply to DRG, Jan 16, 2006

Their web page http://www.planet82.com gives some specs for their new chip:

Resolution: 640 x 480
Pixel size: 4 x 4 ┬Ám
Image area: 2.56 x 1.92 mm
Frame rate: 30 fps
Dynamic range: 80 dB
Power consumption: 82 mW@30fps

So we not talking about high end CMOS chip now, at least for DSLR use.

-jkp-

Reply   Reply with quote   Complain
Zarathustra
Senior MemberPosts: 1,797
Like?
It is here now! CES '07....
In reply to ArtieD, Jan 9, 2007
Reply   Reply with quote   Complain
Andrew dB
Contributing MemberPosts: 970
Like?
CMOS version of an EMCCD? (nt)
In reply to ArtieD, Jan 9, 2007

(no text)

Reply   Reply with quote   Complain
Sheld
Senior MemberPosts: 1,899
Like?
it might have "fovevan" like qualities (not Bayer sensor)
In reply to ArtieD, Jan 9, 2007

A quantum tunnel detector might detect individual photons -- each photos changes the voltage on a switch that is on the edge of tripping already. Moreover, the current would be proportional to the frequency of the photon, which would allow the sensor to record each individual photons time, and frequency. Frequency means no bayer interpolation is necessary. Position is more difficult, since the smaller each photo sensor gets, the more releavent the uncertainty principle becomes. But with a big sensor (say FF, 6 megaixel, with 12 micron pitch), one could in principle be looking at equivalent resolution to a 6 megapixel fovevan 3 color sensor, or about 12 megapixels. with the same or better noise charecteristics as a 12 micron bayer pixel.
--

Reply   Reply with quote   Complain
ledittmar
Veteran MemberPosts: 3,495
Like?
Re: Very low light sensor
In reply to ArtieD, Jan 10, 2007

So where is the finished product, like a dslr?

Reply   Reply with quote   Complain
John Sheehy
Forum ProPosts: 16,026
Like?
Re: But it is...
In reply to DRG, Jan 10, 2007

DRG wrote:

Thomas Nilsson wrote:

Well, the photon noise does get amplified. But as I understand it
isn't the photon noise that makes up the majority of the noise that
we normally see in digital photos. It is thermal, dark current and
readout noise.

Actually, modern imaging sensors are now near the photon-noise
limit, with other noise sources being far lower.

The rumors of the death of read noise are greatly exaggerated, IMO. Read noise is still the limiter of DR at low ISOs and large pixels. The shadows of ISO 100 are garbage compared to the same light levels captured at ISO 1600 on recent Canon DSLRs. Small differences in amplification of different lines shows up in low-ISO images, too, limiting S/N in highlights far more than shot noise does (and 1-dimensional noise is much more powerful than 2D noise, visually). If camera output were truly limited to signal and shot noise, it would be a revolution.

-- hide signature --

John

Reply   Reply with quote   Complain
nickleback
Forum ProPosts: 11,111
Like?
Re: Very low light sensor
In reply to ledittmar, Jan 10, 2007

ledittmar wrote:

So where is the finished product, like a dslr?

I don't know about you, but I'm not interested in a 1/6" 0.25mp sensor in a DSLR.

http://209.85.165.104/search?q=cache:QlW7uQPiVCYJ:www.planet82.com/korean/technology/imagesensor_03.asp+site:planet82.com&hl=en&gl=us&ct=clnk&cd=4

-- hide signature --

Seen in a fortune cookie:
Fear is the darkroom where negatives are developed

Reply   Reply with quote   Complain
wll
wll
Contributing MemberPosts: 533
Like?
In another 3-5 years ..
In reply to ledittmar, Jan 10, 2007

There will be aps-c sized sensors that are FAR better than what we have now and the way things are going flash may be a thing of the past ...

wll

Reply   Reply with quote   Complain
Sheld
Senior MemberPosts: 1,899
Like?
Re: In another 3-5 years ..
In reply to wll, Jan 10, 2007

agreed; any new technology has enormous hurdles to overcome before it can become competitive with today's bayer sensors. my guess, is that it might happen first in some special purpose application like a microwave astronomy sensor or something like that.....

  • Shel

-- hide signature --

Reply   Reply with quote   Complain
John Sheehy
Forum ProPosts: 16,026
Like?
Re: Very low light sensor
In reply to Roland Karlsson, Jan 10, 2007

Roland Karlsson wrote:

By this technique you can of course virtually remove other noise
sources. So - then you can get a super clean sensor without
cooling. But - I doubt that the noise in an ordinary senor is 1000
times larger than the photon nose.

Or is it?

Nearer black, the photon noise increases relative to signal, but decreases in an absolute sense. Dark current noise and blackframe read noise are both independent of signal, and do not decrease in an absolute sense, closer to black signal, and approach an infinite ratio to signal closer to black signal. In truly extreme shadows, the read and dark current noises are far more significant than shot noise.

Read noise often has a patterned component that makes it more visually intense than shot noise of the same statistical value, and this becomes even more true, both visually, and statistically, if you view a print from a distance, or perform software binning or downsampling - the unpatterned noise dissapears more readily.

-- hide signature --

John

Reply   Reply with quote   Complain
Keyboard shortcuts:
FForum MMy threads