Slashdot Mirror
Canon Shows the Most Sensitive Camera Sensor In the World
An anonymous reader writes "Canon announced today that it successfully developed a super high-sensitivity full-frame CMOS sensor developed exclusively for video recording. The new Full HD sensor can capture light no other comparable sensor can see and it uses pixels 7.5 larger than the best commercial professional cameras in existence today."
There doesn't seem to be a gallery of images, but the video demo (direct link to an mpeg4) makes it seem pretty sensitive.
pixels 7.5 feet larger... that's quite a lot. No wonder it can capture so much light.
Slashdot social media options: AIM, ICQ, Yahoo, Jabber and Mobile Text. Why no MySpace?
This is just so awesome. As a Nikon fan, I'm a little upset it's canon, lol. But no, this is awesome.
(in before paranoia about big gubermint surveillance, etc, please go away, just enjoy the cool tech)
My friend will love this for his strip club shots.
....and all CMOS sensors are inferior to CCDs for noise at low light levels, so this is just a good CMOS sensor. CCD noise, when cooled, is measured in electrons per hour.
You accidentally the units. 7.5%, 7.5x, 7.5nm...
const int one = 65536; (Silvermoon, Texture.cs)
SJW, n: "Someone I don't like, and by the way I'm a fuckwit" - AC
... for when my 'subject' turns off her bedroom light.
Whats up with the 1990's 640 x 360 video resolution? This is 2013 for fucks sake. Also, tripods exist.
user@host$ diff
something was missing in the sample video: a scene having a high contrast - eg a dark area + the Moon above, or an interior scene where the sun shines through a window + the back of the room in the dark, or a well-lit city + milky way above. How does the camera behave in that case? Does it record enough information (very HDR) to allow the post processing software to balance areas (using complex algorithms, like current HDR programs on DSLR), resulting in both areas clearly visible to the naked eye? Or, instead, the dark area will be clean visible whereas the well lit area will be burned - ie white? That has been (and still is) the problem even with DSLR, and that could be worse with a ultra-sensitive sensor.
Slashdot, fix the reply notifications... You won't get away with it...
Gosh, you're so full of, well, the latest, most hip, wonderful.... STUFF. You must be a PRO! You know so much about light and stuff.
TFA do not say anything about pixel count.
But a simple computation (24mmx36mm , 19mu/pixel) give 2.5 Mpx.
Probably not a value that Canon want to show too prominently, "Full HD" is better for marketing.
Err, of course the sky is blue under moonlight: it's just reflected sunlight, after all (but see below).
The problem is that the Moon is much fainter than the Sun and thus the overall light level is low. So low that it doesn't significantly activate the colour-sensitive cones in the human eye, meaning that you only really see with the rods in black-and-white.
But take a long exposure with a camera (or a video frame rate with this Canon sensor), and the blue will most definitely come through.
(Actually, the moonlight-illuminated sky is slightly bluer than a sunlight-illuminated one, as the Moon's slightly brown-ish colour first imprints its spectral dependence on the sunlight which bounces off it. That light is then Rayleigh-scattered off the molecules in the Earth's atmosphere, imprinting the well-known 1/lambda^4 dependence which makes the sky blue).
The newly developed CMOS sensor features pixels measuring 19 microns square in size, which is more than 7.5-times the surface area of the pixels on the CMOS sensor incorporated in Canon's top-of-the-line EOS-1D X and other digital SLR cameras.
I guess this is to collect more photons in low light conditions. Of course this means that sensor is physically larger, but that's not a problem for Canon, they have made medium format cameras in the past.
Oh...I guess it did happen.
Still has the potential for tearing during fast motion recording, as the pixels are scanned one by one, not captured all at once. Apparently global shutters for cmos sensors is uncommon.
Have you ever taken a shoot with a 50mm f:1.5 on one of today's pro cameras? If you multiple the light available by 56 times (increasing the pixle size by 7.5) you are looking at a shit load of light.
I can shoot nice outdoor night pics under a full moon with just an F5 @ 5 seconds, drop that down to a F1.5 and that's more than 8 times the light, add this new sensor and that's 280 times more light! or about 6 FPS and note this is on a canon 5D MK I that's almost 7 and 1.2 years old..
Add in the current 3 and half year old generations improvement on the ISO and that goes from the 1600 I shoot at, to 6400, or 4 times the light and you get 24 FPS..
So thats with jsut the sensor.. If they where using F1.2 or even F1 lens and one would expect when showing such a beast and 30 FPS seems like no issue at all..
You have 5 Moderator Points!
Which Helpless Linux zealot/MS basher do you want to mod down today?
There's just that much area in a pixel. So there can only be that amount of photons coming in, no matter how sensitive your sensor is.
Here's the clue: these pixels are BIGGER than before.
Quite aside from that, I don't think we're anywhere near the point where we can detect every single incoming photon, so there's still room for improvement regardless. You may as well argue that there's nothing interesting about better solar panels, because there are only so many photons htiting them.
systemd is Roko's Basilisk.
Also to your full HD comment, the 5D Mk II shoots pictures in 3861 × 2574 which is larger then 4K... so if you can take the shoots as per my other post, 4K video is possible looking like daytime under a full moon if you can get something to store that data for you..
4K = 4096 × 2160 = 8847360 pixels
5D MKII = 3861 × 2574 = 9938214 pixels
--
You have 5 Moderator Points!
Which Helpless Linux zealot/MS basher do you want to mod down today?
You have 5 Moderator Points!
Which Helpless Linux zealot/MS basher do you want to mod down today?
The same video is also available at youtube, which presumably has more bandwidth than Canon's poor server.
That need not even be the case. You could still do it in 35mm, 1080p HD video is roughy 2.1 megapixels, where as the EOS1DX is 18.1mp.
So there is definitely enough room to make pixels 7 times larger than a EOS1DX
Sometimes I wish I was a plumber, then I'd know how to deal with other people's shit.
It's a 35mm full-frame sensor.
It's also explicitly intended "exclusively for video recording" and mentions "full HD". Which would mean - assuming I'm reading between the lines correctly - that the resolution is 1920x1080 - ie. 2 megapixels.
...is there a still version for astronomy?
Of course this means that sensor is physically larger
The sensor isn't physically larger. The specs say it's a full-frame 35mm sensor, and the photo of the prototype camera shows it with a standard EF lens mounted: a larger sensor would need medium or large format lenses, and it would be pretty much dead on arrival in the market if you had to go out and start buying medium or (God forbid) large format lenses to feed the thing. Half of the allure of Canon for video, after all, is that you can reuse your still EF lenses, and demanding huge format glass for HD video would be absurd.
The reason the photo sites are so much bigger in this sensor, presumably, is because the resolution is much lower than Canon's still SLR cameras. It doesn't give the resolution, but since it was only described as capturing "HD video," I wouldn't be surprised to find that the sensor's native resolution is that of 1080p video: 1920x1080 pixels, or about 2 megapixels. The 1Dx, on the other hand, has a native resolution of 18 megapixels.
So far, Canon (and more recently Nikon), have been allowing users to record HD video on their SLR cameras by scaling their massive native resolutions way down to a size that you can reasonably encode and cram onto a memory card in an SLR form factor. This approach, on the other hand, seems to be to build a sensor with a lower native resolution suitable for HD video at the same size as the larger SLR sensors, so you don't have to do any down-scaling and you get massive photo sites, which gives you a huge advantage in sensitivity.
Wouldn't it make more sense to use a system with a larger width lens to gather and capture more photons to increase the ability to get imagery in low-light conditions? Funnel more photons captured with a larger lens onto the same focal plane: more photons come onto the same pixel areas, leading to higher signal levels for the same stop and exposure time, right?
Perhaps the most sensitive in its class.
It's a CMOS sensor. The sop end SCMOS ones are capable of photon counting with a quantum efficiency in the high 70%s. The best EMCCDs push that up to about 85% or so. With cooling, the readout noise is very low.
This may be an excellent standard sensor, but more sensitive ones certainly exist.
Large pixels are great for standard photography though.
SJW n. One who posts facts.
Sensitivity and dynamic range are separate things. You can have an extremely sensitive sensor, pushing an equiv of ISO 12800 or even more, but dynamic range may only be 8-10EV
Even the best of the best have around 13EV of dynamic range(eg Nikon D7000) at ISO 100. As you increase ISO, the dynamic range suffers, and noise increases. Getting to above 14-15EV is very very difficult. You can do it in post processing(HDR combination of multiple exposures)
My Aurora : http://www.youtube.com/watch?v=o91ZsGwJYyg
FB : https://www.facebook.com/TanveersPhotography
I would actually like to see them implement this in a still camera. Cut the megapixels down to 2 or 3 (this is more then enough to put online and ), and make it 7x more sensitive than today's still cameras.
Sometimes I wish I was a plumber, then I'd know how to deal with other people's shit.
Yes, but then you would need special lenses. This sensor works with 35mm (full-frame) lenses.
"A week in the lab saves an hour in the library"
Make it long enough, and the moon will look like the sun
Here you go, an example shot
http://tanveer.smugmug.com/Travel/Ladakh-2012/TSR/i-v7ZtdHc/0/L/DSC_5964_LR-L.jpg
Its all in the exposure
My Aurora : http://www.youtube.com/watch?v=o91ZsGwJYyg
FB : https://www.facebook.com/TanveersPhotography
This sensor would be fantastic for microscopy. The current range of "scientific" cameras are woefully under-specced even compared to consumer DSLRs (tiny sensors, small pixels, high noise level even with peltier cooling). Canon can eliminate Leica from that market with a product like this.
"A week in the lab saves an hour in the library"
Re:Long exposures of full moon often turn out "blue"
Unless it's a video; in which case it will just turn "boring".
Could not doing 3x3 binning on the existing 18MP sensor (if the controller supported it) produce similar results?
Hopefully, it won't plead for too many trips to the psychotherapist.
Use ISO 8601 dates [YYYY-MM-DD]
Should not a 7.5x larger pixel collect 7.5x more light? It's 7.5x the area, not the linear dimensions.
19-micron pixels seem big if you're comparing them to DSLRs, where everything has to fit into a nice little portable package. But it's not at all an unusual size in science-grade detectors used for astronomical instrumentation. At work our instruments use detectors with pixel sizes ranging from 13.5 to 50 microns.
I might be a little more impressed that they're doing this at video frame-rates, and without cryogenics...
Village idiot in some extremely smart villages.
Especially with the sky being blue from the full moon alone.
is the same as the AC who then posted this one:
I obviously didn't object to it being that hue, dumbass. I objected to it being *that* bright. It was a day shot. And obviously so.
then I'd say that's exactly what you did say.
And as for your assumption that I'm an American ... well, you haven't got a clue, mate. You're many thousands of kilometres off. There are other countries in the world where English is the native language, after all. "We have 2013" indeed.
Sometimes I really do wonder whether /. is worth the trouble.
Indeed on the latest pro cameras you already get ISO sensitivities of up to 204800. With f/1.4 and exposure time of 1/25s you'd get good images for illuminations of about
50 * 1.4^2 / (1/25*204800) = 0.01 lux
There are some quite impressive videos already out there, e.g.
http://www.youtube.com/watch?v=lqwbABVxoC8
http://www.youtube.com/watch?v=JXOsueKwjOY
How can the camera capture all those colours in full moon, that makes no sense to me. In low light conditions it should be almost black and white, can anyone more knowledgeable explain this to me?
Bought a Canon Legria HF G10, very pricey! It comes with same claims that it is very good in low light conditions. I didn't find that to be true. Noisy and pixely is more like it. Don't know what to believe in now and who regulates their claims.
no you sir are as gullible as fuck, go look at what a decent 7 year old DSLR can do under a full moon and report back.. Those of us that have them know its not staged at all.
You have 5 Moderator Points!
Which Helpless Linux zealot/MS basher do you want to mod down today?
I'm not familiar with the instruments described in your link, but they don't look either cheap or suitable to use in a consumer video camera. The clips of the meteor shower and the guy with the incense looked like they were taken at video frame rates, that prototype camera looks like it can be hiding a Peltier cooler, thoughttp://hardware.slashdot.org/story/13/03/05/0219240/canon-shows-the-most-sensitive-camera-sensor-in-the-world#h.
... I tend to think that is an important R & D intermediary result, somewhere between 5 and 6 when speaking in Technology Readiness Levels. Canon is not going to sell this "as is", methinks. But it is important proof to them that they are on the right way. Which is typically what you want from reaching TRL 5 or 6....
Religous speak to God. Insane are spoken to by God. When all shut up, one can finally hear Shostakovich in peace
That's the comparison you saw in the video. Binning is generally how the down-sampling is done.
So why have the not been upgrading sensors? Nikon is eating their lunch lately with Sony's latest also beating them. You have to hang your head low when Sony beats you.
All of the camera lines have been stagnant in the sensors and MP for years. the Rebel should be 24MP, the 5D MK III should have been 30+MP and the 1DS should have been upgraded to ungodly nearly medium format resolutions by now.
Nikon is doing that, Sony is catching up.... What the hell is wrong with canon? All I can see is they are busy adding flip out touchscreens and trying to convince people that its a "feature"
Do not look at laser with remaining good eye.
... and avoid sharply worded questions.
...eliminates much of the information that humans cannot see.
(Which is why the big view screen on Star Trek probably does not use an MPEG codec... Klingons would think humans were weirdly colored blocky beings).
I am very small, utmostly microscopic.
Moonlight is just reflected sunlight, you just need more of it to make the colours come out.
E.g. see http://www.flickr.com/photos/dansdata/3074862610/ for an example - this photo was taken under a full moon, 30 second exposure.
I objected to it being *that* bright. It was a day shot. And obviously so.
What else would you expect a night shot from a highly sensitive camera to look like? What differences would you expect to see which you are not seeing?
systemd is Roko's Basilisk.
We're a lot closer than most people think. A hell of a lot closer than we are with solar panels.
Would have been nice to see the video demo in HD.
Bah! Forgot link.
http://www.sensorgen.info/
Don't they say its a video only sensor? If its consumer grade then most consumer cam corders do not have changeable lenses.
Well yes. Bigger lens. More light. But CMOS sensors are not very good. CCD's can be but they are not exactly cheap.
who read that headline in Jeremy Clarkson's voice?
By "larger width" lens are you meaning larger aperture, smaller focal length or larger image circle?
All of these things have effects on the image and have practical limitations. (for example change the image circle area to be larger and you need a new format, all the old 35mm lenses they have can't be used for the larger frame unless you like black edges).
CMOS sensors do not use their full surface area for photosites, unless something fundamental has changed in the last five or so years. For each photosite, there are other surface area needed for things like providing data paths off the sensor. I'm making an educated guess here and saying that having physically larger photosites allows a larger proportion of the sensor area to actually be used to collect light in this case, since less of the surface area needs to be used for other purposes. Additionally, larger photosites are less prone to noise. These are the reasons that 3x3 binning is not equivalent.
If you have to cool a CCD camera you could be looking at silly money. I did astronomy so sticking cameras on the back end of telescopes.
If this is put in super high end video cameras, it doesn't make sense that it is only 2 MP not 4, so you can't do 4K video like the Red cameras.
Think stupidly large optics. And no a 35 mm camera could not cope. 24-inch may work (lowell telescope)
No, becuase of at least 2 problems (may be more I can't think of)
1) With binning, you lose the ability to detect photons that hit between the individual detectors
2) Each individual detector has a certain noise level, and the fewer photons you detect, the lower your S/N (signal-to-noise ratio) becomes. So when you bin smaller detector, you are binning a bunch of low S/N data, giving you a a results that is low S/N. If you have a single larger detector, your S/N is much better.
If you are talking about the size of the camera, I doubt it's that large because of cooling. My guess is, being a prototype, the electronics aren't all refined yet to fit neatly into a compact space.
They use another trick to take care of this, though: "microlenses", miniature optics in front of each pixel that channel light away from the insensitive regions (the data paths) and onto the actual light-sensitive pixels. A recent advance is "gapless microlenses", where nearly all of the light incident on the sensor winds up falling on some pixel or other.
Should not a 7.5x larger pixel collect 7.5x more light? It's 7.5x the area, not the linear dimensions.
You are correct on that. From the article:
The newly developed CMOS sensor features pixels measuring 19 microns square in size, which is more than 7.5-times the surface area of the pixels on the CMOS sensor incorporated in Canon's top-of-the-line EOS-1D X and other digital SLR cameras
Are people seriously still arguing that. There's nothing wrong with CMOS sensors. It really depends upon how well the particular chip is designed, yes you can make cheap sensors, but you can also make some really nice ones as well.
You could probably get a 35mm lens to focus on a larger CCD, and then trim out the "undefined" area outside of that. But most people aren't hackers like that...they'll see the different number, think "not compatible" and move on.
When you see "35 mm", think "8086 segmented real mode" (which was added for backwards compatibility with the 8080, FYI), which you'll still find in modern Core i7's or FXes. Just one of those things that is never going to go away due to so much capital invested in a specific platform with a long legacy.
Yeah, you might be able to get rid of it, but its not "35mm/x86" anymore!
The real path to male liberation
you'll want a bit extra for debayering.
Yes, but the tradeoff for a low f-stop (which I assume is what you're talking about) is smaller depth of field and all that that brings, like it being more difficult to focus on your subject. Not an issue with telescopes but definitely an issue with a low-light home video recorder.
I feel sorry for people that don't drink, because when they get up in the morning, that's as good as they're gonna feel
no you sir are as gullible as fuck, go look at what a decent 7 year old DSLR can do under a full moon and report back.. Those of us that have them know its not staged at all.
The fact that the caption on that shot says "After increasing sensitivity, the newly developed CMOS sensor is capable of capturing video such as this." tells me that it is staged. The wording of that sentence doesn't imply a real shot taken with the sensor, but a mock up of what they expect it can do.
-- ssoorrrryy,, dduupplleexx sswwiittcchh oonn.. -Quote found on actual fortune cookie.
"Video only sensor" doesn't make sense. It may be intended to be used only for video, but that's not a property of the sensor. There's no reason this sensor cannot be used to make photos if you are satisfied with the resolution. Imagine the images you'd get with an exposure time of 1/8 instead of 1/60. Or even really long exposure times like 10s (maybe you'd get a reasonable image at new moon, illuminated just by the light of the stars?)
You're right that I did not mention microlensing. I didn't think it necessary to the point I was making. However, you are greatly overstating the performance benefits from using gapless microlenses.
The link in TFA that says it the pixels are 7.5 x larger than the "best commercial professional cameras in existence" actually points to a page that says:
Each pixel on the new sensor measures 19 microns square, more than 7.5-times the surface area of the pixels on the CMOS sensor incorporated the company most advanced (and expensive) top-of-the-line EOS-1D X camera released last year.
TFA therefore assumes that Canon makes the best cameras in existence. Excluding professional digital backs, the Nikon D800 has 4.88 micron pixels, which is 23.8 square microns--but let's assume that "microns square" means square pixels 19 microns on each side even though it specifically refers to surface area. 4.88 x 7.5 microns = 36.6 microns, which is about twice the size of the pixels in this sensor. Moreover, the camera is a prototype and only for video. The D800 is on the market and capable of both stills and HD video.
I happen to be a Nikon fanboy, so I look at this as Canon hyping their lab results to cover up for the fact that Nikon beat Canon to market with a 36.3 million pixel full-frame sensor, which they responded to with the 22.3 million pixel Canon EOS 5D Mark III. I also happen to own a D800E and have never, ever seen a DSLR punish a lens (because the sensor exposes every flaw) so thoroughly or produce such amazing dynamic range and color depth at ISO 6400.
Actually, I wrote my thesis on life experience.
Werewolves. Definitely werewolves. They come out at night.
they use 19 micron square pixels. A "35mm" format is actually 24mmx36mm (or 24,000x36,000 microns). This works out to be 2.3 megapixels assuming no gap between the pixels. (ie, 1263x1895)
I've lost track of where things stand in the astrophotography world but this sounds like it could be very useful.
Wouldn't it make more sense to use a system with a larger width lens to gather and capture more photons to increase the ability to get imagery in low-light conditions? Funnel more photons captured with a larger lens onto the same focal plane: more photons come onto the same pixel areas, leading to higher signal levels for the same stop and exposure time, right?
Why not do both?
All DSLR's for as long as I can rememberer have come with the ability to increase sensitivity, or as the consumer knows it ISO.. My old crappy one goes from ISO 100 to 3200.. as someone else has posted current generation cameras can go up-to 256000
You have 5 Moderator Points!
Which Helpless Linux zealot/MS basher do you want to mod down today?
I don't always take photographs. But when I do, it's with a Canon.
Stay photogenic, my friends.
IIRC there usually is a small gap between the pixels on a sensor - someone else has also mentioned that you'd want slightly higher resolution for debayering so it sounds like I was probably right - this sensor will produce images of approximately 1920x1080. But it'll be exquisitely sensitive to low-light conditions.
I think this is an attempt to be taken more seriously in the world of professional film recording - historically, companies like Sony and Panasonic have been market leaders here.
More than a few people have made very positive noises about using things like an EOS 5D but that wouldn't be terribly useful in low light situations.
Put this into a body that's designed to be handheld for long periods of time (like you see with shoulder-mounted studio cameras), stick an EOS lens mount on the front and suddenly Canon have the building blocks for a complete portfolio of pro-grade video equipment.
I'm referring to the "capable of capturing video such as this" part of that sentence and not the increased sensitivity part. It just sounds like weasel wording implying that the image is a mock up of what it is capable of rather than it actually doing it.
-- ssoorrrryy,, dduupplleexx sswwiittcchh oonn.. -Quote found on actual fortune cookie.
Note that the difference between 75% and 85% quantum efficiency is not that great. If the pixels at 7.5 the area of a competing CCD, then the CMOS would get (7.5 x 0.75 / 0.85 = 6.6 photons for every photon the CCD got; i.e., the exposures could go down by a factor of 6.6. Even if the CCD had 100% efficiency, that would still be a factor of 5.6.
This could be a boon in observing small asteroids, which are dim and tend to rotate fast (some less than 1 minute) due to YORP.
No.
Larger pixels do not 'collect more light' in real camera systems. This is because in real camera systems, a certain field-of-view is used independent of sensor size (cinematographers do not frame differently with full-frame sensors...they use whatever focal length is needed to achieve the field-of-view they want). In practical terms, this means that the same amount of light is captured from the scene and funneled through the aperture, regardless of sensor size. If you make the sensor bigger, you just spread the same amount of light over a larger sensor. The pixels are bigger, but they are intercepting the same number of photons/second.
If, and only if, making the sensor larger allows a larger % of the sensor to be covered with active sensing area (that is, the 'fill factor' goes up as a result), then maybe the larger sensor is more sensitive. MAYBE. There are other issues working against you when you make the sensor bigger--areal current density is going to go down, which is a big deal if you have a noise floor (digital) and/or your sensitivity is a function of current density (film). You can increase the aperture (D) to maintain F-stop and therefore keep current density the same for the larger sensor, but you can increase the aperture with a smaller sensor too and still win. Practical DOF is a function of aperture (D) and magnification, and we were discussing equivalent camera systems.
There is a reason film formats steadily decreased from 11x14 to 8x10 to 4x5 to 6x7(cm) to 35mm as film became finer-grained. Making the film area smaller increases the areal current density (photons/mm^2) with no other practical difference in real camera systems. This is a big win with film because speed is a function of current. It's a big win with digital because larger sensors are expensive, and since there is NO practical effect on real camera systems, you might as well use a small sensor, since digital sensor pixel pitches are fine enough to support even smaller sensors than film.
I said that sensor size has NO effect on the image (or the speed) with equivalent camera systems; this is slightly complicated by the fact that camera manufacturers continued scaling down aperture (D) as they made smaller sensors, therefore currently-existing cameras with small sensors are not equivalent systems to currently-existing cameras with larger sensors, but this is because of bad engineering and marketing. DX-format cameras my all rights should have f/.7 lenses as standard, but they don't for cost reasons, and reasons related to the retention of the legacy SLR camera design. Camera manufacturers have spun this as 'larger sensors are better' which makes them more money.
Summary:The only way to get more photons to the sensor is to make the physical aperture (D) bigger, or to increase exposure time, or increase the scene lighting. But low-light photography is always a tradeoff between exposure time and depth-of-field. Sensor size is only a factor if a different size gets you a better fill factor AND that fill factor is enough of a boost to overcome the larger noise that is going to result from the lower current density.
You don't get a huge advantage in sensitivity over treating a block of pixels like a single pixel.
What you do get is a huge advantage in processing power, because you no longer need to use so much of it. If you're never going to need higher resolution on the sensor, then it makes sense to just make bigger pixels, because then you can throw the scaler away (or just implement a simpler one to handle downscaling to even lower resolutions, if needed.)
"You're right," Fisheye says. "I should have set it on 'whip' or 'chop.'"
After size, the big advantage of using 35mm is simply being able to use the overwhelming piles of 35mm lenses out there. If you have some big fancy camera you're going to need big fancy lenses. But if you can put the big fancy sensor in a normal camera, then you've got a whole new market there; people who can afford or already have 35mm lenses but can't afford all the kit for a bigger, more expensive camera. That's who this sensor is for. I can't help but think that it would be an excellent thing to sell to schools at only a nominal profit. Then students could buy an SLR-format camera with the same sensor and accepting the same lenses as what they learned on, when they graduated. Get a sort of Apple-like function going on.
"You're right," Fisheye says. "I should have set it on 'whip' or 'chop.'"
Sometimes I really do wonder whether /. is worth the trouble.
It's not worth trouble. If you're not having fun, play another game.
"You're right," Fisheye says. "I should have set it on 'whip' or 'chop.'"
If they could add an additional pixel that's sensitive to IR only and be able to switch it on and off, that would be really useful.
In addition, the fact that making a GOOD large-aperture lens is difficult and expensive - moreso than making a good high-sensitivity sensor.
retrorocket.o not found, launch anyway?
Issues with this:
1) Readout speed. I'm guessing that the 18MP sensor's readout speed is not fast enough to do 3x3 binning - instead there's a good chance that it only selectively reads pixels.
2) There's always a little bit of gap between pixels. As the pixel size gets smaller, this gap becomes a larger percentage of area.
A native 2.1MP sensor greatly reduces readout speed challenges and wasted intra-pixel space.
retrorocket.o not found, launch anyway?
This is a huge deal to the astrophotog community. This new sensor will greatly reduce exposure times and required mirror sizes for doing decent astrophotography.
This was taken sitting on the top of the pyramid of Cheops in Cairo a few weeks ago.
Gibbous moon, 20s exposure - it looks like day.
http://travel.ninjito.com/2013-02-19-Egypt/SLR/qx-cairo-9.jpg
Agreed, but the sibling post says that the math can't work out. It sounds like a better consumer-grade camera sensor, not a professional one if it's not 4K (for a 2014-2015 product that would be essential).
This seems very strange - they just made larger CMOS sensors, as far as I can tell. So, yeah, more photons per sensor.
My God, it's Full of Source!
OUTSIDE_IP=$(dig +short my.ip @outsideip.net)
I was working on high speed cameras for scanning microscopes before I retired and the noise for the sensors was given in equivelant rms noise electrons ie rms charge. At that epoch using cooled sensors and correllated double sampling about 7 rms electrons was possible. Do Canon publish any specs like this or does anyone know.
Making a thing bigger is rarely seen as an engineering breakthrough. This being so, what exactly has Canon done that was not possible before? It has always been known that the bigger the pixel-sensor, the more area it has to capture photons, and thus the more sensitive it becomes to light.
Obviously, with real film, we had 70mm, and then IMAX formats for extreme fidelity in moving image capture. 70mm has actually been (effectively) dead for many years now, and the most recent film to be shot in 70mm couldn't find a commercial cinema to show that format.
A 35mm sensor would certainly 'bloat' the front of the video camera, not least because of the need to support large pro lenses. At only 1080P, this would produce a 'difficult' commercial product, as 4K and 8K cameras hit the prosumer market. True professionals 'light' their sets. Casual users would appreciate a camera with excellent performance in all light conditions, but won't accept a prosumer design (or cost).
Sold in a VERY basic configuration (just the shell and a direct PC interface) this might make a nice tool for a certain class of hobbyist astronomers, and I suppose it has voyeuristic possibilities as well, given that it will give good results from much more powerful zoom lenses. Indeed, maybe this is the answer- the technology used in spy satellites is finally being offered to the plebs in a somewhat reduced form.
Note that the difference between 75% and 85% quantum efficiency is not that great.
Not that great, but for low light stuff like astronomy and microscopy, you want every advantage you can get.
If the pixels at 7.5 the area of a competing CCD, then the CMOS would get (7.5 x 0.75 / 0.85 = 6.6 photons for every photon the CCD got; i.e., the exposures could go down by a factor of 6.6. Even if the CCD had 100% efficiency, that would still be a factor of 5.6.
Not quite. The size of the lens and more specifically the back focal plane determine how much light is collected. As long a you can focus it all onto the smaller pixels without losing resolution due to diffraction limiting, you should be fine.
Large pixels do help the sensitivity a bit because they improve the ratio of light collecting elements to other crap. Though, microlenses also help in that regard.
SJW n. One who posts facts.
Things have indeed changed in the last five or so years. Some manufacturers build "back illuminated" CMOS or CCD sensors where the "infrastructure" is on one side and the light collection takes place on the other side. The silicon is ground thin enough that the resulting charges can be detected on the "top" side. I don't know if any kind of silicon etching/processing/... is performed on the side that points towards the light, or what other process they use to turn the silicon into separate pixels.
They already have processors that easily handle the processing aspect, so I don't think that's really it. No, this is actually aimed at improving the quality of the video recorded. When you use a whole bunch of tiny, high-sensitivity sensors, you get a lot of noise in low-light conditions. You just aren't getting enough photons hitting each sensor to create a good signal - highly variable, noisy images are the result. You can overcome this somewhat by averaging a bunch of sensors together, but you're basically averaging a bunch of known bad data in hopes of creating good data, and the outcome is often much less than satisfactory. This doesn't matter so much in still photography, where you can just keep the shutter open a little longer to collect more photons, but in video there are limits to how long you can open your shutter for each frame.
By using larger sensors, each one is intercepting a lot more photons and given the same sensitivity constraints it will create a much better signal. Anandtech recently did an article relating to this, although they were looking at cell phone cameras and one company that is deliberately decreasing MP in exchange for larger sensors in order to improve image quality for stills and video (though it is a presentation one of their writers gave, and doesn't go into a whole lot of the theory of why fewer, larger sensors can give better results than more, smaller sensors).
Science grade sensors are not cheap, but they're not outrageous. I recall working with engineering grade devices (a decade ago), where the science-grade devices would end up costing our collaboration $500k *each* (since we were effectively being charged for R&D). These were particularly expensive. They were 2k x 2k infared foacal plane arrays that were being calibrated for use on a space telescope. They were ~20 micron x 20 micron pixels as well. They also were operated around 140 Kelvin or so. :P
Illuminated by Earthlight.
I can mend the break of day, heal a broken heart, and provide temporary relief to nymphomaniacs.
What I'd like to see is a DSLR in which the mirror can flip in a different direction to expose one of these sensors under ultra-low-light conditions. That way you could choose, at the time you are taking the shot, whether to prioritize resolution or light gathering.
Check out my sci-fi/humor trilogy at PatriotsBooks.
Except that 35mm isn't 'added for backwards compatibility'. It's still a viable format, it's still in use, it matches a standard professional digital camera sensor size and it's a common reference point across the industry.
Then there's
a 35mm lens
A 35mm lens has a focal length of 35mm. That has absolutely no bearing on sensor size; you can put a 35mm lens in front of a mobile phone sensor or in front of a Leica S2. It's still a 35mm lens.
Of course, you can also put a 35mm (35mm equivalent) lens in front of any sensor. On a mobile phone that's going to be about 4mm in focal length and on a Red medium format could be 50-60mm. Suddenly we're seeing the benefit of using the 35mm reference point, even though we've got a magnitude of difference between actual focal length, they're all giving comparable field of view.
Depth of field, that's a different story. The larger your CCD the narrower your depth of field at 35mm equiv.
The irony is that it's a step back to ye olde times (about a decade ago) in digital SLR pixel counts.
The difference is that they're using modern sensor technology, which means it's a fantastic idea for non-professional video (and possibly as a second-unit camera for some pro stuff). It'll even take stunning still photographs; they just wont look so great printed above around A4 size.
Don't forget the benefits in physical camera size. I shoot Micro Four-Thirds so that my camera with lens attached fits into a coat pocket. People just don't generally want to lug around a heavy medium format camera with humungous lenses.
It could also just mean that they bumped the analog gain of the pre-A/D converter signal amplifiers on the chip up as compared to the previous shots.
There are tradeoffs involved in lens size and sensor size. To get the same S/N and image as a 50mm f/2 lens on a 36x24mm sensor, on an 18x12mm sensor a 25mm f/1 lens with twice the cycles/mm resolution would be required. Design of such lenses is difficult, and design and production of low f-number zoom lenses is extremely difficult. (Also keep in mind that it is impossible to have a lens faster than f/0.5). Signal integrity issues with a smaller sensor are more difficult.
In my opinion, the optimum for high quality with acceptable cost is smaller than 36x24mm, but not much smaller. Small sensors used in point-and-shoots are annoyingly noisy in only moderately dark situations.
Contribute to civilization: ari.aynrand.org/donate
The other way. You'd be able to use a quicker shutter speed due to the increased light capture.
If you want long exposure times, just fit a neutral density filter.
I'm interested in its dynamic range, and whether that's better or worse, but the main point of the sensor is the ability to capture usable images in near darkness. I have several good uses for that!
There is a common misconception that somehow CCD imagers are superior to CMOS imagers in general. In reality, the only main advantage of CCD imagers is that the process technology available to them have the ability to be more sensitive in the near IR range (by increasing the epi-layer thickness). In addition CCD imagers have some special techiques available to them for low-noise gain (electron multiplication and temporal delay integration) that don't have similar low-noise counterparts in CMOS imager (since CMOS conversion circuits are more voltage oriented than electron/current oriented and doing circuit operation on voltages is more noisy than on currents/electrons).
However, if your application doesn't avail themselves of using these techniques most of which are only interesting in extremely-low-light applications (like astronomy), many digital CCD sensor images will end up inferior to their CMOS counterparts. On an CCD, the analog/digital conversion circuitry generally need to have higher bandwidth (because they can't be integrated as easily with the sensor array, you are wire limited and must timeshare the wires for readout and thus need to read-out faster) and higher bandwidth A/Ds have more noise negating much of the CCD advantages. In a CMOS sensor counterparts, the a/d conversion circuitry is more parallel and can be designed for lower-bandwidth and thus lower overall noise (because you aren't amplifying the high-frequency noise).
There is also the general result that CMOS sensor development rides on the coattails of high-volume production characterization of parametric values (because of cell-phones), whereas CCDs being much higher power current-operation devices are not as economically attractive to high-volume markets and thus tend to have low-production runs and suffer more variance in parameteric values.
There is nothing in the article that suggests that this is a Bayer-type system. In fact, one of the theoretically easiest ways to improve S/N is to not filter out 2/3 of the signal like Bayer does.
Contribute to civilization: ari.aynrand.org/donate
Larger pixels do not 'collect more light' in real camera systems
Yes, they do. But skipping past some flawed analysis and misunderstanding,
Summary:The only way to get more photons to the sensor is to make the physical aperture (D) bigger, or to increase exposure time, or increase the scene lighting.
Actually, yes, I'd agree.
Photons : Sensor Ratio at fixed aperture, fixed exposure and fixed lighting is itself fixed for a given sensor size, no matter how many pixels there are.
Now, do the maths. If that ratio is fixed then how many photons hit each pixel if the sensor has 18 megapixels, as compared to the same size sensor having only 3 megapixels.
Larger pixels DO "collect" more light because they have greater surface area and a greater share of the total light hitting the sensor. The sensor as a whole has a fixed amount of light hitting it, but no longer has to split that light between as many pixels.
This means better quality interpretation of the light.
since there is NO practical effect on real camera systems, you might as well use a small sensor, since digital sensor pixel pitches are fine enough to support even smaller sensors than film.
The issue isn't how finely you can split the light to make 18 million readings. It's how accurate those readings are. When shooting using the illumination caused by the glow of a incense stick (as described in the article) you need some pretty sensitive light detection to avoid losing your entire image in noise.
The larger pixels give you that sensitive light detection, by capturing more light per pixel for a given sensor size.
...why kids love Cinnamon Toast Crunch?
For some time we've had cameras, even ones capable of video, that could record single photons. It comes down to resolution, speed, and mostly just cost. If you want huge sensitivity, you can go with photomultiplier tubes, which are quite unwieldy for a large array. But even on an intensified CCD camera, I've had images where you could quite clearly count individual photons coming in. It just costs a lot to get a large number of video frames out of an ICCD camera, as in 6 to 7 digits depending on how many frames you need (although easily at 1+ million fps).
Thats a great presentation video. If it can deliver what it shows in the video, its really good.
Thinking about it some more, there is a professional space where it could come in handy - wedding videographers and the like. Anywhere where the light will be bad and there's nothing you can do about that, and there's not much need for post-production or delivery to any kind of device that runs > 1080p.
For those uses, the light sensitivity should be worth all the other trade-offs.
My God, it's Full of Source!
OUTSIDE_IP=$(dig +short my.ip @outsideip.net)
Alas, the ISO rating for digital sensors is bogus. Double the amplification and you double the rating, but also the noise (except converter quantization noise.) If adequate gain is available, it's the improvement in S/N that is most important area for improvement.
Contribute to civilization: ari.aynrand.org/donate
I did ponder documentary uses - I guess weddings count in that :)
But you're right, anywhere that you can't control lighting is a good candidate, especially if you can't bring in a fullsize TV camera.
After size, the big advantage of using 35mm is simply being able to use the overwhelming piles of 35mm lenses out there.
Don't forget the benefits in physical camera size.
Don't worry, I promise, I won't.
"You're right," Fisheye says. "I should have set it on 'whip' or 'chop.'"
Shows why you need to embrace and innovate in new areas as technology advances. Too bad Kodak didn't figure out this lesson, we see where they went. Bravo, Canon! I hope they put that in their powershot line, I'm on my 3rd powershot and it suits me very well.
I work for the Department of Redundancy Department.
Bah. You win this time, but I'll be watching you.
(I wont, but hopefully you'll feel better thinking I might)
Agree 100%. I also shoot m4/3, and love the smaller size. The difficult part is putting up with high ISO noise (gradually improving with m4/3, but not quickly enough!) which this lens could help fix.
4K video is not 4 Megapixels. It's 3840 pixels × 2160 pixels, or about 8 Megapixels. The 4K is the horizontal resolution (3840 ~= 4k)
I went for the E-PL5. High iso noise isn't upsetting me at all, although I have a max ISO of around 2600 set.
The sensor is actually superb, although obviously nothing like the lowlight capability of this new Canon one. There are issues with the camera but broadly they can be easily summarised as 'user error'.
Binning helps, but one large photosite has more surface area than an equivalent-sized 3x3 set of photosites, because of the gaps between them.
I emailed Schneider Kreuznach once, asking what image sensor pixel size would be appropriate for some of their older lenses designed for use with film, and they told me that 15 micron pixels would match those lenses. So, if a film maker has a favorite older lens from the film days, it might do ok with this new sensor.
Sensitivity is adjusted by the ISO setting.
the newly developed CMOS sensor is capable of capturing video such as this.
This is intended to read as "here, have some footage we took earlier, it is possible for you to do this too.
You could also build a camera out of a medical X-ray sensor, with pixel size around 150 micron. Sensors are typically 43x43cm (17x17 inch) wide, which might be a problem.
In a (D)SLR there is a single light path with a mirror in the way. Moving the mirror out of the way does nothing but restore the single light path. There's nowhere to put a second sensor. However, most serious users of interchangeable lens cameras will have multiple bodies, so I just expect people to carry around another, high sensitivity, body in their bag. Presumably with their f/1.2 50mm on it, so they're ready to just start grabbing photons.
Also FatPhil on SoylentNews, id 863
I think you missed my point. There's no law that says the mirror must have only a single pivot point. You could make a mirror that pivots in the middle, for example, and mount a second sensor down near the bottom of the camera. So in normal mode, the mirror flips up and exposes the sensor, and in low-light mode, the mirror flips 90 degrees and reflects the light down instead of up. Because a 45 degree mirror preserves the distance from each pixel to the lens, the only thing you'd have to do is vertically flip the resulting image, which is trivial to do in software.
Of course, it would be beneficial for the mirror in such a design to be double-sided so that the face that becomes part of the optical path for the second sensor would be moderately protected from scratches and dust. You would probably also want some sort of flip-down or slide-in black cover over the second sensor so that it would not produce reflections of off-axis light on the main sensor (which could remain continuously in the covered position until the second sensor is needed). None of those design constraints seems particularly insurmountable (or even all that difficult) mechanically, though, unless there's something subtle that I'm missing.
As for having multiple bodies, I do, but since I'm not shooting professionally, I usually leave my spare back in the hotel room. If my main camera body fails, I can always fall back on my iPhone. Sure, that's less than ideal, but it just isn't worth carrying around that much extra bulk. My lenses are heavy enough to be obnoxious by themselves without adding an extra 1.25 pounds for a spare body.
Check out my sci-fi/humor trilogy at PatriotsBooks.
"but that's not a problem for Canon, they have made medium format cameras in the past." Ummm, and just when might that have been???
Nikon D800 has 36 Megapixels packed into a 35mm sensor
This camera has 2.1Mega pixels crammed into the same size sensor.
My Aurora : http://www.youtube.com/watch?v=o91ZsGwJYyg
FB : https://www.facebook.com/TanveersPhotography