Canon develops 120 megapixel APS-H sensor

Fascinating, but I guess we will never see that one in a camera that you can actually buy (I'd love to see some sample images though).

...that's going to really put your glass through its paces...

ck3 wrote: Fascinating, but I guess we will never see that one in a camera that you can actually buy (I'd love to see some sample images though).
Very probably not. This is a research prototype and technology demonstrator, for Canon to be able to look ahad on the path of its commercial sensor design. A couple of years ago they made a 50MP APS-H sensor but it was never commercialised - the next commercial APS-H sensor they made was the 16MP one in the 1DIV (which actually performs rather better than the 50MP prototype). It would be commercially nuts for Canon to give us all the next several generations' MP increases in one go - there'd be nothing left for the next model.

Tyrone Wellhung wrote:
ck3 wrote: Fascinating, but I guess we will never see that one in a camera that you can actually buy (I'd love to see some sample images though).
Very probably not. This is a research prototype and technology demonstrator, for Canon to be able to look ahad on the path of its commercial sensor design. A couple of years ago they made a 50MP APS-H sensor but it was never commercialised - the next commercial APS-H sensor they made was the 16MP one in the 1DIV (which actually performs rather better than the 50MP prototype). It would be commercially nuts for Canon to give us all the next several generations' MP increases in one go - there'd be nothing left for the next model.
Correction: the 50MP APS-H sensor has not YET been commercialised. DSLR CPUs are not yet fast enough to process 50MP at 10fps. There's more engineering required to make a successful professional level DSLR than just the sensor.

To tell the truth...I'm excite about this about as much as when a new P&S comes out with a 14 MP sensor.The more MP,the worst quality.Sure,if it would come in a 72 X 96 MM sensor then,it would be different.

Is this sensor the same as a CCD, ( Charged Coupling Device, or chip) ?

Lesl wrote: .Sure,if it would come in a 72 X 96 MM sensor then,it would be different. Now that would be one heck of a camera wouldn't it!! LOL No matter what you're shooting, always keep God in the picture.

tweedle_dum wrote:
Tyrone Wellhung wrote:
ck3 wrote: Fascinating, but I guess we will never see that one in a camera that you can actually buy (I'd love to see some sample images though).
Very probably not. This is a research prototype and technology demonstrator, for Canon to be able to look ahad on the path of its commercial sensor design. A couple of years ago they made a 50MP APS-H sensor but it was never commercialised - the next commercial APS-H sensor they made was the 16MP one in the 1DIV (which actually performs rather better than the 50MP prototype). It would be commercially nuts for Canon to give us all the next several generations' MP increases in one go - there'd be nothing left for the next model.
Correction: the 50MP APS-H sensor has not YET been commercialised.
It will not be. It's performance in terms of QE and read noise (the major measures of sensor quality) were significantly worse than Canon is now achieving in its commercial sensors. There may be a 50 MP commercialised sensor (1DVI ?) but it won't be a commercialisation of that sensor, it will be a new development, very possible incorptorating some of the knowledge gained making both the previous 50MP sensor and this 120MP one.
DSLR CPUs are not yet fast enough to process 50MP at 10fps. There's more engineering required to make a successful professional level DSLR than just the sensor.
Indedd, I made that point myself somewhere. But the 50MP sensor's design will be 5 years old al least before a sensor of that resolution is required commercially, it would no longer be remotely competitive, especially since both Nikon and Panasonic's latest sensors have a higher QE than Canon can manage.

The really interesting thing is that the sensor is described as a commercial product. Hopefully we will be able to look at some sample images some time soon. Also, I wonder if the sensor has or even needs an AA filter and whether the technology can be adapted to the more popular FF and APSC sized sensors. I would also like to see what high ISO results look like and some lenses that do justice to this sensor. I'd love to see this sensor in a reasonably priced DSLR.

Chris59 wrote: The really interesting thing is that the sensor is described as a commercial product. Hopefully we will be able to look at some sample images some time soon.
I think you misread that - they compare it with their commercial sensors, not describe it as a commercial product.
Also, I wonder if the sensor has
generally the AA filter is not part of the sensor, it is a separate assembly that camera manufacturers add.
or even needs an AA filter
120MP equates to 13500x8950 pixels. On APS-H that's 480 pixels/mm or 241 lp/mm. Quite a few lenses give appreciable contrast at that resolution, so an AA filter would be wise. On the other hand, an AA filter designed for 480 pixels/mm will have a very small blurring effect in normal use.
and whether the technology can be adapted to the more popular FF and APSC sized sensors.
No reason to belive it can't be - it's just a case of laying out more or less of these pixels. I suspect the route to this sensor is the use of the pixel designs that Canon is developing for its P&S CMOS sensors, laying out enough of them for an APS-H sensor.
I would also like to see what high ISO results look like and some lenses that do justice to this sensor.
The 50MP sensor they developed was presented at the IISW, this one probably will also, but not until June 2011. When this is done, we will know how the sensor performs.
I'd love to see this sensor in a reasonably priced DSLR.
Not coming soon. At present the amount of processing it would require is probably prohibitive.

Wonder how much memory it will need to store and process an image.

If it is of comparable dynamic range to current sensors, it'll be about two bytes per pixel in RAW, or 240 mb uncompressed, normally less when compressed. As an uncompressed RGB image, it could be 120,000,000 x 3 bytes for 24-bit (360 mb) or 120,000,000 x 6 bytes for 16-bit (720mb), as stored on disk. Compressed, normally less. In memory, in order to take advantage of processor efficiencies, it would probably be stored in units of 4: red, green, blue, and a dummy or possibly an alpha channel per pixel. So here you'd have it consuming 480 mb as an 24-bit image, or 960 mb as a 48-bit (16-bit/channel) image. And in this case, it would be uncompressed, most likely. Not at all impossible for a machine with 4 gb of memory to throw around; but it's a great deal of data to process in one go, and you'd see that in increased execution times for anything from simple table-based white balance remapping to image re-display. And of course, this is just the start of memory consumption. Add a layer, you can easily double the memory cost - per layer - unless it's just a remapping layer. Undo records cost memory that are (at least) proportional to the area that was affected by the last operation. Sometimes more.

fyngyrz wrote: If it is of comparable dynamic range to current sensors,
per pixel, it probably isn't. An upper bound on the number of electrons collected per pixel is about 6000 (the Nikon D3S collects the most photons per unit area of any DSLR sensor at about 90000 per 12M pixel on FF). The best read noise per pixel achieved so far is about 1.6 e- on the 1DIV . This would shrink if this sensor includes a scaled version of the 1DIV's read transistor, but in the last Canon 50MP prototype Canon adopted a shared, larger OP transistor. If they use a 1 e- read transistor then the DR per pixel is 6000, requiring 13 bits per pixel to digitise it completely (no point having 14 bit digitisation).
it'll be about two bytes per pixel in RAW, or 240 mb uncompressed, normally less when compressed.
Usually raw files are packed down to the actual data needed, so this can be reduced by a factor of 13/16.

This is great, such advancements push the whole industry forward. Props to Canon. -- The Pentax Kx is a monster.

...is that we will see a new differentiation in medium size sensor camera models. There is no doubt in my mind will see this pixel density in medium-format sensors in cameras in due time: if Canon won't do it, someone else will: no technology advance, if otherwise commercially viable, can remain exclusive for a long time. However, to me this suggests that in due course the differentiation between FF and DX will become less and less important: what will become important is whether high-end cameras with this and similar sensor sizes (and here I consider everything from APS-C to FF to be all of the similar, "medium" size) are built big and heavy with sensors similar to this one, for relatively slow operation but maximum image resolution (landscape, architecture, fashion...) or light and portable, with sensors of lesser density but faster and less power hungry (travel, sports, photojurnalism) - i.e., the distinction which in a gentler age would have been indicated by a choice between Leica and Rolleiflex. bremzer

Medical and astrophotography spring to mind. Imagine the detail !! -- Given up on the whole "regain my original ID". Seems DPR don't give a flying fucher

bremzer wrote: ...is that we will see a new differentiation in medium size sensor camera models. There is no doubt in my mind will see this pixel density in medium-format sensors in cameras in due time: if Canon won't do it, someone else will: no technology advance, if otherwise commercially viable, can remain exclusive for a long time.
I would very much doubt this will happen. MF cameras have signally failed to keep up with the sensor advances of the 35mm DSLR sector, and I don't see any reason why they should start now. No MF manufacturer has the volume to fund development of a sensor like this, so they use Kodak and DALSA sensors developed for the science and aerospace industries, which are obloscent by DSLR tech. My guess is that it's quite likely that even by the time we see a 120MP Canon 1D, there won't be any MF manufacturers left. The Pentax 645D is temporarily popular, but I can't see it being sustained.
However, to me this suggests that in due course the differentiation between FF and DX will become less and less important: what will become important is whether high-end cameras with this and similar sensor sizes (and here I consider everything from APS-C to FF to be all of the similar, "medium" size) are built big and heavy with sensors similar to this one, for relatively slow operation but maximum image resolution (landscape, architecture, fashion...) or light and portable, with sensors of lesser density but faster and less power hungry (travel, sports, photojurnalism) - i.e., the distinction which in a gentler age would have been indicated by a choice between Leica and Rolleiflex.
I disagree completely. The differentialtion in sensor size between FF and APS-C is 2.56 in area, always more than a stop better noise performance, whereas the small sensored MF cameras have just 1.6 times the area of a FF camera. There is more difference between APS-C and FF than between FF and MF. More likely that MF will just fade away and FF will take its place (in essence, this happened with the 1DsII, which essentially pulled the rug from under the MF market. It hasn't died yet, but the doctor's given up hope.

Tyrone Wellhung wrote:
bremzer wrote: ...is that we will see a new differentiation in medium size sensor camera models. There is no doubt in my mind will see this pixel density in medium-format sensors in cameras in due time: if Canon won't do it, someone else will: no technology advance, if otherwise commercially viable, can remain exclusive for a long time.
I would very much doubt this will happen. MF cameras have signally failed to keep...
I was not talking about the class of cameras usually called MF (Pentax 645D etc.) at all. I simply consider that at present we have three broad categories of sensors commonly used for general photography: small, medium and large. "Small" are for me sensors typically used in P&S digital cameras, "medium" are those with sizes between (and including) APS-C and FF, and "large" are anything bigger, used in MF cameras (like Pentax you mentioned, digital backs etc.) With that context-specific use of physical sensor size attributes "small", "medium" and "large" clarified, I'd like to point out that I was speculating on the future of what we today commonly call DLSRs (i.e., both DX and FF), not the future of the camera class we today call "MF digital cameras". (My prediction for those, if I was to make it, might well be even grimmer than yours). bremzer

bremzer wrote:
Tyrone Wellhung wrote:
bremzer wrote: ...is that we will see a new differentiation in medium size sensor camera models. There is no doubt in my mind will see this pixel density in medium-format sensors in cameras in due time: if Canon won't do it, someone else will: no technology advance, if otherwise commercially viable, can remain exclusive for a long time.
I would very much doubt this will happen. MF cameras have signally failed to keep...
I was not talking about the class of cameras usually called MF (Pentax 645D etc.) at all. I simply consider that at present we have three broad categories of sensors commonly used for general photography: small, medium and large. "Small" are for me sensors typically used in P&S digital cameras, "medium" are those with sizes between (and including) APS-C and FF, and "large" are anything bigger, used in MF cameras (like Pentax you mentioned, digital backs etc.) With that context-specific use of physical sensor size attributes "small", "medium" and "large" clarified, I'd like to point out that I was speculating on the future of what we today commonly call DLSRs (i.e., both DX and FF), not the future of the camera class we today call "MF digital cameras". (My prediction for those, if I was to make it, might well be even grimmer than yours). bremzer
If you're predicting that the 'large' sensor size will be 24x36 and the 'medium' will be APS-C, I'd agree with you. I don't know what the small will be, though Nikon's 2.5x EVIL might catch on - especially if Canon does something similar. Acting together to go on a tangent the others haven't, the big 2 could change the market (though there's probably something illegal about that).

Good afternoon,
bremzer wrote: ...is that we will see a new differentiation in medium size sensor camera models. There is no doubt in my mind will see this pixel density in medium-format sensors in cameras in due time: if Canon won't do it, someone else will: no technology advance, if otherwise commercially viable, can remain exclusive for a long time. However, to me this suggests that in due course the differentiation between FF and DX will become less and less important:
Allow me to stop you right there, my good man. You won't see this on 35mm sensors without major changes to the manufacturing processes. And that's from Canon themselves. Check out this Wikipedia article: http://en.wikipedia.org/wiki/Sensor_sizes As Canon say: "Production costs for a full frame sensor can exceed twenty times the costs of an APS-C sensor. Only 35 full frame sensors can be produced on an 8 inches (20 cm) silicon wafer that would fit 112 APS-C sensors and there is a significant reduction in yield due to the large area for contaminants per component. Additionally, the full frame sensor requires three separate exposures during the photolithography stage, which requires separate masks and quality control steps. The APS-H size was selected since it is the largest that can be imaged with a single mask to help control production costs and manage yields." The reference Wikipedia has for that text is a 2006 Canon press release (which makes interesting reading if you know nothing about silicon wafers), so this isn't a competitor attempting to spread FUD. The manufacturing techniques are pretty much unchanged, so there's very little wiggle room for "maybe it's different today", I'm afraid. In a nutshell, the economics are such that 35mm sensors are always going to be much more expensive than APS-C.

Philip Storry wrote: Frankly, I'll come straight out and say that I regard 35mm as becoming an anachronism. For almost all photographic needs, APS-C is a better solution as a sensor size. The few advantages 35mm has are either niche (stupidly shallow DOF) or fleeting, as history has shown that APS-C is always hot on the heels of 35mm sensor sizes in performance terms.
This isn't the case. History shows that each generation of sensor technology, FF and APS-C remain exactly the same distance apart, the distance dictated by the 2.5x sensor area of FF, which means that FF gathers 2.5 times the number of photons as APS-C at the same exposure. Of course, at the same exposure FF needs less DOF or longer shutter speed, but that's often a trade people will make for a stop better noise.
It seems that only the desperate marketing attempts of incumbent companies with significant 35mm lens designs is keeping 35mm alive as a dream, and even then it's an expensive one.
The success of Nikon's FX line and the 5DII suggests that FF is enough of a necessity for many to pay for it.
I wish that they'd just kill it and offer decent APS-C designs - something that some companies have been much better at than others.
There are some superb cameras. Where APS-C systems fall short is highest quality designed for APS-C lenses, offering smaller size or the same apertures as equivalent FF lenses.
It particularly bugs me that the only reason 35mm is "better" is that it's the sensor chosen for flagship cameras -
it's that 2.5 times the sensor area.
when for a sports professional, for example, a smaller APS-C system could be lighter,
if there were APS-C specific lenses
process more frames per second at a given Mhz of circuitry,
that's down to pixel count, not sensor size.
and would still be more than acceptable for just about every publication purpose.
you might be right. But I think the camera manufacturers remeber just how successful was Nikon's attempt to address the pro marketplace with APS-C cameras against competion with larger formats.
However, my opinions on the reality of 35mm aside, the economics simply mean that this technology is always going to be in favour of APS-C designs and compact camera designs. If Canon could do this "technology demonstrator" on 35mm, they would have.
Not necessarily. Simply, if you're making a prototype, the stitching required for an FF sensor would be an unnecessary complication.
But frankly, at approximately 20 times the cost, I suspect they're quite happy to do it on APS-H and let the huge MPixel count take the headlines.
I think you'll find wafer processing costs are a very small portion of the costs of this sensor. 20x the wafer costs would have made little to no difference in the overall cost of the exercise.
(And you'll probably see this technology in compact cameras as a 1/2.5" sensor long before you ever see it in an APS-C camera, too... Assuming you could get Canon to admit any such thing about a 1/2.5" sensor, that is!)
This is very likely simply an exercise in using Canon P&S pixels to make a large sensor for publicity purposes. On 1/1.7" this pixel size yields a 9MP sensor. If the G series continues to drop in MP count, this could be the pixel from the G12.

Tyrone Wellhung wrote: This is very likely simply an exercise in using Canon P&S pixels to make a large sensor for publicity purposes.
I disagree. Assuming the image quality is good, then the ability to pull 122 MP off a sensor at 9.5 fps is a significant achievement for Canon.

DSPographer wrote:
Tyrone Wellhung wrote: This is very likely simply an exercise in using Canon P&S pixels to make a large sensor for publicity purposes.
I disagree. Assuming the image quality is good, then the ability to pull 122 MP off a sensor at 9.5 fps is a significant achievement for Canon.
I'm not completely convinced it's that significant achievement. It just means having a lot of output channels, which isn't particularly hard in a prototype implementation. In production, the cost of very many off chip A to D's would probably be prohibitive. In fact, it could well be a sign that this is a prototype assembled from existing designs, they just take a column readoff design intended for a smaller sensor and replicate it many times.

Phillip There seem to be plenty of people that will spend a lot of money buying huge cameras and massive lenses to look like what they regard as a "pro" (Something like someone who uses a D3s with a fast 400mm lens at high fps) Oh I forgot, it now seems your camera is not "pro equiptment" unless it can be used to shoot a black cat in a coal cellar hand held in available light as well. Anyway there seem to be plenty of people prepared to buy these large cameras and lenses at huge prices - even "upgrade" already expensive lenses for tiny improvments that will almost never matter. So long as there are a lot of people who can convice themselves they need them, the manufactures will surely go on producing them with even higher and higher specs. I've always liked nice cameras, but if I can get prints with IQ comparible to those I could from a good medium film camera using a moderate iso and aperture on a digital camera I don't care about format - in fact the smaller the better. A full frame camera like Sony's A850 (which I find fairly big, heavy and obtrusive) now does this easily - if similar resolution IQ and DR to that can be achieved with a smaller APS-C camera all the better.
  • C

Philip Storry wrote: ... I wish that they'd just kill it ... ... It particularly bugs me ...
For us to wish that the behavior of industrial conglomerates to be this or that, or to be annoyed when it is not, strikes me as unreasonably emotional; we might as well wish for a nicer climate and be annoyed that our wishes are ignored (well, in some parts of the world :). All we can do is carefully observe the technology (or weather) trends and make our personal decisions accordingly. bremzer

Philip Storry wrote: Good afternoon,
bremzer wrote: ...is that we will see a new differentiation in medium size sensor camera models. There is no doubt in my mind will see this pixel density in medium-format sensors in cameras in due time: if Canon won't do it, someone else will: no technology advance, if otherwise commercially viable, can remain exclusive for a long time. However, to me this suggests that in due course the differentiation between FF and DX will become less and less important:
Allow me to stop you right there, my good man. You won't see this on 35mm sensors without major changes to the manufacturing processes. And that's from Canon themselves. Check out this Wikipedia article: http://en.wikipedia.org/wiki/Sensor_sizes As Canon say: "Production costs for a full frame sensor can exceed twenty times the costs of an APS-C sensor. Only 35 full frame sensors can be produced on an 8 inches (20 cm) silicon wafer that would fit 112 APS-C sensors and there is a significant reduction in yield due to the large area for contaminants per component. Additionally, the full frame sensor requires three separate exposures during the photolithography stage, which requires separate masks and quality control steps. The APS-H size was selected since it is the largest that can be imaged with a single mask to help control production costs and manage yields." The reference Wikipedia has for that text is a 2006 Canon press release (which makes interesting reading if you know nothing about silicon wafers), so this isn't a competitor attempting to spread FUD. The manufacturing techniques are pretty much unchanged, so there's very little wiggle room for "maybe it's different today", I'm afraid.
There are reasons to believe that since the time of that comment larger steppers and other equipment used during photolithography allow for a single pass to be made.
However, my opinions on the reality of 35mm aside, the economics simply mean that this technology is always going to be in favour of APS-C designs and compact camera designs. If Canon could do this "technology demonstrator" on 35mm, they would have. But frankly, at approximately 20 times the cost, I suspect they're quite happy to do it on APS-H and let the huge MPixel count take the headlines. (And you'll probably see this technology in compact cameras as a 1/2.5" sensor long before you ever see it in an APS-C camera, too... Assuming you could get Canon to admit any such thing about a 1/2.5" sensor, that is!)
The Canon S95 sensor, if scaled up, would be 200MP at full size. The S95's sensor is 23MP/cm2, this 120MP APS-H is 20MP/cm2. It would seem that the bigger sensors are still catching up with the pixel resolution of the smaller ones.

Now that the cats out of the bag,I'll bet it wont be far off before it really is available. Of course for a 120MB price too. -- No matter what you're shooting, always keep God in the picture.

tweedle_dum wrote:
Philip Storry wrote: Good afternoon,
bremzer wrote: ...is that we will see a new differentiation in medium size sensor camera models. There is no doubt in my mind will see this pixel density in medium-format sensors in cameras in due time: if Canon won't do it, someone else will: no technology advance, if otherwise commercially viable, can remain exclusive for a long time. However, to me this suggests that in due course the differentiation between FF and DX will become less and less important:
Allow me to stop you right there, my good man. You won't see this on 35mm sensors without major changes to the manufacturing processes. And that's from Canon themselves. Check out this Wikipedia article: http://en.wikipedia.org/wiki/Sensor_sizes As Canon say: "Production costs for a full frame sensor can exceed twenty times the costs of an APS-C sensor. Only 35 full frame sensors can be produced on an 8 inches (20 cm) silicon wafer that would fit 112 APS-C sensors and there is a significant reduction in yield due to the large area for contaminants per component. Additionally, the full frame sensor requires three separate exposures during the photolithography stage, which requires separate masks and quality control steps. The APS-H size was selected since it is the largest that can be imaged with a single mask to help control production costs and manage yields." The reference Wikipedia has for that text is a 2006 Canon press release (which makes interesting reading if you know nothing about silicon wafers), so this isn't a competitor attempting to spread FUD. The manufacturing techniques are pretty much unchanged, so there's very little wiggle room for "maybe it's different today", I'm afraid.
There are reasons to believe that since the time of that comment larger steppers and other equipment used during photolithography allow for a single pass to be made.
Could you cite some evidence for that? So far as I can see, all the new steppers and scanners from Nikon and Canon conform to the 26x33mm reticule size that they've had as standard for some time. Canon has produced a single type with a 50mm square reticule size, but its for 500nm geometry, which is useless for a sensor of this density (although it could be used for patterning the microlens mould). Nikon also makes scanners with much larger reticules for the LCD industry, but again the geometry would be far too coarse for a sensor like this.

Tyrone Wellhung wrote:
tweedle_dum wrote:
Philip Storry wrote: Good afternoon,
bremzer wrote: ...is that we will see a new differentiation in medium size sensor camera models. There is no doubt in my mind will see this pixel density in medium-format sensors in cameras in due time: if Canon won't do it, someone else will: no technology advance, if otherwise commercially viable, can remain exclusive for a long time. However, to me this suggests that in due course the differentiation between FF and DX will become less and less important:
Allow me to stop you right there, my good man. You won't see this on 35mm sensors without major changes to the manufacturing processes. And that's from Canon themselves. Check out this Wikipedia article: http://en.wikipedia.org/wiki/Sensor_sizes As Canon say: "Production costs for a full frame sensor can exceed twenty times the costs of an APS-C sensor. Only 35 full frame sensors can be produced on an 8 inches (20 cm) silicon wafer that would fit 112 APS-C sensors and there is a significant reduction in yield due to the large area for contaminants per component. Additionally, the full frame sensor requires three separate exposures during the photolithography stage, which requires separate masks and quality control steps. The APS-H size was selected since it is the largest that can be imaged with a single mask to help control production costs and manage yields." The reference Wikipedia has for that text is a 2006 Canon press release (which makes interesting reading if you know nothing about silicon wafers), so this isn't a competitor attempting to spread FUD. The manufacturing techniques are pretty much unchanged, so there's very little wiggle room for "maybe it's different today", I'm afraid.
There are reasons to believe that since the time of that comment larger steppers and other equipment used during photolithography allow for a single pass to be made.
Could you cite some evidence for that? So far as I can see, all the new steppers and scanners from Nikon and Canon conform to the 26x33mm reticule size that they've had as standard for some time. Canon has produced a single type with a 50mm square reticule size, but its for 500nm geometry, which is useless for a sensor of this density (although it could be used for patterning the microlens mould). Nikon also makes scanners with much larger reticules for the LCD industry, but again the geometry would be far too coarse for a sensor like this.
Indeed, that would be a major revelation if it were true... The fact of the matter is that the scanning exposure field is limited in size by the size of the image that the projection lens can produce. To up scale this lens would be prohibitively expensive to produce. This is in fact, why stepper became scanners in the first place.

martin_vernon wrote:
Tyrone Wellhung wrote:
tweedle_dum wrote:
Philip Storry wrote: Good afternoon,
bremzer wrote: ...is that we will see a new differentiation in medium size sensor camera models. There is no doubt in my mind will see this pixel density in medium-format sensors in cameras in due time: if Canon won't do it, someone else will: no technology advance, if otherwise commercially viable, can remain exclusive for a long time. However, to me this suggests that in due course the differentiation between FF and DX will become less and less important:
Allow me to stop you right there, my good man. You won't see this on 35mm sensors without major changes to the manufacturing processes. And that's from Canon themselves. Check out this Wikipedia article: http://en.wikipedia.org/wiki/Sensor_sizes As Canon say: "Production costs for a full frame sensor can exceed twenty times the costs of an APS-C sensor. Only 35 full frame sensors can be produced on an 8 inches (20 cm) silicon wafer that would fit 112 APS-C sensors and there is a significant reduction in yield due to the large area for contaminants per component. Additionally, the full frame sensor requires three separate exposures during the photolithography stage, which requires separate masks and quality control steps. The APS-H size was selected since it is the largest that can be imaged with a single mask to help control production costs and manage yields." The reference Wikipedia has for that text is a 2006 Canon press release (which makes interesting reading if you know nothing about silicon wafers), so this isn't a competitor attempting to spread FUD. The manufacturing techniques are pretty much unchanged, so there's very little wiggle room for "maybe it's different today", I'm afraid.
There are reasons to believe that since the time of that comment larger steppers and other equipment used during photolithography allow for a single pass to be made.
Could you cite some evidence for that? So far as I can see, all the new steppers and scanners from Nikon and Canon conform to the 26x33mm reticule size that they've had as standard for some time. Canon has produced a single type with a 50mm square reticule size, but its for 500nm geometry, which is useless for a sensor of this density (although it could be used for patterning the microlens mould). Nikon also makes scanners with much larger reticules for the LCD industry, but again the geometry would be far too coarse for a sensor like this.
Indeed, that would be a major revelation if it were true...
http://www.nikon.com/products/precision/lineup/fx/fx-803m/index.htm

That was Nikon's thinking too, when they made D2h. All photographers just went and bought Canon instead. The rest of your post wasnt any good either, btw.
Philip Storry wrote: when for a sports professional, for example, a smaller APS-C system could be lighter, process more frames per second at a given Mhz of circuitry, and would still be more than acceptable for just about every publication purpose.

I can see it now: Canon Powershot GXX, 20 MP, ISO 25600, RAW, 3 FPS, HD 24P Video !! -- http://hdr.strivearth.com | canon fodder

Strange huh? Sony "throws down the gauntlet" with their new cameras, and Canon scrambles to make the news by announcing a sensor that has no practical use for anybody! Seems like more of a PR stunt, rather than a true announcement.

Go back to your cave Sony troll.
DavidVogt wrote: Strange huh? Sony "throws down the gauntlet" with their new cameras, and Canon scrambles to make the news by announcing a sensor that has no practical use for anybody! Seems like more of a PR stunt, rather than a true announcement.

The Sony troll with a Nikon signature, huh? Interesting theory you got there.

Hey, at least I got half of it right. Whether it's Nikon or Sony, your attitude is all TROLL. So get lost.
DavidVogt wrote: The Sony troll with a Nikon signature, huh? Interesting theory you got there.

Seaclam wrote: Hey, at least I got half of it right. Whether it's Nikon or Sony, your attitude is all TROLL. So get lost.
Why? The truth hurts? The truth is that canon is 1-2 steps behind the competitions. No innovations in last few years from canon. Maybe they lost the engineers to sony or pansonic.

Oh PLEASE! Get a frick'n life will ya.....
sorinx wrote:
Seaclam wrote: Hey, at least I got half of it right. Whether it's Nikon or Sony, your attitude is all TROLL. So get lost.
Why? The truth hurts? The truth is that canon is 1-2 steps behind the competitions. No innovations in last few years from canon. Maybe they lost the engineers to sony or pansonic.

This is not a PR stunt. It seems they improved upon their aps-h 52MP image sensor they made news with about 2 years ago. Canon is usually #3 in the world in patents filed each year. They produce lots of advancements.

Absolutely, you are completely right. But nobody can tell me that it's a coincidence that this announcement was made on the same day that Sony "throws down the gauntlet" on Canon. (DPR's wording, not mine.) Don't get me wrong, I understand it's just a smart marketing move to take some of the attention off of Sony.

Back in 07 they did the 50mp sensor and no 50mp camera at all. So i think this time around it may be the same thing.
DavidVogt wrote: Strange huh? Sony "throws down the gauntlet" with their new cameras, and Canon scrambles to make the news by announcing a sensor that has no practical use for anybody! Seems like more of a PR stunt, rather than a true announcement.

The chip is listed as 29.2 mm and 13280 pixels wide which I calculate as 2.2 micron pixels. This is equivalent to 227 lp/mm since 2 pixels are required per line pair. Calculating the MTF due to diffraction for .55 micron green light at this resolution using the diffraction MTF formula: 2/pi * (acos(f) - f * sqrt(1-f * f)) Where f is the fraction of the diffraction cutoff frequency of 1/(wavelength * f-stop) Yields the following result in diffraction MTF percent contrast: f-stop: MTF 2.8 : 56.4 4.0 : 39.1 5.6 : 18.8 7.9 : 0.2 Since for low-ISO images sharpening can be used to increase the contrast of high frequency details I would consider contrasts at about 10 percent and above before sharpening to be useful. This sensor should therefore be well matched to those lenses which have optimum performance in the f4 to f5.6 range, which includes f2.8 and faster prime lenses: especially macro lenses and super-telephotos.

what is the exact formula for those MTF values (pixel size included)? the larger the sensor, the worse the lens scaling issues (which is why pinhead sensor cameras are not that whole lot worse than much larger ones); so an f/2.8 is first geometry limited, rather than diffraction.

A sensor like this could clearly resolve a lot of detail at low ISO. I wonder if they have some HW pixel binning capabilities to improve the noise performance at high ISO. After all, even with 4:1 pixel binning, this will still be a 30MP sensor! The linear resolution of this sensor is roughly twice that of EOS 7D. In order to take advantage of the increased resolution, Canon may have to revise a substantial portion of the EOS lens portfolio. Most of the supertele lenses that take a 2x TC well should perform excellent even on a sensor of this resolution (w/o the TC, of course), but the WA lenses will be a different story...

This is just a sensor. -gt

ISO is any combination of hardware and software signal amplification that yields the desired level of tonal output from the given exposure (light intensity).

The way I knew it, ISO relates to signal amplification done before quantization. If A/D conversion would have electron precision, this wouldn't hold an advantage to PP. On a CMOS sensor (highly integrated) the ISO range may be known, on a CCD it may not.

Add new comment

Image
More information
  • Files must be less than 2 MB.
  • Allowed file types: png gif jpg jpeg.
Attachment
More information
  • Files must be less than 2 MB.
  • Allowed file types: zip rar.
By submitting this form, you accept the Mollom privacy policy.