Previous news story    Next news story

Canon develops high sensitivity full-frame CMOS sensor for video

By dpreview staff on Mar 4, 2013 at 19:30 GMT

Canon has developed a 35mm full-frame CMOS image sensor designed for low-light video capture. The 16:9 sensor features a 1920x1080 pixel array, meaning each pixel measures a huge 19 microns along each edge - 7.5 times larger than the ones in the EOS-1D X. The large pixels and low readout-noise circuitry allow the sensor to capture light around 10 times less bright than current CCDs used for astronomy. The sensor will first be shown in public at a security show in Japan. Other applications the company is looking into are astronomy and medical research.

Interestingly, the 36 x 20mm sensor's use of 19μm pixels means it creates its video footage from a 1920 x 1080 array. This means it has only one photosite per output pixel, so won't offer the same color resolution as Canon's EOS C300 professional video camera, which has four capture pixels (a Red, Green, Blue, Green quartet) per output pixel.


Press Release:

Canon develops 35 mm full-frame CMOS sensor for video capture

Photos taken before and after increasing sensitivity using the CMOS sensor on a full moon night. Click through to view Canon's video demonstrating more sample images.

TOKYO, March 4, 2013—Canon Inc. announced today that the company has successfully developed a high-sensitivity 35 mm full-frame CMOS sensor exclusively for video recording. Delivering high-sensitivity, low-noise imaging performance, the new Canon 35 mm CMOS sensor*1 enables the capture of Full HD video even in exceptionally low-light environments.

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. In addition, the sensor's pixels and readout circuitry employ new technologies that reduce noise, which tends to increase as pixel size increases. Thanks to these technologies, the sensor facilitates the shooting of clearly visible video images even in dimly lit environments with as little as 0.03 lux of illumination, or approximately the brightness of a crescent moon—a level of brightness in which it is difficult for the naked eye to perceive objects. When recording video of astral bodies, while an electron-multiplying CCD,*2 which realizes approximately the same level of perception as the naked eye, can capture magnitude-6 stars, Canon's newly developed CMOS sensor is capable of recording faint stars with a magnitude of 8.5 and above.*3

Using a prototype camera employing the newly developed sensor, Canon successfully captured a wide range of test video,*4such as footage recorded in a room illuminated only by the light from burning incense sticks (approximately 0.05–0.01 lux) and video of the Geminid meteor shower. The company is looking to such future applications for the new sensor as astronomical and natural observation, support for medical research, and use in surveillance and security equipment. Through the further development of innovative CMOS sensors, Canon aims to expand the world of new imaging expression.

Canon Marketing Japan Inc. will be exhibiting a prototype camera that incorporates the newly developed 35 mm full-frame CMOS sensor and sample footage captured with the camera at SECURITY SHOW 2013 (www.shopbiz.jp/en/ss/), which will be held from Tuesday, March 5, to Friday, March 8, at the Tokyo International Exhibition Center in Tokyo, Japan.

*1 An imaging element (aspect ratio: 16:9) that supports the largest image circle size possible when shooting with a Canon EF lens.
*2 A CCD sensor with a readout mechanism that multiplies electrons after being converted from light. Applications include nighttime surveillance and the capture of astral bodies and nighttime nature scenes.
*3 The brightness of a star decreases 2.5-times with each 1 magnitude increase.
*4 Recording of test video footage was made possible through cooperation from ZERO Corporation.

Comments

Total comments: 103
12
RichRMA
By RichRMA (7 months ago)

For low light video, check-out the Malincam.

0 upvotes
Sad Joe
By Sad Joe (7 months ago)

Cool stuff !

0 upvotes
utomo99
By utomo99 (10 months ago)

Is there any chance that some improvements bring to new Canon S series which will replace the Canon s 110 ?

0 upvotes
Sad Joe
By Sad Joe (Mar 8, 2013)

Well, now we know what Canon's Sensor Engineers have been working on rather than improving on the (now very old) 18 MP sensor in most of their cameras. Still no reason however what Canon's Autofocus Engineers have been up to and why their happy to release at least one new product (EOS-M) with duff focusing and why their also happy to have a brand new full frame DSLR (6D) with focusing only as good a 20D from 8 years ago....

0 upvotes
yabokkie
By yabokkie (Mar 8, 2013)

pixel size means little if any. the readout circuitry is one of Canon's major weak points and should have been improved years ago.

0 upvotes
MaleManGuy
By MaleManGuy (Mar 6, 2013)

Wow. Good for security cameras..... Get away from the ir LEDs and such. 2MP is more than enough.

0 upvotes
spatz
By spatz (Mar 6, 2013)

Too bad, it looks like it's not a BW sensor. The presence of an RGB matrix makes this much less attractive for astrophotography with narrowband filters.

1 upvote
maxnimo
By maxnimo (Mar 6, 2013)

Aaaaaah.... videos by natural candlelight. Very super cool!

0 upvotes
MarkByland
By MarkByland (Mar 5, 2013)

Canon has stiff competition with Finger Lakes Instruments, SBIG, and any other Kodak sensor patent holding camera manufacturer in the world of AP. I don't understand their mindset any more. Leave the CCD market already dominated by QSI, FLI, SBIG, and many others to the people that have spent a lifetime developing those cameras.

It would seal the deal for me if Canon attempted to walk in to a market and crush the competition in order to establish yet another half-hearted attempt at market share. Pay attention to your current customer requests and move advancements further down the line up in DSLRs.

0 upvotes
Mandibela
By Mandibela (Mar 5, 2013)

But you still get that CMOS jelly goodness?

1 upvote
Sdaniella
By Sdaniella (Mar 5, 2013)

when i think out loud... others ignore me, met with deaf ears and closed minds, as though my ideas are 'unappealing'...

but Canon listens

http://forums.dpreview.com/forums/post/50980532

sdyue

0 upvotes
Royce - Into the Night Photography

I believe we are going to see this low-light technology spill over quickly to Canon's high-end, full-frame sensor DSLR cameras. In their demo video, Canon ends with this statement: "Canon aims to expand the world of new imaging expression through the further development of innovative CMOS sensors."

You can see photos and a link to their amazing demo video at my blog:
http://intothenightphoto.blogspot.com/2013/03/new-see-in-dark-camera.html

0 upvotes
Lucas_
By Lucas_ (Mar 5, 2013)

This is good stuff. I wonder how Sony will respond in order to keep being the leader in professional video. Isn't competition awesome?

1 upvote
iae aa eia
By iae aa eia (Mar 5, 2013)

the most interesting thing about this sensor is that it is not about technology, as usual, but using the head only. "ok, let's just develop a sensor for an ff video camera. if they want a picture, just extract a frame out!"

it is not going to be an expensive sensor to produce, so i guess the camcorder won't be expensive, unless it's a heck of a loaded one.

i still don't understand why they don't make an affordable 10 mp ff camera. it would have enough res for photos and hd video, and cost about a grand.

don't they understand there a lot of people who can't wait to see ff sensor cameras something normal again?

0 upvotes
silentstorm
By silentstorm (Mar 5, 2013)

I'm not sure how they derived the 7.5x larger pixel size than 1Dx. My estimated guess should be just about 4x.

Let's rewind to the 1st 1Ds with 12MP, I remembered it to have 9.x microns

Fast forward to 1Dx with 18MP, & improve fabrication technology, I would guess it to be somewhat around 5 microns.

So how is 19 vs 5 microns be 7.5x???

Lastly, full HD only offers 2MP. This sensor is really not meant for your common everyday camera.

Where's the challenge? By reducing pixels & enlarging them, & then claim that to be a breakthrough in sensitivity? I don't get it at all???

0 upvotes
dimsgr
By dimsgr (Mar 5, 2013)

1d X has a pixel size of 6,95 microns - also sensitivity is analogous to the surface of the pixel, so Canon is right about being 7.5 times more sensible than 1d X

I agree that enlarging pixels is not breaking through in a logical way, especially considering this sensor is really a 960x540 one

0 upvotes
Brian Lund
By Brian Lund (Mar 6, 2013)

It is as much 960x540 as the sensor in the 1Dx is 2592x1728...

0 upvotes
trungthu
By trungthu (Mar 5, 2013)

"... This means it has only one photosite per output pixel, so won't offer the same color resolution as Canon's EOS C300 professional video camera, which has four capture pixels (a Red, Green, Blue, Green quartet) per output pixel... "

I don't know how it can record color information in only one photosite. While almost all others use three color filter in four pixels (RGBG).
Please help me. Thanks.

1 upvote
Brian Lund
By Brian Lund (Mar 6, 2013)

Google "Bayer filter" - it is used in pretty much every still image camera today! More likely than not, yours too! (Except for Sigma with their Foveon sensor) No big news, this has been around for many years!

0 upvotes
trungthu
By trungthu (Mar 6, 2013)

Yes, I see.
Bayer filter use four capture pixels to record color informations.
""" .. This means it has only one photosite per output pixel, so won't offer the same color resolution as Canon's EOS C300...""" what does i mean?
Only one photosite per output pixel""" ???
Thanks Brian Lund for your respond.

0 upvotes
Brian Lund
By Brian Lund (Mar 7, 2013)

I'm not an expert on the C300 but I'll try to explain how I understand it..!

In all "normal" cameras a bayer filter is used, and one "photosite" is one pixel in the finished image, colour interpolation is used to create a full colour image, ie. a red photosite/pixel will get it's green and blue colour from the neighbouring photosites/pixels!

The C300 sensor has four times as many pixels as the output image! It also uses the bayer filter but it uses four photosites for one pixels to get more accurate colour information in the pixel itself - ie. no colour interpolation!

0 upvotes
malcolm82
By malcolm82 (Mar 7, 2013)

For this reason 7680x5120 sensors should become the standard to provide near optimal image quality for the new 3840x2160 ultra hd tv standard in the same way as the 3840x2160 sensor of the c300 provides near optimal full hd quality.

0 upvotes
jaygeephoto
By jaygeephoto (Mar 5, 2013)

Demo video is intriguing but dull. Whenever I think of great low light photography I cannot forget the candle lit scenes from Stanley Kubrick's Barry Lyndon, using a NASA spec. 50mm 0.7 Zeiss lens. - using Kodak 5254 color negative film stock, ISO 100!

Now that's some stunning low light photography - still gives me goosebumps.

http://www.youtube.com/watch?v=hJahPDkvivc

Maybe we'll see this sensor in RED cameras: http://www.red.com/

Comment edited 8 minutes after posting
2 upvotes
sportyaccordy
By sportyaccordy (Mar 5, 2013)

Why are people so obsessed with binning? Using actual larger pixels wins out both logistically and optically

Plus lets be honest. Whats more of a priority for 99% of photographers... pixel peeping a 30 by 40 foot print, or speed? I would be OK with a 4MP sensor if it legitimately went up to something like ISO204800. When in doubt, K.I.S.S.

0 upvotes
alexisgreat
By alexisgreat (Mar 5, 2013)

You can already detect planets orbiting other stars with current (even relatively cheap astrocams) so this can only be better!

0 upvotes
spatz
By spatz (Mar 6, 2013)

Direct imaging? Can you please post a link to that?

0 upvotes
Brian Lund
By Brian Lund (Mar 7, 2013)

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

Although I think they have all been imaged in the infra-red spectrum and not the visible light!

0 upvotes
alexisgreat
By alexisgreat (Mar 5, 2013)

I use an astrocamera and this is welcome news! Hopefully they will have a monochrome version with no Bayer filter because those are the most sensitive :-) we imagine with LRGB filters which really cut through light pollution and combine the images to produce colors :-)

0 upvotes
Sdaniella
By Sdaniella (Mar 5, 2013)

glad to see this sensor

sdyue

1 upvote
Sdaniella
By Sdaniella (Mar 5, 2013)

http://www.canon.com/news/2013/mar04e.html

0 upvotes
Pat Cullinan Jr
By Pat Cullinan Jr (Mar 5, 2013)

Tondelayo like.

0 upvotes
SeeRoy
By SeeRoy (Mar 5, 2013)

There's surely got to be a huge and lucrative market for surveillance use. Every time I see surveillance footage in connection, for example, with crime investigation, it's amazing how poor the quality of the images are - particularly in relatively low light. A lot of the hardware installed is probably cheap, nasty and out of date, however it also applies to a lot of corporate and municipal installations.

0 upvotes
T3
By T3 (Mar 5, 2013)

I remember there was a large group of people on these forums who were adamant that large high resolution sensors with *smaller* pixels (as opposed to larger pixels) would yield better low-light image quality. Wonder where those people are now.

2 upvotes
Anastigmat
By Anastigmat (Mar 5, 2013)

They were probably talking about binning, or combining signals from several pixels into one. That is of course not as good as having a single large pixel the same size as those small ones combined. For pure video applications, what Canon has done with this sensor makes more sense. The images will be stunning, because if a lens only needs to resolve in low resolution, contrast goes up.

0 upvotes
mgrum
By mgrum (Mar 5, 2013)

This sensor has more than just very large pixels, some of the other technologies may in fact require pixel pitch to be increased.

There some pretty good evidence that large numbers of small pixels can yield a better downsampled image when a sophisticated noise reduction algorithm in used. This increase processing workload somewhat, which is not desirable in a video camera.

4 upvotes
Virvatulet
By Virvatulet (Mar 5, 2013)

I for a one would always prefer signal purity, integrity and naturality over artificial edge definition, which is achievable with advanced NR processing.

Micro-optoelectronics physics has not changed, with a given technology and generation, larger pixels can collect also relatively more photons. But I do admit this is a complex question in real circumstances.

1 upvote
mariuss
By mariuss (Mar 5, 2013)

There is for sure no coicidency that this news is comming right after Aptina/Sony news.
Its good to know that Canon is continue researching on sensor tehnology but the question is, will Canon use this tehnology in any other DSLR-s?
And how about DR at low ISO?
From my understand, this shows that Canon concentrate its research on High ISO tehnology and not in high DR at Low ISO.

0 upvotes
rrccad
By rrccad (Mar 5, 2013)

actually it is a coincidence, it was announced at an appropriate trade show.

not everything canon does revolves around DSLR sensors or even DSLR imaging.

"The company is looking to such future applications for the new sensor as astronomical and natural observation, support for medical research, and use in surveillance and security equipment."

and lo, announced at a security show.

0 upvotes
rrccad
By rrccad (Mar 5, 2013)

this is intriguing .. doing some napkin maths .. .01 to .03 lux is somewhere around -6 to -8 EV .. correlating that a bit more .. that's around ISO 500K to 800K at a reasonable shutter speed of 1/60th and aperture (F1.4).

from an astrophotography aspect being able to take a very short video clip, stack and drizzle it up to 2x2 the size .. which such sensitivity would be amazing. usually the video astrophotography is limited to brighter objects ie: planets.

0 upvotes
Camediadude
By Camediadude (Mar 5, 2013)

I am mostly a visual observer, but I am most intrigued by the amateur astrophotography applications of these new super sensitive breeds of sensors! Better cosmic captures by dedicated amateur space photo bugs are to the benefit of us all ... new discoveries are waiting to be made, and sensors like these might help! Exciting times to be in with all this tech, no? :)

1 upvote
iamachair
By iamachair (Mar 5, 2013)

I wonder how difficult it will be to actually under-expose a picture if you WANT it to be under-exposed. I guess just adjusting the exposure compensation or purposely choosing settings that will under-expose it.

0 upvotes
Jahled
By Jahled (Mar 5, 2013)

Hahahaha This is serious stuff

0 upvotes
Woodlink
By Woodlink (Mar 5, 2013)

Regardless if this tech ever makes it's way into consumer DSLR's, you have to admit the results presented in the sample video are quite dramatic.

0 upvotes
ImagingPro
By ImagingPro (Mar 5, 2013)

Great high-sensitivity sensor for full HD stills and video.
Full intensity resolution, 1/2 chroma resolution, which normally looks OK.
Sure, an 8MP sensor would give full 1080P intensity and chroma resolution, but binning an 8MP sensor does not give the same sensitivity as a 2MP sensor, it gives 1/2 the sensitivity. That might be a better compromise for some, but not for "seeing in the dark" like this imager. Cool.

1 upvote
martian1
By martian1 (Mar 5, 2013)

Colour resolution at full 1920 x 1080 can be provided by a 3 CMOS system using a prism assembly to split the 3 colours red, green and blue onto 3 CMOS sensors. This is today employed by many professional video cameras, albeit using relatively small 1/3", 1/2" or 2/3" sensors.

Note that splitting the incoming light using a prism assembly utilises ALL incoming light (except for minor reflection losses), while any colour filter in front of a pixel loses a major portion the incoming light by filtering out other colours, e.g. for red one has to filter out green and blue, for green one has to filter out red and blue, etc - in principle this loses 2/3 of the incoming light, however note that in a standard Bayer pattern the filters are setup a bit different.

Additionally, at each pixel full colour information is available through the 3 sensors, while for a single sensor using Bayer pattern filters, colours are always interpolated, as each pixel has either a filter for red, green or blue.

0 upvotes
AllMankind
By AllMankind (Mar 5, 2013)

A 3CMOS array would drive the cost up considerably (3 FF sensors) and I would have to wonder if Canon FF lenses would even work, as I am not sure what that (3CMOS) would do to the registration distance.

Of course, if low light capability is your primary goal, then color fidelity is of secondary importance.

0 upvotes
chn_andy
By chn_andy (Mar 5, 2013)

This comparison with EMCCD, in my opinion, is biased. A large pixel can always beat a small pixel sensor given a equivalent QE and read noise. All single photon detection these days are using EMCCD or sCMOS. Canon should publish its QE curve, read noise and dark current data.

0 upvotes
Lucas_
By Lucas_ (Mar 5, 2013)

Interesting, but I really don't know anything about video, so I wonder if this new Canon psensor has any similarity with the sensor on the new Sony Pro video cameras F55 and F5, just made available at B&H:
http://www.bhphotovideo.com/c/product/898428-REG/Sony_PMW_F55_CineAlta_4K_Digital.html/
The specs call for :
"8.9MP Super 35mm CMOS Image Sensor and Internal 4K/2K/HD Recording" .
I apologise if that's something else, as I said I'm not at all into video...!

0 upvotes
AllMankind
By AllMankind (Mar 5, 2013)

A Super35 sensor is closer in size to APS. A tad smaller actually ~25x14mm. (APS ~24x16mm).

1 upvote
plasnu
By plasnu (Mar 4, 2013)

Looks like this originally is a munition technology, which Canon makes huge money these days.

0 upvotes
HubertChen
By HubertChen (Mar 4, 2013)

It is always interesting to see concepts and technologies pushed into extremes :-). This could be a candidate to expand dynamic range, either to add a smaller photo sites in parallel or to reset the photo site once full and count the resets. Night time pictures with one point light source have an extreme dynamic range. Thanks for this post!

0 upvotes
iAPX
By iAPX (Mar 4, 2013)

So DPReview is expanding into surveillance camera?

Because I don't see the point to shoot stars with 1080p sensor, it's irrelevant even if doable. Or doing a "movie" with still stars on sky. And moving stars short movies has been done with DSLR, using long exposure times, with great success (and not limited to 1080p).

Next subject will be about VGA (640x480) identification camera sensor on laptop?
It's camera sensors, but it's not relevant here!

0 upvotes
dpLarry
By dpLarry (Mar 4, 2013)

Just because you're not interested doesn't mean others won't be.

I for one find an announcement of a Canon ultra low light sensor to be interesting.

11 upvotes
ultimitsu
By ultimitsu (Mar 5, 2013)

the problem is precisely "you dont see the point?"

if this camera is as good as they say, it will be able to take very good quality images at high shutter speeds where light is not abundant. wide life, sports, deep water photography can all benefit from that.

0 upvotes
marike6
By marike6 (Mar 5, 2013)

No offense, but how is the existence of this article detracting at all from your DPR reading experience? It's not like this HTML document about new Canon sensor technology is taking the place of another HTML document about the latest m43 camera released or the newest P&S. If DPR is offering to host all these documents, they can publish all three articles, and you can click on the one that interests you. That's how websites work, they are huge webs of documents that can grow, and offer something for everybody.

I enjoy reading about the latest developments in digital video and photography so I'd likely click on all three articles.

0 upvotes
hammerheadfistpunch
By hammerheadfistpunch (Mar 4, 2013)

That demo movie is pretty questionable. So they want to compare a super sensitive CMOS created specifically for high sensitivity to what? a dated 4x3 3ccd chip setup? And what a surprise, the ccd looks like crap, um, duh. Although I have no doubts this will be an amazing capture tool, and I genuinely mean that I am WAY excited to see this against something like a 1dx, the video demo is terrible You know what takes better night time video than my old Canon GL2? ALMOST ANY MODERN CMOS CAMERA.

0 upvotes
dosdan
By dosdan (Mar 4, 2013)

Pity the DR value is not stated. This would show how low the total read noise floor is in relation to the FWC.

Also I wonder what trade-offs have been made to achieve this, besides the low resolution?

Dan.

Comment edited 2 times, last edit 3 minutes after posting
0 upvotes
RedFox88
By RedFox88 (Mar 4, 2013)

If this is indeed an image sensor with just resolution for HD video, I think the target buyer would be... movie companies not military or security. I read or saw a thing on the Navy Seal movie which was shot with a 5D2 and how being able to use ISO 3200 saved so much time and money to eliminate much special lighting to properly illuminate dark scene for capture with convential movie making gear.

0 upvotes
Macx
By Macx (Mar 4, 2013)

This would probably make a good security camera. :)

This is a very specialised piece of equipment. There are probably a lot of broadcasters and TV and YouTube production crews (essentially reality-based TV and video) who will have use for a low-light camera of high quality, but it's limited resolution is going to limit it's longevity and it's usefulness.

0 upvotes
Canadianguy
By Canadianguy (Mar 4, 2013)

"Canon Marketing Japan Inc. will be exhibiting a prototype camera that incorporates the newly developed 35 mm full-frame CMOS sensor and sample footage captured with the camera at SECURITY SHOW 2013 (www.shopbiz.jp/en/ss/)"

Its not targeted at the consumer - most likely Military / Security applications...

0 upvotes
ZAnton
By ZAnton (Mar 4, 2013)

Good circuitry is OK, but why can't they make a proper pixel binning for HD video on an existing sensor?
Instead of roughly 6000px*4000px, make a pixel binning 3*3 clusters to appr. 2000*1300px.

Comment edited 23 seconds after posting
0 upvotes
Richard Murdey
By Richard Murdey (Mar 4, 2013)

As I understand it, there are two problems: the time required to read all 24,000,000 pixels off the sensor, and the computing effort required to bin those down to 2,666,667 pixels. At 60fps, the system would have to be capable of handling 1.4 billion pixels of data per second.

2 upvotes
ZAnton
By ZAnton (Mar 5, 2013)

OK, not 24MP, then 8 MP, they already have this sensor for a canon C-video camera. If 1Dx is capable of 12fps on 18MP, 24 fps on 8 MP should not be a problem (except of the power required, hehe). Well anyway, as an engineer I am very curios about HD-FF, the ISO performance must be astonishing, but if I were marketologist, I would have some very serious doubts about such product. Using it as a HD-camera for normal filming make no sense - night lighting is very very bad, I mean not intensity, but aestetic attractivity. So only astrophotographers left. How much of them do we have? How many of them would pay astronomical prices for that camera?

0 upvotes
Gully Foyle
By Gully Foyle (Mar 4, 2013)

Astronomical video captures expected to be phenomenal!
At long last, Canon produces something cool and meaningful. As others already said, we can only hope (wait?) for the same tech on a DSLR (FF or not).

A new megapixel race to begin soon? The other way around this time!

Comment edited 8 minutes after posting
4 upvotes
G G
By G G (Mar 4, 2013)

Looks more like a "my pixel is bigger than yours" type of race.

I don't see any true innovation here, it merely applies a well known recipe: larger pixels get more light.

0 upvotes
Naveed Akhtar
By Naveed Akhtar (Mar 4, 2013)

Why they are hitting the extreme ends .. why only 2MP .. hmm ok for austro
but couldn't they produce something like 8MP (4k native video) .. which can be combined on demand like Fuji is doing for ages .. to produce even lower noise when needed; and possibly a HDR mode too for HD recording. They can deploy the same high quality (lower noise) underlying assembly in that too! or not?

0 upvotes
rrccad
By rrccad (7 months ago)

well considering it's running around 1M ISO .. i'd say they were interested in low light applications more than photography.

0 upvotes
CameraLabTester
By CameraLabTester (Mar 4, 2013)

This new sensor has been developed exclusively for video recording.

Hopefully the next set of posts will reflect opinions based on that.

.

5 upvotes
HowaboutRAW
By HowaboutRAW (Mar 4, 2013)

Okay, so there's something special about this sensor that makes it best for video only.

But by your logic a Canon 5D MIII wouldn't be particularly good for video, and that's simply not the case.

So it would still be nice to see a full frame DSLR with a sensor doing a similar fewer MP count thing.

Comment edited 34 seconds after posting
0 upvotes
pacc
By pacc (Mar 5, 2013)

4k is video too, but full HD is good for family pictures.

0 upvotes
Trollshavethebestcandy
By Trollshavethebestcandy (Mar 4, 2013)

In a few sensor generations this could mean something powerful for regular photography. Imagine your slow mega zoom lenses able to become much brighter via sensor sensitivity. If they do this with a Foveon like sensor with photo site stacking you can have meaningful resolution.

0 upvotes
Photato
By Photato (Mar 4, 2013)

Canon, bring the 70D with a 8MP sensor. Great for low light hand held and of course video.

4 upvotes
HowaboutRAW
By HowaboutRAW (Mar 4, 2013)

Nice idea, but would that 70D be full frame?

4 upvotes
Photato
By Photato (Mar 4, 2013)

Nope, just APS-C, full frame sensors are expensive. But 8MP is a sweet spot for my uses. Would be super nice to shoot compact raw files with it and the video would come super clean which is ideal for compression.

3 upvotes
acidic
By acidic (Mar 4, 2013)

WRT pixel density:
8 MP APS-C = 21 MP 35mm FF

So if an 8MP APS-C 70D comes out, you can expect noise performance comparable to a cropped 5D3 or 6D.

5 upvotes
DaytonR
By DaytonR (Mar 4, 2013)

ROFL ! If the 70D had those specs I would buy one immediately !

3 upvotes
JKP
By JKP (Mar 5, 2013)

You can always buy a used 20D...

0 upvotes
HowaboutRAW
By HowaboutRAW (Mar 4, 2013)

Cool, hope they put something similar (okay say 6MP) in a full frame DSLR; how about the Canon 5D IV.

0 upvotes
thomas2279f
By thomas2279f (Mar 4, 2013)

How about doing a 4k = 8.3mp / 8k = 33 mp version for next generation of video capture devices and for the 8k video you can capture high resolution 33mp still image.

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

1 upvote
HowaboutRAW
By HowaboutRAW (Mar 4, 2013)

Well one thing is that uncompressed capture of even 1080 is still uncommon. So work that out first, bet it's something like a terabyte per hour.

So imagine what uncompressed 4k would be.

I saw that big new Sony monitor/tv doing something like 4K display from a big file and it wasn't that impressive. OLED at lower resolution will likely be much more impressive.

0 upvotes
thomas2279f
By thomas2279f (Mar 5, 2013)

Thanks HowaboutRaw.

Still impressive by Canon and future and we are all winners on this - yes 4k & 8K are very new and processing pipe line + storage power / capacity would need to catch up to handle this.

But in 10 years time this feature by Canon + 8k will be common place in digital capture + viewing.

0 upvotes
Photog74
By Photog74 (Mar 4, 2013)

"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."

Not the best wording on Canon's part - they first quote a 19-micron figure that refers to the pixel pitch (i.e. a linear measurement), and then go on to talk about how the photosites compare to those of the 1D X in terms of surface area.

0 upvotes
Gully Foyle
By Gully Foyle (Mar 4, 2013)

I can't see where they mention 19 microns pixel pitch. They say 19^2 microns surface area.

The same in the dpr intro:
"...meaning each pixel measures a huge 19 microns along each edge."

They only talk of surface area, not pixel pitch. Correct me if I'm wrong.

0 upvotes
joejack951
By joejack951 (Mar 5, 2013)

Gully Foyle, a "micron" is a linear measurement. It's one millionth of a meter, or one thousandth of a millimeter. If you divide 36mm by 1920, you'll get .01875mm, which rounds to .019mm or 19 microns. The surface area of each pixel is then (19uM)^2=.00035 uM^2. A 1DX pixel is 7 microns (36mm /5184) which gives a surface area of .000048 uM^2 which is 7.3X smaller than .00035uM^2 (close enough to Canon's figure). There's nothing confusing or inaccurate about this press release.

1 upvote
Gully Foyle
By Gully Foyle (Mar 5, 2013)

?????
I commented on the OP's argument of bad wording on Canon's part.
They only refer to the 19μΜ long pixels to compute the total surface area (the way you did it); the latter being ~7.5 bigger than a D1X. It sounds quite straight-forward wording to me.

0 upvotes
ThePhilips
By ThePhilips (Mar 4, 2013)

So. After all. They CAN produce a sensor with meaningful number of pixels.

0 upvotes
viking79
By viking79 (Mar 4, 2013)

Not sure why they didn't make it 4 times the pixels and bin them together if they want lower resolution. This would give you 2 greens, 1 red, and 1 blue pixel at each pixel location for added color resolution, but maybe they have their reasons.

Edit: I see the C300 does exactly this, so not sure what the advantage to the larger pixels is here?

Comment edited 6 minutes after posting
1 upvote
Revenant
By Revenant (Mar 4, 2013)

I don't think colour resolution is that important for the extreme low-light applications Canon has in mind.

1 upvote
Total comments: 103
12