Slashdot Mirror
Canon Develops 8 X 8 Inch Digital CMOS Sensor
dh003i writes "Canon has developed a 8 x 8 inch CMOS digital sensor. It will be able to capture an image with 1/100th the light intensity required by a DSLR and will be able to record video at 60 fps in lighting half the intensity of moonlight. There are already many excellent quality lenses designed to cover 8 x 10 inches, although Canon may develop some of their own designed specifically for their requirements."
what is the spectral response?
The article did not explain if this would be incorporated into a camera anytime soon. Also I wonder how it compares to the Hasselblad digital backs and cameras. http://www.hasselbladusa.com/
I assume this means a would-be digital Ansel Adams will need to drag around a camera the size of a bread machine? I'm not too confident the market size is large enough for anything other than highly specialized scientific equipment. I don't see large format digital cameras even for professional photographers because of what it will probably cost to produce.
There is currently no information about the sensor's resolution.
Darn, that was my biggest question. Low light photography has always been one of my interests, so I would have a lot of fun with a camera based on this technology :D ... Actually, I'd be rather keen to have a try making my own... Maybe that's for another day though. ;)
Funnyhacks - Wierd, unusual, and fun hacks
Moonlight on the earth surface or moonlight of the moon?
Taking photos of the moon is same thing as taking photos of the bright sunlight of theearth surface. Like 1/125 f:11 ISO 100.
No but really, that is impressive but depends from the aperture and lens quality do we get better than f:0.4 or something. But that just means the A/D conversion is impressive at that size of sensor so we might see very noiseless ISO of 250 000 setting.
But there really is demand to get a old formats back. Especially if the megapixel amount would be same as with negative. What is not going to happend because Canon likes more to make bigger sensors than tight megapixels. Thing about A4 (197x210mm) sized full size architecture camera. On such negative with ISO 50-100 you can capture more details than what you could even think about with digital cameras.
it's be great if it were something lame like 6 megapixel
.8 mp and it takes amazing pictures because the sensor is huge. Like this thing.
Why is 6 mp lame? Do you know the Hubble is something like
Actually, they use software to merge the photos. Otherwise the photos would suck.
The article says that there's currently no information available about the resolution. I just wondered if this might be the same sensor from the story the other day about the 120MP sensor...? Two stories for the price of one? Agree? Disagree?
Funnyhacks - Wierd, unusual, and fun hacks
even 640 by 480 would be pretty amazing if it was low-noise, 60 FPS, in moonlight.
I can't be arsed to post-process everything like NASA can.
Yes, 6 Mp is a lot. It is three 1920x1080 monitor screens.
Not sure about the Hubble's image processing, but I had some dealings with satellite images several years ago. Our images were created from a combination of a high resolution monochrome image to provide detail then a sequence of lower resolution colour images potentially ranging from UV through to IR. The images were then combined, the colour spectrum compressed and/or shifted to fit within the range visible to the human eye before being output as high-resolution (for the time) colour images.
UNIX? They're not even circumcised! Savages!
Well for video yeah. For photos, not so much. However photomerging in Photoshop is a two step process. Select the photos, and hit OK. It does all the work for you.
"It will be able to capture an image with 1/100th the light intensity required by a DSLR"
I'm reading that as ultra fast shutter speeds being available for fast moving photography. Cool.
TFA doesn't say how many pixels it is.
One?
I can see this used by many professional photographers who would have much less post-processing to do as a sensor that size could produce an image of poster size with very little manipulation. Those pros who make a good living selling fine art photographs will be drooling over this as the size of the print possible with this sensor will be very large with very little pixelation.
"while democracy seeks equality in liberty, socialism seeks equality in restraint and servitude." de Tocqueville
Ahem. Remember that you have to account for the Bayseian Filter in front of the sensor.
It's not a "Bayseian filter" [sic], it's a Bayer matrix.
That's not the whole story. The actual size of the HST sensor is something like 45mm square (or maybe diagonally). Hubble takes amazing pictures for a few reasons. 1. It's got an 8 foot (2.4m) collecting mirror, so its light gathering prowess is amazing compared to normal cameras - like most telescopes. This means that the sensor is only effective because Hubble can direct so much light onto it. 2. It tracks the sky - like motorised ground based telescopes it is incredibly good at pointing in the same place for extended periods of time. So it can take longer exposures to get more light in. The Deep Field was taken with exposure times of roughly 1200 seconds, for instance. I assume it could expose for longer if it was at a Lagrange point and didn't have to contend with orbiting the Earth. 3. It's in space.. so there is very little in the way of light pollution (besides the sun!) and no atmospheric diffraction limit. Presumably they also make "panoramas" of the images to make them appear larger in print. The famous "Pillars of Heaven" shot is certainly not one image.
Perfect for capturing the Sorority girls in the next dorm over that turn-off the lights, but never close the curtains. "No honey I can't see you, but my camera can."
I think Infra-red cameras will work better for your case. Sure, the colors aren't natural, but it works much better in low light.
And how is a higher resolution sensor going to undo lens aberrations? That would be nice.
...the future crusty old bastards are already drinking the Kool-Aid.
You have this exactly backwards. The more you can stop down your lens, f2.8 wide open and f60 stopped down, means less light to your sensor, the greater your depth of field. This sensor means you could shoot at ISO 25, a shutter speed of 1/500 or 1/1000 of a second, and an fstop of 60 very easily in a lot less than full light conditions. That's a great depth of field, a shutter speed fast enough to reduce the effects of any vibration, and still get enough light to get a good exposure. I'm just guessing on what the fstop and shutter speeds would be with a sensor that light sensitive, but with a modern dslr you couldn't even get close to those settings in anything less than bright sunlight without very low shutter speeds that require the use of a tripod and higher ISO settings that tend to induce noise.
"while democracy seeks equality in liberty, socialism seeks equality in restraint and servitude." de Tocqueville
The higher res gives you the data to allow the software to do that. Canon's Digital Photo Profesional already does a lot of this automatically.
That's not true. Canon's professional cameras have larger sensors combined with higher pixel counts, however, once you hit a certain point where you're out resolving the lens, you're not going to get a whole lot out of adding more pixels without enlarging the area by more than that. Which is why the full frame 35mm format will always be capable of having more pixels than the APC-S or 4/3 formats will, at some point you hit the point of out resolving the lens at which point you're only option is to go larger. No technical wizardry in chip or in the camera hardware will ever make up for that.
Same goes for lens aberrations of various sorts, you can make them less obvious, but at the end of the day, you're still sacrificing image quality and counting on the camera system to do the right thing. But you're still going to lose detail and introduce other image problems.
Precisely, the best photographers whether they be pro or serious amateurs know that the best chance to get the shot is in camera. You can do a lot of those things in photoshop, but it really and truly isn't the same. Best case you limit the size at which you can print and more likely you get something of inferior quality. Beyond that, it just takes less time to do it right the first time, than to try and figure out how to fix it later on, even with raw formats it's still just not as good.
There was one shown off at trade shows in the late 1990s but it never went into production despite a lot of interest. Anyone with a microscope plus 35mm camera mount would have paid thousands for it instead of the tens of thousands to attach digital cameras.
Depth of field doesn't just relate to the aperture size (less DoF for wider aperature/small F-number), but also focal length of the lens (less DoF in telephoto lense or the larger SLR versus compact cameras) and the distance to focussed object (less DoF if it is closer to the camera). Your statement, as the other person points out is wrong. If it requires 1/100th of the light, you might have to close the aperture somewhat, increasing the DoF. This is a really cool DoF Calculator.
An SLR shutter doesn't expose the whole frame at the same instant: It's like a scanning line running down the frame, so if your gap (between the separately-controlled curtains) is small enough, you can have _any_ shutter speed you want - just don't expect the whole frame to be recording the same instant in time. Also, you don't need to put the shutter immediately in front of the film/sensor plate (but it helps give a clear image).
Yeah I have a good Nikon camera at home which I will never use again because it uses film. I feel bad about it but the sad fact is I can get better bang for my buck by buying a new DSLR.
http://michaelsmith.id.au
the merge process has to be built into the photo taking process to work with any level of confidence.
Inches? Could we please have that in useful units. Like football fields.
Have gnu, will travel.
With a Beowulf Cluster of these....
"Eve of Destruction", it's not just for old hippies anymore...
The whole point of sensors like this is that what may look to you like black with a few white dots is in fact a much more vivid image, but at intensities too low for human perception. When your sensor is capable of discriminating the differences in intensity or color between stars, stitching the exposures is very easy.
it's not that it's very easy most of the time... it's that it isn't anywhere near perfect all of the time.
A reasonable assumption would be that the sensitivity of the sensor is proportional to the area of the photosites (to a first approximation), so if this sensor is 100x as sensitive as, say, a D5MkII, then you would expect the photosites to be about 100x as big. Coincidentally, the 8in x 8in sensor is on the order of 100x the area of the 5D sensor, so the number of pixels is probably about the same (20M or so). To a first approximation, anyway.
I love Mondays. On a Monday, anything is possible.
Agreed. This sensor in a view camera, with lenses that would stop down far enough to really take advantage of the light sensitivity of the sensor, would be a mighty sweet camera. With the tilt/shift/swing of a good view camera and a very deep depth of field the creativity possibilities would be incredible.
"while democracy seeks equality in liberty, socialism seeks equality in restraint and servitude." de Tocqueville
Sure, you can do anything in software if you have all the necessary information. If the software is going to take out microscopic grinding defects in every lens, then you will have to characterize each lens with science-grade calibration equipment, and record what lens, what zoom (rotation) and everything when you take the picture, so that the software can back out the aberrations after the picture is taken. This is essentially what NASA does with all their telescopes. Removing the problem in hardware with a larger CCD and lens system is much, much simpler, especially for a field photographer with a variety of lenses and configurations.
8x8? Sigh, Back to Commodore-64 porn. "I don't know if it's a tit, but close enough, I'm horny!"
Table-ized A.I.
Lenses which cover 8*8 are basically large format lenses which include leaf shutters. Leaf shutters have a couple of problems - limited size, and a limited upper speed. Typically 1/500th is the fastest a leaf shutter will operate, and the limited diameter means you typically are down to f5.6 or f8 as a maximum aperture.
The maximum aperture will limit the speed advantage against a 35mm DSLR or medium format where f/2 and faster is common (f/1 can be achived at standard lengths if you compromise on image quality, say a noctilux). f/1 vs f/5.6 is 5 stops, or 32 times the amount of light. So an f/1 lens vs an f/5.6 will accept 32 times as much light to start with
The usual complaint about fast lenses is the limited depth of field. However, large format at f/5.6 will also suffer this problem as the larger image format will also offer a limited DOF, but in addition, a slow lens. I guess the answer will be to run the sensor at a higher ISO equivalent, make it more sensitive, and hence allow a smaller aperture to be used, but the tradeoff isn't obvious from the specifications
My guess is that this is a technology demonstrator, and will not be available to average punters on real cameras. Saying that an old large format camera with a 8*8 back would be very cool
You have to take into account noise on fingernail sized sensors. On this scale at 6MP, the noise floor would be very low.
ISO doesn't work like that, the sensors are celebrated so that the ISO values are roughly the same across all sensors (and the same as film too), so a compact camera set to ISO 100, f4 will require the same shutter speed as a full frame DSLR at ISO 100, f4
ISO 25 in broad daylight with a lens set to f60 would require (quick back of a fag packet calculation) 1/6 sec exposure no matter what the camera.
What I think you mean is that the sensor's high iso (say iso 12800) is the equivalent quality of what iso 25 would be on a standard DSLR
That's why uncropped Hubble images (like this one) have black squares in the corner - the Hubble camera can't take pictures there.
No sig today...
Pixel count isn't everything, especially these days.
That's why a 6 megapixel APS-C DSLR will blow away most 10-14 megapixel point and shoots in terms of image quality.
retrorocket.o not found, launch anyway?
If they use this for scientific work (which I imagine they might), something tells me they won't have any Bayer matrix on it, and will instead do multiple shoots with different whole-image filters, to avoid artifacts due to demosaicing.
Program Intellivision!
This is obviously a stunt to show off their process controls in production. They're telling the competition (and investors) that they can make an entire 12 inch wafer worth (40) of their 21 megapixel sensors without a defect anywhere on the wafer. Of course it'll be an expensive wafer, and they haven't said how many other wafers were rejected, but it's still impressive. It marks pretty much the ultimate maturity level for that production process and tells their competition, "You won't beat us on price for this geometry, go try something else." The only applications for something like this are wide-field low-light surveillance (e.g. for asteroid hunting).
Oh, I'm sorry sir, I thought you were referring to me, Mr. Wensleydale.