Previous news story    Next news story

Canon releases sample video from high-sensitivity full frame CMOS sensor

By dpreview staff on Sep 13, 2013 at 17:38 GMT

Canon has released a short video produced with a unique full frame CMOS sensor. Announced in March, it's a sensor dedicated to extreme low light video recording with unusually large pixels. The test subject? A group of Yaeyama-hime fireflies on Japan's Ishigaki Island. The sample video was captured without artificial lighting in just 0.01 lux, a light level that challenges the naked eye. Beyond natural and astronomical applications, Canon says the high sensitivity sensor may one day be used for medical and security purposes.

Canon staff used a prototype camera carrying the new CMOS sensor to produce a sample video featuring an extremely dark scene. The 35mm full frame CMOS sensor uses a 1920 x 1080 pixel array with individual pixels measuring a whopping 19μm.

The high sensitivity chip uses large pixels and low readout-noise circuitry, allowing it to capture light around 10 times less bright than current CCDs used for astronomy. Each pixel is 7.5 times larger than those of the Canon EOS 1DX.

Via: Canon Rumors, Source: Canon Global

Comments

Total comments: 228
12
refillable
By refillable (1 week ago)

...Cough... Sony ...Cough... A7s ...Cough...

0 upvotes
nimrod1212
By nimrod1212 (6 months ago)

One of these would be great to have in a general purpose large sensor HD video camera, instead of DSLR. No more interpolation of too numerous pixels and undreamed of low-light performance.

0 upvotes
yabokkie
By yabokkie (7 months ago)

does anyone have experience with this 2/3" camera?
http://www.panasonic.com/business/provideo/AJ-SDC615.asp

0 upvotes
guatitamasluz
By guatitamasluz (7 months ago)

"Sensor like this one has probably a NATIVE ISO of 204.800 or 102.400, so, YES, it is impressive in every way." ... Just two words: great news, if these ISO are correct.

0 upvotes
Macx
By Macx (7 months ago)

The interesting bit is how noisy the images will be at those ISOs. The numbers themselves are sort of meaningless in this context.

0 upvotes
yabokkie
By yabokkie (7 months ago)

0.01 lux is about the illuminance that all of our vision is gone. we can see nothing put on the ground in this darkness (maybe except a large piece of mirror or shiny metal).

and the slowest shutter speed is 1/24s.

0 upvotes
Francis Carver
By Francis Carver (7 months ago)

In other words..... when you stumble along in total darkness out there with this wunder-sensor housed in a camera, you are probably going to be breaking your leg and/or neck BEFORE you line up that first shot, huh?

0 upvotes
Looca
By Looca (7 months ago)

It's interesting how some people (those who know nothing about videography) keep talking about 4K that nobody uses. For those who do not know yet, NOBODY uses 4K video recording for TV ads, documentaries, or anything else except hollywood movies, which are still mostly done with film cameras btw. 4K cameras are nothing else than early bait for enthusiast and amateurs to pay off entire 4K resolution project in advance. 4K isn't supported by anything anywhere, and it won't be any time soon.

So, to sort this about resolution, it's obvious that Canon was thinking about pro's, not amateur videographers who are chasing resolution instead of a good frame.

I'll just try to explain you this in numbers, just from my point of view:
To capture light of 0.01 lux, even if it had an aperture of f1.0, it would need something like ISO-409.600, even if the sensor refresh is as low as 25.
Sensor like this one has probably a NATIVE ISO of 204.800 or 102.400, so, YES, it is impressive in every way.

1 upvote
yabokkie
By yabokkie (7 months ago)

better we never go 4K for 8K is coming in 6-7 years.

0 upvotes
Francis Carver
By Francis Carver (7 months ago)

These days, poor videographers might still have to shoot in HD.... whilst somewhat better-off videographers can now afford to shoot with any number of (Sony's) 4K cameras. And yes, after 4K video, we will have 8K video. Looking ahead -- instead of rearward, that is.

Comment edited 16 seconds after posting
0 upvotes
Francis Carver
By Francis Carver (7 months ago)

Canon just can't seem to get it right. For starters, a 1920 x 1080 pixel resoltuion sensor, in late 20013, is considered a LOW RESOLUTION imager by most folks into digital photography.

Now, let's assume for a second that Canon was planning to further develop and then use in video cameras this experimental low-light friendly sensor of theirs. In that case, they picked an arcane resolution, as the world of digital videography and filmmaking is fast moving away from good old 2.2MP HD sensors and towards 8.8MP 4K sensors.

I suppose if they had made the sensor's resolution to something like 640 x 360, then the individual pixel photosites would have been even larger and the sensor even more light sensitive. This thing would have made sense 8-10 years ago -- but today?

1 upvote
Plastek
By Plastek (7 months ago)

"is considered a LOW RESOLUTION imager by most folks into digital photography" - that's because it's NOT a sensor for digital photography.

"hey picked an arcane resolution" - arcane?! What are you talking about? It's currently most commonly used resolution, and it will be like that for at least new next years. What world do you live in? lol

6 upvotes
Francis Carver
By Francis Carver (7 months ago)

Who even knows what this is -- but whatever it is, it came a few years ex post facto, it seems. Assuming it will make it into a practical picture or video taking device one day, it will be noted for its low-light recording abilities.

Unfortunately, for Canon video is still HD for resolution and 30 frames per second maximum for refresh rate/frame rate. Check out their EOS Cinema C100 and C300 models -- yeah, 1080p30. Whereas for Sony, they have 4K at 60p = 2160p60. Big difference, you know. Not sure what Canon is up to with this one, but if, like you said, Plastek, this is indeed NOT a sensor for digital photography -- then pray tell, why are we reading about it here on DIGITAL PHOTOGRAPHY Review?

0 upvotes
JackM
By JackM (7 months ago)

Francis you ignorant dolt.

3 upvotes
Looca
By Looca (7 months ago)

Relax Francis, you're obviously not a videographer, this is not a sensor for you. Those who are videographers do not talk much about 4K resolution, because it does not exist.
Technically, even in the very cameras that do have 4K recording, it's half-interpolated :D

But honestly when I see this, I would not mind to have 1080p sensor for PHOTOGRAPHY if it could do this low light, it would be funny. I remember getting spectacular prints from 3mp old cameras :)

1 upvote
Francis Carver
By Francis Carver (7 months ago)

Okay, Looca, so Canon here has announced an "experimental" sensor with 1920 x 1080 photosite resolution. Problem is, we have got those in spades already for over a decade now, really we do.

When it comes to video -- Canon's world in the autumn of 2013 consists of 1080p HD and 30 frames per second maximum refresh rate. Fortunately for us, the world of other mfrs include other stuff as well.

Okay, so this low-light sensor is NOT for videographers, like you have said -- in that case, why the strangely videoish 1920 x 1080 pixel count?

"Technically, even in the very cameras that do have 4K recording, it's half-interpolated."

Okay, Looca, but only because YOU said so.

"I would not mind to have 1080p sensor for PHOTOGRAPHY...I remember getting spectacular prints from 3mp old cameras."

Unfortunately, Looca, this sensor would not even be a 3MP sensor.... more like a 2.2MP sensor. I guess some pro photographers dream about that still.... while others have moved onto the 2010s.

0 upvotes
SRT3lkt
By SRT3lkt (7 months ago)

comments here are unnatural

2 upvotes
Rodrigue Zahr
By Rodrigue Zahr (7 months ago)

amazing innovation, can't wait to see it implemented in future DSLRs

2 upvotes
Andy Crowe
By Andy Crowe (7 months ago)

It's only a 2 megapixel sensor so it's far more suitable for video than for stills.

0 upvotes
Plastek
By Plastek (7 months ago)

I'm quite sure that ppl would find thousands of uses for camera that can shoot in pitch black.
From paparazzi market all the way to nature photography. And god knows what else.

0 upvotes
Francis Carver
By Francis Carver (7 months ago)

"I'm quite sure that ppl would find thousands of uses for camera that can shoot in pitch black."

Question for you, Plastek: if it is indeed so dark out there that it is pitch dark, then pray tell, how you gonna SEE with your own eyes what you are supposed to be shooting in the first place? See the fallacy of this whole heady concept here?

0 upvotes
Lord metroid
By Lord metroid (7 months ago)

Night vision googles.

0 upvotes
Francis Carver
By Francis Carver (7 months ago)

Wow, a "Full-HD" 1920 x 1080 pixel count sensor for videography, how revolutionary! Quick, someone tell Canon that Sony already has 6 different models of digital video cameras/camcorders, each with native 4K resolution video recording capabilities. And Canon is now experimenting with the previous generation resolution HD? Wow....

1 upvote
JamesHurst
By JamesHurst (7 months ago)

Evidently you didn't comprehend what this is. This is not a high-pixel-resolution camera - it's design to be sensitive to low levels of ambient light. To achieve that requires LOW resolution so that the individual pixels (the light-sensors) can be BIG. Which is precisely what a lot of us have been clamoring for. For most of us, any resolution above HD is just a waste.

9 upvotes
Nukunukoo
By Nukunukoo (7 months ago)

It's not the resolution, it's the light sensitivity and wider dynamic range. In fact, the "low" HD resolution is the big feature if you understood the context of the article.

Would love to see one in APS-C as well. Been spoilt with a loaner Pocket Cinema at Super-16. The RAW and Dynamic range where color grading is a pleasure! Too bad for the orbs and black dots (deal breaker).

8 upvotes
Andy Crowe
By Andy Crowe (7 months ago)

And now show us a sensor that can record at 0.01 lux.

2 upvotes
Francis Carver
By Francis Carver (7 months ago)

First off, this is not a camera, rather, an experimental CMOS sensor. They have been around for many, many years now, right?

This new sensor, if ever implemented by the company behind it in an actual video camera, will be able to handle low light fine. Problem is, Canon is still stuck with HD resolution video and with 30 frames per second video.... while Sony is offering 4K resolution video cameras and camcorders that can record at 60 frames per second... and another model that does Full-HD at 240 fps.

I don't care if this experimental sensor of theirs records at 0.0000001 lux of available light, Canon's HD resolution and 30 fps maximum frame rates no longer cut it for us, videographers.

0 upvotes
wflan
By wflan (7 months ago)

@Francis

Show me a monitor capable of displaying 4k. Show me a good delivery method to distribute your ultra-hd footage. Show me a real use that doesn't involve IMAX, cropping, or down-sampling.

This sensor was never intended to to ultra-hd slow mo recording. What it's good at lies outside that realm, and the sensors competitive in that spec range are not good at what this one excels at. You are so off base, I kinda suspect you are trolling.

This is for creative, scientific, and medical fields, and will be extremely useful therein. Yeah, you have to use legs/lenses to change your framing, rather than a post-production crop. To shoot a naturally lit scene on a lake under a cloudy moon, in vivid color trumps those other technical abilities. For this sensor, in its specialized applications.

3 upvotes
Mike_V
By Mike_V (7 months ago)

4K monitors are available now.

Consumer models are being released very soon.

4K streaming is available now with 4K Bluray available very soon. It seems like the Bluray is going to be pretty heavily compressed though.

4K is the next big consumer electronics push.
I'm not sure if it will really take off, but the the manufacturers are going to give it a red hot go at making it a success.

1 upvote
deemoss
By deemoss (7 months ago)

@wflan
Actually, ultra slow motion requires very bright lighting since frame exposures are so short. This sensor is super sensitive and if it would allow quick data readouts then slowmo could be a good match for it...

0 upvotes
Francis Carver
By Francis Carver (7 months ago)

@ wflan: "Show me a monitor capable of displaying 4k."

Well, I can show you one... or else I can show you dozens of native 4K resolution flat panel displays, and more than a dozen 4K resolution projectors available today. Why just pick one?

Sony is streaming 4K content already, as does a satellite provider. For Canon to have an "experimental" sensor at 1920 x 1080 resolution, no matter how low-light sensitive, and to have this announced in September 2013, just shows that they do not get today's videography and digital cinematography trends at all. In fact, Canon as a company has dozens (hundreds?) of video recording capable digital cameras and camcorders out there now.... but only one (EOS Cinema C500) that they claim to be 4K ready. And even that one is "4K ready" only if you teeter it into a third-party external video recorder.

This new sensor would be for the military and nightlife shooters -- and probably even military shooters would prefer to shoot in 4K now.

0 upvotes
Anastigmat
By Anastigmat (7 months ago)

Finally, we will get a camera that can photograph a black cat at night inside a coal mine.

12 upvotes
AstroStan
By AstroStan (7 months ago)

I've not seen any useful specs for this chip (esp QE and readnoise). But Canon implies that it has very low noise, in which case it may be their version of "sCMOS" (scientific CMOS). Google/Bing "sCMOS"

If the Canon chip is competitive with and similar to sCMOS then many people here may be greatly disappointed because those sensors are very expensive. The sCMOS sensor alone originally cost over $10,000, though I recently heard it has dropped to just under $7,000 (chip only).

Maybe Canon has pioneered a revolutionary manufacturing process to match sCMOS quality for less $. Or maybe the Canon CMOS is not as good as implied (these promos never have any details about the lens and such). It will be interesting to see if anything ever becomes of this...

0 upvotes
AstroStan
By AstroStan (7 months ago)

On the other hand, that performance may be possible by simply reducing the number of pixels in an otherwise normal fixed size sensor (FF). A FF sensor with 2MP (mega-pixels) will generate about 1/3 the angular read noise of a 20MP FF sensor (10x more pixels = sqrt(10) more noise).

If that's the case then that sensor can be put into cheap DSLR &/or video cam with a niche market that may or may not generate enough profit to be worthwhile for such a large company. I would probably get one. And I would definitely buy a full-spectrum non-Bayer version!

Comment edited 2 minutes after posting
0 upvotes
yabokkie
By yabokkie (7 months ago)

> by simply reducing the number of pixels

it cannot be simply reduced, not with extra effort which is why Canon's announcement "we made a great achievement."

higher performance is possible by increasing the number of pixels, not simply but easier than going the opposite direction.

Comment edited 25 seconds after posting
0 upvotes
AstroStan
By AstroStan (7 months ago)

by "simply" I meant the simple math of holding read-noise constant while decreasing the pixel count (i.e. increasing pixel size). Indeed the promo material boasts that they are able to increase the pixel size without increasing noise. But it does not say that the noise is actually decreased (compared to smaller normal pixels). So upon a more careful reading it does appear that this chip may not compete with sCMOS.

But it is still a potentially important achievement. In re-reading the original press release I see that it was to debut at a "security conference". Many high end security camera are very pricey. But of course there are also many super-cheap ones...

Comment edited 2 times, last edit 6 minutes after posting
2 upvotes
yabokkie
By yabokkie (7 months ago)

looks we talked about different things, image quality or pixel quality. image quality is measured against (a fraction of) the whole image and won't change along with pixel size.

Comment edited 1 minute after posting
0 upvotes
Priaptor
By Priaptor (7 months ago)

Now that is awesome. Awesome enough to get me back into shooting wildlife in the dark

1 upvote
ThePhilips
By ThePhilips (7 months ago)

Pretty cool stuff. Some scientist should probably start preparing cash to part with.

Would have been even cooler is it hasn't used the Bayer filter. But that tech is apparently way underdeveloped yet.

0 upvotes
ARTASHES
By ARTASHES (7 months ago)

with my limited acknowledges about sensor I think this one isn't any better than any actual FF sensor on the market, but the it's advantage is that it will use total surface of the sensor to create an fullhd video because the whole sensor has exactly the same resolution as the fullhd while the 20mp+ sensors can't do this, in fact to create a fullhd D800 for example will have to take only several pixels from given surface to avoid re-sizing from 36 mp and so it uses only some part of whole surface that's impact the IQ in low light, conclusion : the size of individual pixels is important for video making but is irrelevant for photography where you can resize and so this penalize big resolution sensor for video making and that's why Canon made this one with such big pixels

2 upvotes
Anastigmat
By Anastigmat (7 months ago)

To have unusually large pixels on a full frame sensor, you just reduce the number of pixels on the sensor. It is very simple, folks. Of course, fewer pixels means less resolution when taking still photos, but if you use the camera to capture videos only, then it makes sense, because the cameras doesn't throw away information when recording videos.

3 upvotes
Andy Crowe
By Andy Crowe (7 months ago)

With a bayer sensor it is advantageous to be able to record an image at higher resolution than you "need" because it makes up for the sharpness loss of having to spread light over several different coloured pixels (the AA filter)

0 upvotes
deemoss
By deemoss (7 months ago)

On a FF dslr sensor when you record a movie in HD the frames are normally grabbed from the whole sensor not just a cropped portion. Then the frames are down sampled.

HOWEVER, in theory it makes sense that if you consider a 1080 HD movie on a FF sensor, the pixel size should not matter - only the sensor size! The light gathering ability of he sensor is roughly related to the surface area illuminated. What difference does it make if you gather the light of one MOVIE-pixel from one large SENSOR-pixel or ten smaller SENSOR-pixels? In this example you are dividing the total sensor area by 2 million MOVIE-pixels.

I will speculate that you could get similar performance from the D800 sensor or any other FF dslr by just through a firmware update, doing "sensitivity-optimised" interpolation.

0 upvotes
Silvarum
By Silvarum (7 months ago)

"grabbed from the whole sensor not just a cropped portion" Not a cropped portion, but not the whole sensor either. Just skips some pixels to keep bit-rate in check.

0 upvotes
Steve Ives
By Steve Ives (7 months ago)

I have no doubt that Canon will be sacking all of their sensor R&D staff and several people here will be receiving job offers from them, on account that they obviously know a lot more about the subject!

5 upvotes
JJ Rodin
By JJ Rodin (7 months ago)

The comments here blow me away! I noticed when this was announced earlier that many of the lower luminance IQ crowd did not comprehend the ramifications. Even now the smooth knife ppl do not see the advantages of such a sensor. Just stick with your P&S 1/2.3" & 1/3" video sensors and be happy!

The grownups will lovingly use this sensor and astound those with only a handful of synapse, unless drunk or drugged out, in that case watch your sports and forget photography/video, plz.

Surprised that this type sensor is not made by all major players, much more useful than those 1/3" 'professional' sensors! :)

4 upvotes
Francis Carver
By Francis Carver (7 months ago)

Another full-frame sensor from Canon. Yippy doo!

0 upvotes
Plastek
By Plastek (7 months ago)

Money well spent.
They don't waste time with 1" sensors like Nikon does.

1 upvote
Funduro
By Funduro (7 months ago)

Woah, that's some super duper sensor there !

3 upvotes
Sangster
By Sangster (7 months ago)

Ground breaking. Infrared camera would not have worked since fireflies use bioluminescent flashes. Looking forward to commercialization of this technology.

6 upvotes
Andy Crowe
By Andy Crowe (7 months ago)

And with IR you don't capture the natural colours either.

1 upvote
petepictures
By petepictures (7 months ago)

Bravo Canon , push the limits

2 upvotes
jdc562
By jdc562 (7 months ago)

Way too much noise for my use. What was gained here?

0 upvotes
zos xavius
By zos xavius (7 months ago)

Its shooting in pitch blackness and you complain about noise. Wow.

7 upvotes
yabokkie
By yabokkie (7 months ago)

our fully dark adapted vision (after some 30 minutes in the dark) can barely see the outlines of large objects (trunks but hardly branches if not against the sky) at illumination 2 or 3 stops brighter than this.

2 upvotes
samhain
By samhain (7 months ago)

@jdc562 you have no idea what your even looking at, do you?

1 upvote
Francis Carver
By Francis Carver (7 months ago)

Or else just use an LED light to shine on your subject, they are not that pricey any more, you know.

0 upvotes
ageha
By ageha (7 months ago)

Pretty cool!

0 upvotes
Photato
By Photato (7 months ago)

Nokia can do it better by combining 41 small megapixels ;-)

0 upvotes
HowaboutRAW
By HowaboutRAW (7 months ago)

Except the 808 can’t match the Canon 5D Mark II for high ISO still shooting.

The 1040 produces stills that look to me near unuseable at ISO 1600.

So this Nokia+more small pixels claim equals better image quality is getting tired, particularly since those Nokia cameras don’t shoot raw. While all DSLRs shoot raw stills and some can shoot raw video.

Canon is right to pursue this. And particularly for still shooting it’s not like Nikon ever forgot to. Sony on the other hand.... And no Sony does not make the sensor in the Nikon D4 or the D3.

Or perhaps you were being ironical:)

2 upvotes
yabokkie
By yabokkie (7 months ago)

now phone cameras got 40MPix and I hope entry level P&S's should follow. premium P&S's can have 80MPix as well as Nikon CX, 160MPix for 4/3" to APS-C and 320MPix for full-frame.

0 upvotes
HowaboutRAW
By HowaboutRAW (7 months ago)

let's hope not unless electricity runs cold--perhaps that Panasonic prism thing will really work out.

0 upvotes
JadedGamer
By JadedGamer (7 months ago)

Well, all camera sensors shoot raw - they cannot help not to - but whether that raw data is made available to the user afterwards in a file format is what differs.

0 upvotes
HowaboutRAW
By HowaboutRAW (7 months ago)

JadeGamer--

Only sort of all cameras shoot raw, what you mean is that all sensors see raw.

Then after the sensor sees the raw data many cameras simply toss out the data and one gets a jpeg recorded.

So it's that not all cameras process and record raw, which effectively means that jpeg only cameras don't shoot raw.

Now look of course Canon allows hacks so jpeg only P&S Canons will shoot + record raw, so it's possible for Nokia to allow such a hack. Time for Nokia to start.

0 upvotes
guatitamasluz
By guatitamasluz (7 months ago)

if it won't be translated into DSLRs, I still hope that they have some other similar tech in their sleeve to equip upcoming 6D/5D and 1D bodies with...

1 upvote
cameron2
By cameron2 (7 months ago)

Pretty cool ..

0 upvotes
AstroStan
By AstroStan (7 months ago)

It's been some time since the Canon announced the first version of their large-pixel/low-noise sensor. But no products have appeared. Maybe this teaser is a hint of an actual product brewing?

If it does ever become a product, I think it is unlikely it will be encased in a DSLR. It sounds like they are positioning it as a scientific video cam. That could easily exceed $10,000. They may try to capture some scientific applications of EMCCD (typically costing $20,000-$40,000). But if so then they must produce a non-Bayer version.

1 upvote
Dave Luttmann
By Dave Luttmann (7 months ago)

And yet Canon can't design a sensor with decent dynamic range.

4 upvotes
Photomonkey
By Photomonkey (7 months ago)

Actually they do have decent range, its just that others have gotten better. What I feel is overlooked is that the public clamored for better low ISO performance and then when that came along they spin around and scream for better DR.
The sort of thing that drives manufacturers crazy.

I have yet to be convinced that the better DR of even MF cameras shows up in final images.

Comment edited 37 seconds after posting
3 upvotes
photogeek
By photogeek (7 months ago)

I would think this should make manufacturers happy. They get to sell stuff to you over, and over, and over again, extracting tens of thousands of dollars in revenue. :-)

0 upvotes
HowaboutRAW
By HowaboutRAW (7 months ago)

Photomonkey:

Dynamic range, colour generally, and even high ISO shooting can be vastly improved with a better lens--Zeiss in this case.

The Nikon D3s body that I used with both Zeiss and NikonED lenses demonstrated this difference really clearly--or not so clearly with the Nikon glass. (I'm sure the same basic thing applies to Canon+I'm not taking sides about Canon sensors' dynamic range.)

1 upvote
David Elliott Lewis
By David Elliott Lewis (7 months ago)

This low light sensor is an amazing technology that appears to extend photography into areas previously unavailable - providing high definition video in near total darkness - light so low that even the unaided eye cannot see.

I believe there is huge commercial potential including in motion picture production. I hope Canon will bring this sensor to market.

One suggestion would be to create a dual sensor camera containing both this new low light sensor and a traditional high resolution sensor mounted on a rotating device that can quickly switch between them. Possibly, the same type of mechanical device that moves neutral density filters into place could be used alternate between sensors.

P.S. The reason I am suggesting a dual sensor camera design is based on the assumption that it will not perform well under daylight conditions. With such a low pixel count, visual artifacts like moire might degrade the image.

Comment edited 9 minutes after posting
0 upvotes
David Elliott Lewis
By David Elliott Lewis (7 months ago)

Continuing in my quest for the ideal video camera, an even better design would be a 3 sensor camera with the third sensor dedicated to black and white imaging by eliminating the Bayer color filter.

0 upvotes
Eike Welk
By Eike Welk (7 months ago)

An other idea for switching sensors, would be a mirror.

This could be a new application for the mirror in SLR cameras, and would leverage the the properties of existing SLR-lenses in the age of the electronic viewfinder.

A possible camera could be constructed like this:

The high resolution sensor is in the top of the camera, light reaches it via the mirror. The high sensitive sensor is in the back of the camera, light reaches it when the mirror is moved upwards.

This kind of camera could even have a separate focus sensor, with a similar arrangement like in current SLRs: The primary mirror is half transparent. There is secondary mirror behind the primary mirror that directs the light downwards to the focus sensor, which is in the bottom of the camera.

0 upvotes
ET2
By ET2 (7 months ago)

This is a video camera (and even then special purpose video camera). For stills "large pixels" cameras like that will be a big failure.

5 upvotes
Barry
By Barry (7 months ago)

Hope this makes handheld night shots with very little light and no camera shake possible.

0 upvotes
tecnoworld
By tecnoworld (7 months ago)

In less than 10 years, from nikon D1s to D4, cameras gained almost 2 stops for high iso performance, not to mention the increased DR and color depth. I think the same trend will hold in the next 10 years, more or less, so there will be consumer cameras able to shoot noise free up to 12800 or even more and with acceptable noise for small/medium prints up to 102400 which is incredible for todays standards.

1 upvote
ageha
By ageha (7 months ago)

No

4 upvotes
Eric Hensel
By Eric Hensel (7 months ago)

Yes.

0 upvotes
NetMage
By NetMage (7 months ago)

You may want to study the quantum nature of light.

3 upvotes
zos xavius
By zos xavius (7 months ago)

Physics....worth studying....

1 upvote
Eric Hensel
By Eric Hensel (7 months ago)

Stay tuned, kids

0 upvotes
DUSTY LENS
By DUSTY LENS (7 months ago)

This is finally getting down to serious light gathering Photography , worthy of owning in a second specialized body . I love it .
So I would personally want a very Hi resolution Sensor body too , for good lighting uses . maybe 50MP to do the daylight quality photo work .

What could be a better reason for owning two bodies ?

Makes me think about the possibilities of offering a FF body having the ability of Sensors switching by the owner , in a clean clear plastic bag .

0 upvotes
Silvarum
By Silvarum (7 months ago)

Why would you bring video camera if you're doing photo work?

0 upvotes
pcblade
By pcblade (7 months ago)

Light painting failure...

0 upvotes
stevens37y
By stevens37y (7 months ago)

This will be never used in consumer cameras.

0 upvotes
Silvarum
By Silvarum (7 months ago)

Probably because consumers don't go shooting video in total darkness.

2 upvotes
Optimal Prime
By Optimal Prime (7 months ago)

Never say never.

3 upvotes
stevens37y
By stevens37y (7 months ago)

>> don't go shooting video in total darkness.
They don't have a camera for that.

0 upvotes
-Teo-
By -Teo- (7 months ago)

Now we need a level on the camera to switch day and night sensors :)

0 upvotes
T3
By T3 (7 months ago)

I remember there was a big discussion a while back where people were claiming that the key to lower noise was having more pixels, and of smaller size. Wonder where those people are now?

2 upvotes
Ola Forsslund
By Ola Forsslund (7 months ago)

To manage 30 fps, other FF-sensors SKIP a large portions of pixel in video mode, effectively reducing the light capturing area. THAT is the reason for lower performance.
It is the sensor read-out speed that is the limit to the number of pixels used in FF sensors. This can be seen in the low pixel count in the 1Dx and D4, as compared to the low-fps 5D and D800.

Comment edited 12 seconds after posting
9 upvotes
RP McMurphy
By RP McMurphy (7 months ago)

Those people are still here, laughing at you for failing to understand the science

5 upvotes
Eike Welk
By Eike Welk (7 months ago)

These people never knew what they were talking about. Or when they were from the industry, they tried to protect their work on ever smaller pixels.

Big pixels are necessary for high sensitivity - unless you can accurately count photons. But as long as there is read noise each pixel introduces additional noise into the image. You can reduce the read noise by averaging neighboring pixels but this is much less efficient than big pixels.

The only noise that can be efficiently reduced by averaging is shot noise. Shot noise is additional noise from quantum mechanics - the phenomenon that photons are individual particles that arrive at random.

1 upvote
Silvarum
By Silvarum (7 months ago)

@Eike Welk: Apparently it's you who don't know what you are talking about. It doesn't matter how much pixels you divide your sensor by. Quantity of light hitting the sensor will not change. As long as sensor size (and efficiency) remains the same, overall noise will not change as well.
By the way, "averaging neighboring pixels" (I believe correct term would be downsampling) in general produces much more pleasing noise.

2 upvotes
Eike Welk
By Eike Welk (7 months ago)

@Silvarum:

For each pixel, a certain amount of noise is created, each time an image is taken. This noise originates from imperfections in the analog circuitry and from the analog to digital converter. This noise is called "read noise". It is independent of the amount of light.

As you are writing about the amount of light, I think you are confusing read noise and shot noise.

Shot noise is an other source of noise that comes from the quantum nature of the world. The world itself is not smooth but made of particles that are subjected to random. Photons, for example, arrive at random points in time at a certain pixel. This random arrival of photons creates shot noise, which is significant in low light.

I write averaging because averaging is at the core of most noise reduction and down sampling algorithms. These algorithms usually contain smoothing steps which are usually weighted averages of neighboring pixels.

1 upvote
T3
By T3 (7 months ago)

Well, I guess the "smaller pixels are better for less image noise" fanatics are still around. LOL. The proof is clearly in the pudding. But they still want to cling to theory that is unsubstantiated by product results.

Clearly, these people don't understand that, no matter how small or large a pixel is, there is still circuitry within each pixel that takes up space and limits the area within that pixel (or photosite) dedicated to capturing light. As the pixel gets smaller, this circuitry takes up a disproportionately larger percentage of space within that pixel. Thus, light capturing efficiency decreases. But all these nuts think about is the total "quantity of light hitting the sensor". Yeah, total light hitting the sensor remains the same, but pixel efficiency goes down because each pixel isn't simply an empty well whose entire area is dedicated to capturing light! Leave it to these armchair theorists to tell Canon that they're doing it all wrong by using larger pixels. LOL.

Comment edited 2 times, last edit 10 minutes after posting
0 upvotes
Silvarum
By Silvarum (7 months ago)

@Eike Welk: check data at sensorgen.info and see that read noise is pretty much the same for D4 and D800. Also look for "Equivalence" article from Joseph James.
@T3: First of all, I mentioned that sensor efficiency should remain the same. Second of all, circuitry size simply does not make any difference for big sensors. Unless you go to the P&S pixel size that is. As I said below, it is the reason why Sony claims that BSI technology will not give much advantage for big sensors. And you always can use CCD sensors to go even smaller. "LOL"

0 upvotes
JadedGamer
By JadedGamer (7 months ago)

@Silvarum, the rest of us know that the information read by the sensor is the light, and if there is less light "data" per pixel to read then there is less accuracy in the reading. Adding complex algorithms to try and compensate is just fixing the wrong problem.

Look for "signal to noise ratio" articles since the beginning of electronics.

0 upvotes
Silvarum
By Silvarum (7 months ago)

@JadedGamer: Bigger pixels have bigger noise! There is less "light data" per pixel, but also less noise. Therefore, SNR remains the same.

0 upvotes
Renzokuken
By Renzokuken (7 months ago)

I wonder how much of this technological improvement will translate to better stills??

Canon seems to have their hands full on video and their cine-line recently...

0 upvotes
yabokkie
By yabokkie (7 months ago)

I don't know anything can be used for stills directly but video is everything in the long run, because our eyes always see video and never still.

0 upvotes
RPJG
By RPJG (7 months ago)

yabokkie, you make a lot of "interesting" comments on this forum, but that takes the cake!

0 upvotes
BJL
By BJL (7 months ago)

None: the only technological change here is having fewer, bigger photo sites; about 2MP, which is good enough for HD video, but too few for almost any still photography with a se SLR that big.

0 upvotes
BJL
By BJL (7 months ago)

Typo: with a _sensor_ that big (36x24mm)

0 upvotes
yabokkie
By yabokkie (7 months ago)

fewer, larger pixels need special efforts to make,
which are more difficult than smaller ones.

also they improved readout to have better SNR and DR,
which may or may not be used for consumer (if too costly).

Comment edited 2 times, last edit 9 minutes after posting
0 upvotes
Dougbm_2
By Dougbm_2 (7 months ago)

Hmm…so bigger pixels are better? Does that mean the Mp war is over and my 5D mk1 is back in vogue?

1 upvote
JadedGamer
By JadedGamer (7 months ago)

The megapixel wars ended when Canon made a 120 megapixel proof-of-concept sensor. Not that they will ever use that thing, though.

1 upvote
Dan Tong
By Dan Tong (7 months ago)

Impressive. The Canon people are doing some serious innovation.

7 upvotes
Marksphoto
By Marksphoto (7 months ago)

just confirms that all we need is a 2-3mp sensor on the full frame Canon 5d MK 4 for the general purpose photography in low light and still be able to get great enlargements under daylight conditions.

Finally the consumers need to understand that there is no need for 21mp at an expense of noise.

Comment edited 1 minute after posting
2 upvotes
Juck
By Juck (7 months ago)

You don't know much about sensors, cameras,photography,, eh? Seriously,, lol.

13 upvotes
groucher
By groucher (7 months ago)

It depends what you want to do but you're more or less right. If you only ever want to view photos on an HD television then 2mp is fine although you can't zoom in. 4mp is fine for printing to A4 and 8mp for A3. Very high res cameras really come into their own for large format printing and for telephoto work - e.g. a 300mm f4 on the D800 becomes 600mm equiv f4 if the image is cropped to 9mp.

3 upvotes
Silvarum
By Silvarum (7 months ago)

"There is no need for 21mp at an expense of noise" - I agree. But there is no such expense :P

2 upvotes
RP McMurphy
By RP McMurphy (7 months ago)

2-3mp will give you fine pictures but not for enlarging
Sensor size determines low light gatjhering ability whether you divide that area by many or few MP, after that it depends on how large you want to view it

I dont think you understand the relationship, high MP for low light is fine, low MP for low light is fine, it depends how large you want to view your image

The trade off at larger viewing sizes is noise versus softness

0 upvotes
yabokkie
By yabokkie (7 months ago)

there are about 150M pixels in each of our eyes. maybe you only need 3M but there is no operation to remove the rest.

1 upvote
D1N0
By D1N0 (7 months ago)

What we need is a camera with interchangable sensor.

Comment edited 22 seconds after posting
0 upvotes
JadedGamer
By JadedGamer (7 months ago)

Ricoh made one (interchangeable sensor and lens modules). Some are fans of it, but not everyone...

0 upvotes
yabokkie
By yabokkie (7 months ago)

> Some are fans of it, but not everyone...

Sony is one of them.

0 upvotes
nofumble
By nofumble (7 months ago)

Nobody need to buy fast lens, that is what Canon is afraid.

0 upvotes
Plastek
By Plastek (7 months ago)

You don't understand what fast lens does, do you?

2 upvotes
Juck
By Juck (7 months ago)

Yeh,, canon is terrified. lol, clown.

5 upvotes
nofumble
By nofumble (7 months ago)

No I don't understand why it takes too long for the folks at Lytro to perfect their artificial bokeh algorithm.

0 upvotes
Plastek
By Plastek (7 months ago)

nofumble - that's simple: Money. They don't have much of that.

0 upvotes
Edmond Leung
By Edmond Leung (7 months ago)

Cool!
But no surprise! Because long time ago, Canon had already developed the 30cm extremely high sensitivity sensor.
This is just a baby of that sensor.... But the technology of this baby sensor is still out of reach by other camera manufacturers.

0 upvotes
limlh
By limlh (7 months ago)

Canon admits that the sensor may be useful for medical and security purposes. Unlikely to see it in digital camera. I think for sensitivity in digital imaging to take a big step forward, a new base material for sensor like Graphene would be needed. In fact, it may be more practical for sensor makers to use Graphene in their current process than making organic sensor, which Fujifilm is very likely to use in the next iteration of the X-series cameras.

Comment edited 2 times, last edit 4 minutes after posting
3 upvotes
gillamoto
By gillamoto (7 months ago)

graphene imaging sensor surely is a great idea. but I doubt we will see it as commercial product in a near future. and maybe the cost will be much higher. my wish for now is the quantum dot film.

1 upvote
limlh
By limlh (7 months ago)

Graphene is just the base material. The existing manufacturing process can be used. In fact, Graphene can be as much as 5 times cheaper when mass produced. We would most likely see organic sensor first before Graphene.

Comment edited 12 minutes after posting
0 upvotes
Juck
By Juck (7 months ago)

limlh ,, just stop,, you clearly have no clue what you're babbling about.

3 upvotes
Plastek
By Plastek (7 months ago)

Cool, only noone right now has a production capacity for that. And building factory for semiconductors takes roughly 2 years till it starts. Not to mention that you need to develop all of the equipment and design the sensors themselves, what might be very complicated when there are no working units anywhere on a commercial landscape.
Than there are issues specific to graphene sensors, but for that you'd have to google, as I don't have enough knowledge, even though my Uni been working on graphene.
I would say that an optimistic scenario is 15 years till any consumer cameras with graphene sensors will hit the market. If they will at all cause it might be some other technology will come out that will be better for imaging sensors.

0 upvotes
Anadrol
By Anadrol (7 months ago)

Intel just announced that graphene processors will be made around 2020, that doesn't mean big processors though !

1 upvote
limlh
By limlh (7 months ago)

So, Intel is the first one to make Graphene sensor. A smart way to diversify.

0 upvotes
Plastek
By Plastek (7 months ago)

CPUs are not image sensors. Each of them got very different challenges in the design. You just can't compare them.

0 upvotes
Bogdan Spineanu
By Bogdan Spineanu (7 months ago)

Wow .. if you make sensor pixels larger you can capture more light ! Why didn't I think of that ? Revolutionary !

2 upvotes
Plastek
By Plastek (7 months ago)

If you would read the dPreview forums you would know that in did - it is an revolutionary information.

2 upvotes
Zerg2905
By Zerg2905 (7 months ago)

@Mr. Spineanu: you did not because you have nothing to do with sensor technology R&D. Obviously. So criticism without solutions is not quite constructive, you know... Cheers! :)

1 upvote
Silvarum
By Silvarum (7 months ago)

You can capture more light only if you make your sensor larger.

2 upvotes
RP McMurphy
By RP McMurphy (7 months ago)

You didn't think of that because you dont understand it
See Silvarum's response

2 upvotes
Bogdan Spineanu
By Bogdan Spineanu (7 months ago)

@Silvarum, bigger sensor is a consequence of larger pixels, but the sensor size is SET at 24x36mm. This is the given situation. Not at all subject of discussion. The sensor size it is that, and that's all there is about it.
So you increase the receptor size. So what did Canon do ? Made a sensor with 1920 pixels on the long side, and if you do a bit of math, you are going to come with 18.75um available per pixel.
And this is how you capture more light. Of course, you could put 1920px on a full frame camera back, but this is another discussion.
@all I still don't see what is so amazing to make a sensor with big pixels. Bigger pixels, yes, you are gonna see fireflies ... What took them so long ?

0 upvotes
utomo99
By utomo99 (7 months ago)

I hope canon use the technology to improve the cameras sensor. so it can improve the low light photos.
By using this new technology and also fast lens canon maybe No need to use ISO higher than 200, and the low light photo still look good.
Let wait for the new cameras

0 upvotes
Plastek
By Plastek (7 months ago)

Uh? Another person that doesn't know what ISO is?

4 upvotes
Eric Hensel
By Eric Hensel (7 months ago)

How do you manage to balance that giant brain on that skinny neck, Plastek?

0 upvotes
falconeyes
By falconeyes (7 months ago)

0.01 lux is -8 EV.

I heavily doubt it was this dark, as it would correspond be about a moonless starry night. But the fireflies then would illuminate the scenery (or their direct surrounding at least) as main light source and I didn't notice the effect in the video.

Anyway, while dark, -8 EV corresponds to (F/1.4, 1/24s) ISO 1,200,000. I saw extreme noise at 640px web resolution. Scaling web resolution to full 24 MP resolution, the would be the screen pixel noise (24MP) at ISO 14,000. I'd say a normal full frame 24MP dSLR doesn't have this much pixel noise at ISO 14,000.

Therefore, I conclude that the demo shown by Canon is technically lame, more of a marketing gig than anything else.

2 upvotes
Donnie G
By Donnie G (7 months ago)

This sensor will be great used in a security system camera where IQ need only be good enough to get a recognizable face recorded in spaces that are poorly lit or not lit at all. Canon intends to be a major player in the security industry and this video is merely a proof of concept offering. They state very clearly that the sensor was designed strictly for video usage, so comparing its IQ to that of a purpose built stills camera is pointless.

2 upvotes
Frank_BR
By Frank_BR (7 months ago)

>Therefore, I conclude that the demo shown by Canon is technically lame, more of a marketing gig than anything else.

Yes, I think the same way. Possibly Canon used a lens with very large aperture, like F/0.7, to shoot the video. Who knows?
Moreover, notice that the scenes are so dark as if they were underexposed by 2 or 3 stops. Finally, when Canon claims the illumination was 0.01 lux, certainly it refers to the AVERAGE light level. The fireflies and stars were at much higher light levels because they produce their own light!

Comment edited 3 minutes after posting
0 upvotes
jhinkey
By jhinkey (7 months ago)

Well, let's see:
- Yes, from the beginning frames of the video where they show the moonless starry night sky, then by your math it IS 0.01 lux.
- The fireflies are not illuminating the scene as they are not on continuously as is obvious from the video. They are small intermittent point sources of light and don't contribute the the overall illumination.

So I highly suspect that Canon was not telling a lie when it described the situation and they are not technically LAME.

Could they do better by combining pixels of a higher resolution sensor - perhaps, but it would take a lot more computing power and maybe that's the point.

2 upvotes
Rad Encarnacion
By Rad Encarnacion (7 months ago)

@falconeyes
Perhaps you were distracted by the marketing text in the center of the frame, but the scenes at 0:09-0:15 was of the night sky, precisely to show that this video was taken during a "moonless, starry night."

1 upvote
yabokkie
By yabokkie (7 months ago)

a moonless clear night sky provides about 0.001 to 0.002 lux (Wiki says 0.002 but I'm not sure about that).

if fireflies are the major source of light, it could change wildly between 0.001 and say 0.03 lux (if 0.01 was average).

Comment edited 4 minutes after posting
0 upvotes
falconeyes
By falconeyes (7 months ago)

I was only wondering if it really was pitch black dark as they claimed. Because fire flies then actually illuminate a scene, I have seen it with my own eyes. But in my actual argument, this plays no role, I assumed 0.01lx to be a true assertion.

It is lame because any full frame dSLR can do this already, in still mode at the same shutter speed. The news is that Canon works around a bug how they produce video. And this IS lame.

0 upvotes
zos xavius
By zos xavius (7 months ago)

Its hard to say from one video what this is truly capable of I think. Capturing the stars impressed me personally, but I've never had a D4 or something of that caliber to try insane high isos on, so it may be possible to do this with normal full frame already. (disclaimer: I am by no means a canon fan)

Comment edited 38 seconds after posting
0 upvotes
lylejk
By lylejk (7 months ago)

Definitely cool stuff. I guess the large pixels have it when it comes to both low noise and light gain. :)

1 upvote
Leandros S
By Leandros S (7 months ago)

** Double posting, pls delete **

Comment edited 14 minutes after posting
0 upvotes
audijam
By audijam (7 months ago)

perfect for wedding.

1 upvote
Roland Karlsson
By Roland Karlsson (7 months ago)

Few and large pixels is a very good choice for ultra high ISO work, like this one. But, when there is more light, more pixels gives better IQ. So, if you want few and large pixels or more and smaller pixels is a choice. Do you want to maximise high ISO performance or do you want to maximize low ISO performance.

1 upvote
falconeyes
By falconeyes (7 months ago)

As I said below, the fact that the pixels are large has nothing to do in this. It is that Canon doesn't throw away any pixels for video in this case as they do in their consumer dSLRs.

Look up sensorgen.info to see that larger pixels gain no extra sensitivity for SLRs.

3 upvotes
Dave Luttmann
By Dave Luttmann (7 months ago)

Agreed Falcon....I'm surprised how many people still believe in the old myths.

2 upvotes
SRT3lkt
By SRT3lkt (7 months ago)

so, is it impossible to take movie of fireflies flying without this sensor?

0 upvotes
Total comments: 228
12