What's Next from Canon?

[Diko]

CIPA aggregates the data, there are many years' worth here:

http://www.cipa.jp/stats/dc_e.html

Enjoy!

You are the best... but I guess you know that! ))

 

[jrista]

The Q.E. is indeed high. I don't know about 90%, even with a BSI design unless he is supercooling, there is going to be a certain amount of loss due to dark current.

Actually that is the idea: the Q.E. to be at almost 100%. Here is an extract of some more recent materials about QIS:

Fossum writes:
QIS "vision" is to count every photon that hits the sensor, recording its location and arrival time, and create pixels from bit-planes of data.

That sounds to me as 100% Quantum Efficiency ;-) No?

There is a very significant difference between "vision" and "reality". The vision may indeed be to count every photon. The reality is, in order to achieve that, the sensor would have to be superconducting. That's the only way you can count every photo. The Titanium Nitride video sensor I linked is a photon counting, position recording, exact wavelength detecting sensor. It is about as close to Fossum's vision as modern technology gets. The only difference is it doesn't use jots and dynamic grains. The reality is, that TiN sensor is cooled to superconducting supercool temperatures.

If your thinking that someday Fossum's QIS is going to pan out to a hand-holdable photon-counting DSLR (or for that matter even a DSLR with 90% Q.E.), your gravely mistaken. It isn't possible to cool electronics to a fraction of a degree above absolute zero in a hand-holdable package, and room-temperature superconductive materials, as much as they are the holy grail of the electronics industry, simply haven't been discovered, and the more time passes, the liklihood of finding a room-temperature superconductor diminishes (research has been ongoing for decades, and every time someone "discovers" a room temp superconductor, it always pans out to be false.) This is all assuming that actual photon-counting is possible with any superconducting material above absolute zero...dark current is the photon counting killer.

Having a high Q.E., however, does not change the notion of digital grains. In the presence of low light, you have low incident photon counts. The whole entire DFS/QIS design is based not just on jots, but on the fact that jots are organized into dynamic grains. In low light, all it takes is ONE jot to receive a photon in a grain for the ENTIRE grain to be activated. Let's say grains start out containing 400 jots each (20x20, a 16µm pixels...HUGE). Lets say were shooting in very low light, starlight. The moment one jot in each 20x20 size grain receives a few photons (lets say 50% Q.E., so two photons), then all 400 of those jots are marked as active! So, under low light, it might seem as though you actually received 800 photons, rather than just two! Big difference...you are now simulating the reception of a lot of light, however it is at the cost of resolution. At 16µm a grain, your resolution is going to be pretty low by modern standards...roughly 3.375mp.

Now, lets say a crescent or half moon comes out, and we take the same picture again. We have about two to three more stops of light. Instead of two incident photons, we now have ~8 incident photons per grain. Lets say a dynamic grain division is set at 8 photons. Once our jots receive and convert eight photons, our grains all split. We now have four times the resolution (10x10 grains, or 100 jots per grain, four times as many grains). We have a stronger signal overall, but roughly the same signal per grain as we did before. However we now have an image with four times as many megapixels, 13.5mp to be exact.

Now a full moon is out, and we take the same picture. We have another two stops of light. We get about 32 incident photons. Our grain size is now 5x5, or 25 jots per grain. Our resolution has quadrupled again. Same overall SNR, but our image resolution is 54mp.

This is what Eric Fossum has designed. A totally dynamic sensor that adjusts itself based on the amount of incident light, maintaining relative signal strength and SNR regardless of how much light is actually present. It does this by dynamically reconfiguring the actual resolution of the device...very low light, very low resolution, low light, low resolution, adequate light, good resolution, tons of light, tons of resolution. Technologically it is pretty advanced, conceptually it is relatively strait forward.

I've greatly exagerrated the scenario above...you wouldn't be able to have 54mp under moonlight. You would probably have something closer to 0.8mp under starlight, maybe 3mp under full moonlight, 13.5mp under morning or evening light, and maybe finally be able to achieve 54mp under full midday sunlight.

Actually he intends to put more than 4K jots in 1 pixel However I believe:

At 4k jots per "pixel" (pixels don't really exist in the DFS concept, not sure about any more recent QIS papers), if we assume he is using an 800nm jot pitch, that would mean you have 45,000 jots across and 30,000 jots down in a 36x24mm sensor. That makes a grain/pixel with 4000 jots (63x63 jots per grain) over 50µm in size. I mean, there are physical limitations here. We can't break the laws of physics, not even if we are Eric Fossum. Make jot size much smaller than 800nm, and you'll start filtering out red light. You can't have a color accurate sensor if you do that...not at room temperature anyway.

(Based on one of the papers you linked, it isn't actually 4096 jots per pixel. It is 16x16 jots read 16 times to produce one frame, 16 times 16 (physical dimensions) times 16 (time) is 4096. Based on other information in the article, it sounds like his jots are about 1µm in size, or 1011nm to be exact.)

Now, if we move back into the realm of superconductive TiN sensors at absolute zero, you could probably make jots 100nm, maybe even 10nm in size, have near-perfect positional measurement accuracy by measuring broken cooper pairs. Since your measuring the exact energy in each incident photon, your jots are recording the exact wavelength of the disturbance in the superconductor. The TiN sensor I linked here uses the same general concept Fossum put forward on 2005...taking minuscule time-slices of an exposure by reading the sensor hundreds or thousands of times per second, and integrating the result. That gives it essentially infinite dynamic range if you expose for long enough (although you would still be limited if you needed shorter exposures, however that limit would still be considerably higher than modern day sensors...18-20 stops maybe.)

There are other physical problems that have to be solved before this technology would even be viable. According to another one of Fossum's more recent papers, were talking about excessive data throughput rates. The fastest data throughput rates we have today for storage are on the order of gigabits per second. A high end PCI-E type SSD can move around a gigabyte and a half per second or so, which is about 12 gigabits per second. Fossum's QIS concept requires 100 TERRABIT per second throughput. That is 12.5 terrabytes per second!!! That is unprecedented transfer speed. I mean, MASSIVELY UNPRECEDENTED. I don't even know that I've heard 1/100th of those kinds of throughput rates for single supercomputer throughput channels. You would have to bundle hundreds if not thousands of the high speed fiberchannel connections usually used with supercomputing in order to achieve a terrabit of throughput. Fiberchannel, one of the fastest transfer channels , tops out at around 25-gigabit per second, or about 0.00025x as fast as would be necessary for a QIS sensor to operate effectively. To be really clear about this, the fastest data channel on earth can currently only achieve 0.00025x what is necessary to support Fossum's QIS concept. It is still 0.025x the necessary throughput rate to achieve even 1 terrabit/second throughput. Even the on-chip data cache of a modern CPU is still only pushing a couple hundred gigabits/second throughput to the CPU registers, and that is thanks to exceptionally short physical data paths. In a digital camera, the image information has to be shipped off-sensor to a processor powerful enough to integrate hundreds of individual jot frames per second. Even if were talking about an integrated stacked sensor and DSP package, the distance those frames have to travel would make achieving 1-100tbit/s throughput difficult without some radical new breakthrough in bit transfer technology.

There are significant challenges in order to make Fossum's DFS/QIS concept a reality. Which is why, even after at least nine years, it is still just a concept.

1/ your info might be a little out-of-date.

2/ Fossum knows what he is doing if he is doing it for more than 10 years now. And he already has created something befre (the CMOS).

3/ I hope you will agree that we both are a little bit behind - no matter how much we know, with our understanding of this TO-EMERGE technology ;-)

MAY-EMERGE technology. As I said above, there are some pretty massive physical and technological issues to overcome. I've never heard of a photon-counting sensor that used sensing elements as small as a jot that wasn't supercooled. A data throughput rate of 1-100tbit/s is not only unprecedented, but could very possibly be impossible without integrating the entire concept onto a single die, and even then, at a tbit/s throughput, that little sensor+dsp die is going to get exceptionally hot (even supercooled, your producing a hell of a lot of energy in an extremely concentrated space, meaning you run a high risk of heating the sensor above the point where it can behave as a superconducting device...either that, or you need orders of magnitude more power to keep it all cool).

You don't need to know what technological advances may come down the road in the future to know that the QIS concept is running up really close to the laws of physics. It's very likely it is bending them as far as they can go, and it may not be enough.

...
Absolutely. I'm 100% sure. It makes no sense for Canon to try to break into a niche market that already has not only it's dominant players, but dominant players with a HELL of a LOT of loyalty among their customers. There have been Canon MF rumors for years. I remember reading MF rumors here back in the 2005 era. Nothing has ever come of them, despite how often Northlight tends to drag the subject back out.

The only way Canon could make a compelling entry into MF is if they launched an entire MFD system. Cameras with interchangable backs, image sensors that at least rival but preferably surpass the IQ of the Sony MF 50mp, a wide range of extremely high quality glass (they are certainly capable here, but it still is a MASSIVE R&D effort), and a whole range of necessary and essential accessories like flash. Canon has to do this all UP FRONT, on their own dime, to cover the massive R&D effort to build an entirely new system of cameras that can compete in an already well established market.

Now, they've done that once. They did it with Cinema EOS. But the cinema market is a lot broader with more players, and is a significant growth market with the potential for significant long-term gains, even for a new entrant like Canon. The medium format market is not a growth market. It's a relatively steady market, that has its very few players and it's loyal customers. Since there are so few players who already dominate the market, breaking in for a new player like Canon would be a drain on resources, and there is absolutely zero guarantee of any long-term payoff.

So, yes, I'm sure. Canon won't be offering a medium format camera any time soon.

OK.
- Yes about SYSTEM, of course. I have never imagined CANON selling digital backs, or sensors to anyone )))
- Yes about glass
- No about light
- Perhaps CANON has been in the MF R&D since 2001 with the introducing of 12"' Si wafer
Let us not forget the BIG SENSOR or the BIG 120 MPs APS-H sensor - the 2007 success?

If Canon had been developing an MF system for 13 years, then they have already failed. They've failed multiple times over. Sorry, but I find that idea exceptionally unlikely.

The "BIG SENSOR" was designed for an entirely different division of Canon for use in scientific grade imaging. It will never be used in a DSLR type camera. The 120mp APS-H was an APS-H sensor...that is smaller than FF, and significantly smaller than MF (MF sensors start around 44x33, and get as large as 60x70...anything smaller, and were not talking medium format.)

Silicon Wafer Sizes Trend The picture I provide is more relevant to intel then to SONY or CANONn and yet it is a trend:

I know all about wafer size trends. Canon still uses 200mm wafers for their APS-C and FF sensors as far as I know. Another indication that they haven't and are not moving into MF any time soon. On a 200mm wafer, you can fit 24 44x33mm sensors. On a 300mm wafer, you can fit 54 sensors, with less waste. Assuming Canon somehow skipped 300mm and went strait to 450mm (highly unlikely, the 450mm wafer size still seems to be somewhat fragile), you could fit 130 sensors, with even less waste. I don't see Canon manufacturing MF sensors on 200mm wafers.

Let me make another comparison exactly with the small Cinema EOS success. It's like an early bird. FF sensor from DSLR equipment against ARRI, RED & SONY APS-C Cinema solutions..... Hmmm... Who knows... ;-) Extra dollar is always welcomed. Even if it is from 0.5 market share. If CANON succeeds to sell 2k MF bodies in 3 years, let's say 10K$ each.... 2 million extra dollars... I ask

You don't seem to understand the market dynamics involved here. Cinema is a big, huge growth market into the future. There is massive potential for Canon, who already has an exceptional reputation in photography, to make big inroads into the Cinema market space as it grows, grabbing up new customers, many of whom are already familiar with Canon video from using their video-capable DSLRs. Canon already has a name in that industry thanks to the 5D II, which has been used in a number of relatively big name productions for TV and even a few big name movies.

Canon's break into the cinema market is easy. It cost them little to integrate video into the 5D II, which gave them their initial foothold, paving the way for them to expand that foothold into a legitimate presence. Since the market is a growth market, the risk is relatively low compared to an entry into MF, because you can grab new customers who are just moving to digital cinema cameras and have yet to buy into a system.

This is in contrast with the medium format market. It's growth opportunities, such as they are, are small. That means the primary source of market share gain has to come from existing dominant players. The market is relatively closed, with relatively few needs and a small base of customers to start with.

You could compare Cinema EOS as sprinting up a gentle slope, the wind of the 5D II success at their backs, with Canon MFD as climbing up an ice cliff with the wind buffeting them around their precarious perch. Medium format is a loss/loss for Canon. Excessive up-front cost, uphill climb once they finally enter the market full of very few customers and low growth.

Sorry...still don't believe it's going to happen. Not with Canon, anyway, not right now, not with global economies still in the pits relative to their 2007 peaks.

WHY NOT? ;-)

See above.

 

[Don Haines]

If your thinking that someday Fossum's QIS is going to pan out to a hand-holdable photon-counting DSLR (or for that matter even a DSLR with 90% Q.E.), your gravely mistaken. It isn't possible to cool electronics to a fraction of a degree above absolute zero in a hand-holdable package

Not even for winters in Canada....

 

[Diko]

;D ;D ;D ;D ;D

You remind me of this:

Your knowledge is adorable... And yet quite interesting that you DO continue to fight with the idea that he may have did it.

If you stop just for a second using your knowledge from here whereby this is an improvement of the CMOS:

And the regular physics knowledge... Yeah I had that moment with the jots and the wavelength just as you did. But I do know that he is quite longer in this business than you and me together ;-)

DO you believe that he will reveal every detail of his study to the world so someone could steal it from him? ;-) And do you believe that he would continue for 10 years to research in that field within a reputable university and some sponsor (it could be even Samsung)?

IF scientist were so sure like you and so negative... we would be in the medieval age, no offence.

There is a pdf (if in those presentations not included back dated from 2010, if I recall correctly where Fossum represents that they are to try to implement that technology (of course quite away from Q.E. of 100%) in 3 stages....
The first one on a regular CMOS. The last one on a new superconducting material... so there you go...

As for:

There are significant challenges in order to make Fossum's DFS/QIS concept a reality. Which is why, even after at least nine years, it is still just a concept.

Please go back where I first mentioned the QIS. I said that at the moment the current technology is not ready yet for QIS or something along those lines.

Additionally I said that our brains would be half a kilo smaller on average. Let me elaborate - that means about 20 -30 years from now... Unless you are younger than me ;-) CHEAT: live a healthy life!

As for the CANON.. Officially 200mm is what I have head as well about CANON... but you have to admit that if you were CANON you wouldn't reveal of you are already on a 300 or 450mm wafer, now would you?

A proof that is this very topic here - we even are not sure what the new 1Dx m2 and 7D m2 would be look like... we are pretty confident they will include dual pix though....

Which reminds me that we even didn't know about the DUAL PIX just before the release of 70D - a few months before that we had some rumor about new focus tech... And you, as well as the others are quite aware that Dual PIX AF didn't emerge like that in the last 6 months before the 70D, now did it? ;-)

UPDATE: Since Jrista is as curious as I am and I respect that and like these people - here is the latest on QIS

 

[Rienzphotoz]

NEURO, Could you be so kind to provide some links for those statistics about the 14 million DSLRs sold in 2013. I can put them in good use for personal doings

Thank you in advance.

CIPA aggregates the data, there are many years' worth here:

http://www.cipa.jp/stats/dc_e.html

Enjoy!

Interesting stats ... thanks for sharing Neuro. Looking at those numbers, I see there were over 3.3 million non-reflex/mirrorless camers sold in 2013, now that's over 20% of the DSLR market share and growing ... but it is baffling that Canon does not seem to be interested in the Lion's share of that market :-\ ... surely, with their experience as a successful camera manufacturer, Canon CAN make a great mirrorless camera ... but for some reason they seem to have deliberately crippled the EOS-M. I see a company like Sony come up with a compelling camera like an a6000 (granted that Sony is desperate to try new ideas for their survival, nevertheless, they have produced some AWESOME mirrorless cameras), so why is Canon not interested. With the shrinking P&S camera market and a growing mirrorless camera market, there is clearly money out there to be made ... yet Canon ain't interested? ... are they deliberately sabotaging the market to give mirrorless a bad name? :-\ :-\

 

[Sella174]

... but it is baffling that Canon does not seem to be interested in the Lion's share of that market :-\ ... surely, with their experience as a successful camera manufacturer, Canon CAN make a great mirrorless camera ... but for some reason they seem to have deliberately crippled the EOS-M. ... yet Canon ain't interested? ... are they deliberately sabotaging the market to give mirrorless a bad name? :-\ :-\

Maybe Canon doesn't own certain key patents regarding mirrorless cameras and thus don't want to be walloped into bankruptcy by a no-account company like Olympus. ;D

 

[tron]

... but it is baffling that Canon does not seem to be interested in the Lion's share of that market :-\ ... surely, with their experience as a successful camera manufacturer, Canon CAN make a great mirrorless camera ... but for some reason they seem to have deliberately crippled the EOS-M. ... yet Canon ain't interested? ... are they deliberately sabotaging the market to give mirrorless a bad name? :-\ :-\

Maybe Canon doesn't own certain key patents regarding mirrorless cameras and thus don't want to be walloped into bankruptcy by a no-account company like Olympus. ;D

I wonder why you mention sabotaging. As if it isn't their right not to put resources to mirrorless...

 

[jrista]

;D ;D ;D ;D ;D

You remind me of this:

That's not what I'm saying at all. The 640k deal was a choice, not due to physical limitations. Building a system that can scan incoming light at subwavelength and transfer information at 1000 to 100000 times faster than the fastest we've ever been able to achieve boils down to physics. MAYBE we can do it. MAYBE, if we produce the necessary technology by 2015 (which is when Fossum, based on that latest paper you linked, which I've already read BTW). Sorry, but I truly do not believe that even by 2017 or 2020, we will have the technology to transfer data at 100tbit/s. We won't even be close.

Your knowledge is adorable... And yet quite interesting that you DO continue to fight with the idea that he may have did it.

He hasn't done it. It's a theory. It's a concept. It isn't an actual prototype. If it was, I GUARANTEE YOU it would make waves. It would be on every sensor-related news site and probably every technology site everywhere. Fossum wouldn't keep it under wraps. Not a chance. (You clearly don't read ImageSensorsWorld...this kind of technology, if it reaches prototype stage, will be HUGE.)

http://m.eet.com/media/1081272/SARGENT1433_PG_46.gif
If you stop just for a second using your knowledge from here whereby this is an improvement of the CMOS:

I'm not sure what that has to do with anything. It is simply an alternative approach to making photosensitive electronics. Your readout logic is still built with standard silicon at standard sizes, transferring information at standard speeds. And the sensitivity improvement, from what I know about it, doesn't allow photon counting. There have been other improvements that increase the light gathering capacity of silicon without resorting to wet tech, such as black silicon. Even black silicon isn't going to solve the data transfer rate issue, or allow photon counting, though.

http://m.eet.com/media/1081272/SARGENT1433_PG_46.gif
And the regular physics knowledge... Yeah I had that moment with the jots and the wavelength just as you did. But I do know that he is quite longer in this business than you and me together ;-)

DO you believe that he will reveal every detail of his study to the world so someone could steal it from him? ;-) And do you believe that he would continue for 10 years to research in that field within a reputable university and some sponsor (it could be even Samsung)?

Fossum is a researcher. He produces patents. It's what he does. So absolutely. I do believe he will reveal every detail of his research. I believe he has already revealed everything he knows. Besides, due to things like prior art, no one can really steal it from him. It's his work. He has the prior art. Even if someone tried to patent it, he could prevent it in court. He has about a decade of research and documentation to clearly prove the concept is his, and therefor not the unique invention of someone else. So yes, he absolutely would reveal all the details. He reveals details all the time, and again, if you read ImageSensorsWorld, you would know this.

Fossum doesn't HAVE the specific details for things like high speed tbit/s data transfer. NO ONE DOES. There are undoubtedly people researching it. People have been researching 500gbit and tbit data transfer rates since the 80s. I remember a Byte Magazine article from the late 80s that talked about organic memory and hundreds of gigabits per second data transfer rates. Well, were some 25 years on, and it still hasn't happened. The organic memory concept died, it just wasn't viable, and SSD offered realistic, tangible gains in performance without being unrealistically hopeful.

http://m.eet.com/media/1081272/SARGENT1433_PG_46.gif
IF scientist were so sure like you and so negative... we would be in the medieval age, no offence.

I'm not negative. I'm realistic. You can be as hopeful as you want, but it doesn't mean your hopefulness means success. Your hopefulness is simply unrealistic given how far technology has come, and how close the walls of physics are. This isn't the 90's. Back then we couldn't even see those walls. It's now the 2010's. Two decades on, at the relentless rate we've been pushing technological advancement, those walls are right in front of us now. And not just in the case of QIS...technological advancement via traditional means (i.e. primarily via reduction in size) is going to come to a crushing halt relatively soon. Certain problems have already forced some radical changes to how we manufacture CPUs, for example, and all that's done is stave off the inevitable for a little while longer.

You also cant' forget, Fossum has been trying for a decade to design this type of sensor. A DECADE. That is a very long time to even prove a concept can work. A LOT of CIS patents files in the 80's were viable, and we knew we could eventually shrink die sizes and increase data transfer speeds to levels where we could eventually achieve them. Many of the new technologies being implemented today were actually discovered decades ago. However back then, transistor sizes were hundreds of microns in size, and data transfer rates were so low we had absolutely no question we could improve them.

Today, we've been riding the limits of Moores Law on a continuous basis. The effort involved in developing new advances costs and order of magnitude more money each time we develop a new fabrication process (i.e. it used to cost a few hundred million to build a CMOS fab, today it costs tens of billions.) Transistor sizes are approaching physical limits...the next die shrink is 14nm, and the one after that is 7nm. Gate sizes, even with 3D/finFET, are now only a couple ATOMS across. Even with stretching, that poses a real problem for current flow, hence finFET, and that is only a stopgap measure (and it imposes it's own limitations as well.) The hard, impassable physical WALL is looming very close. There are a couple generations left before there is no such thing as a die shrink anymore. We have a decade, two at most (assuming two to four years between die shrinks) before efforts begin in earnest to develop full multi-layered CPUs and the like, because that will be the only remaining option.

You call me negative. I'm just a realist. There are significant physical limitations that computer technology is already riding close to. If Fossum said he would need 100gbit/s transfer rate, I'd say "When does the technology hit?!?" Why? Because 100gbit/s is only 10x (or less) faster than the fastest transfer speeds we already have today. It's realistic, it's doable, there is already research that indicates it's possible, and it could be ready by late 2015/early 2016. Fossum said he needs 100tbit/s transfer rate, and his timetable showed 2015-2016 as his target date for QIS. Sorry, but you don't suddenly go from 10gbit/s (the fastest ethernet, and also the transfer speed of Thunderbolt) to 100tbit (100000gbit/s) overnight. It just aint gonna happen. Maybe a decade or so from now. But not by 2016. It simply isn't realistic.

There is a pdf (if in those presentations not included back dated from 2010, if I recall correctly where Fossum represents that they are to try to implement that technology (of course quite away from Q.E. of 100%) in 3 stages....
The first one on a regular CMOS. The last one on a new superconducting material... so there you go...

I believe he said the same thing in 2005. In 2010, the new superconducting material was actually just announced as something called a superinsulator. Superinsulators had been hypothesized for years, and we knew they had to exist if superconductors existed. We just didn't know how they worked. Ironically, they really don't work all that differently than superconductive material...just in the opposite (instead of encouraging cooper pairs to attract, superinsulators cause them to repel). The other sensor I've been talking about, the Titanium Nitride sensor, uses both superinsulating and superconducting properties. TiN IS the new superconducting/superinsulating material. However for those properties to exhibit, the sensor has to be cooled to absolute zero.

Again, not being negative. Being realistic. We won't have DSLR-sized cameras with supercooled sensors by 2016. Not a chance. The power and material requirements necessary to cool anything to absolute zero are immense, not to mention rather unique.

As for the CANON.. Officially 200mm is what I have head as well about CANON... but you have to admit that if you were CANON you wouldn't reveal of you are already on a 300 or 450mm wafer, now would you?

Of course they would! You really don't understand either the technologies involved here, nor the economics. Canon moving to 300mm wafers for their FF and APS-C fabrication would be a huge boon to their stock price. OF COURSE THEY WOULD OFFICIALLY ANNOUNCE IT! It's ludicrous to think otherwise, and exceptionally naive.

A proof that is this very topic here - we even are not sure what the new 1Dx m2 and 7D m2 would be look like... we are pretty confident they will include dual pix though....

Canon already has 300mm fabs for their small form factor sensors, which are built on a 180nm process rather than a 500nm process. Canon has had those fabs for years. There has been speculation that the 70D DPAF pixels would need a smaller process in order to be produced. The pixel size shrink on the 70D, however, is not actually that great. At first I thought the shrink was more significant, however the size of the 70D sensor also grew. Prior sensors were around 22.2x14.8mm in size. The 70D is 22.5x15.1mm. The increase in size means instead of a 3.9µm pixel, they actually have 4.1µm pixels. The pixels are only 0.2µm smaller than the 7D. There isn't any need for Canon to reduce their process size to split those pixel in two...they are still more than large enough. Even at 3.9µm, they would still be large enough.

As much as I personally hoped Canon would move the 70D to a smaller process, there just isn't any evidence to support that theory. Physically, Canon still has space to use a 500nm process, even with dual photodiodes. When you think about it, doping the photodiodes really is not that big of a deal, as the photodiodes themselves are a couple thousand nanometers in size, which is about four times larger than the smallest 500nm etching possible with a 500nm process.

If Canon had moved FF and APS-C manufacturing to a 300mm wafer fab, they would have announced it. It would be a massive move, and a move for the better, for Canon as a company, for their shareholders, for their customers. A move to 300mm means more FF sensors manufactured faster with less waste, reducing cost, allowing more electronics on-die at a smaller transistor size, etc. It would be big news, for everyone. No way in hell would Canon hide that fact.

Which reminds me that we even didn't know about the DUAL PIX just before the release of 70D - a few months before that we had some rumor about new focus tech... And you, as well as the others are quite aware that Dual PIX AF didn't emerge like that in the last 6 months before the 70D, now did it? ;-)

The technology had to be in development for more than 6 months before the 70D hit the streets. Canon has patents on the technology. If someone was digging, they would have found them (quite possibly LONG before the 70D hit the streets, as patents have to be requested and then filed quite some time before they are granted. You don't know about the request, but once they are filed, it's all public knowledge...you can find it if you want to. I used to go digging through CIS patents...I don't have enough time to do that any more, but I don't doubt that the patents were out there before the 70D hit the streets.)

 

[Loswr]

The technology had to be in development for more than 6 months before the 70D hit the streets. Canon has patents on the technology. If someone was digging, they would have found them (quite possibly LONG before the 70D hit the streets, as patents have to be requested and then filed quite some time before they are granted. You don't know about the request, but once they are filed, it's all public knowledge...you can find it if you want to. I used to go digging through CIS patents...I don't have enough time to do that any more, but I don't doubt that the patents were out there before the 70D hit the streets.)

In the US, it's 18 months between filing a patent and the publication of that patent. So, 18 months before the patent issued, it was filed. Most of the research had to be completed before the patent was filed, so that goves you an idea of the lead time (there are provisional patents, too, which give you an extra year to fully develop the invention, but I don't know that Canon uses that mechanism all that much).

 

[jrista]

The technology had to be in development for more than 6 months before the 70D hit the streets. Canon has patents on the technology. If someone was digging, they would have found them (quite possibly LONG before the 70D hit the streets, as patents have to be requested and then filed quite some time before they are granted. You don't know about the request, but once they are filed, it's all public knowledge...you can find it if you want to. I used to go digging through CIS patents...I don't have enough time to do that any more, but I don't doubt that the patents were out there before the 70D hit the streets.)

In the US, it's 18 months between filing a patent and the publication of that patent. So, 18 months before the patent issued, it was filed. Most of the research had to be completed before the patent was filed, so that goves you an idea of the lead time (there are provisional patents, too, which give you an extra year to fully develop the invention, but I don't know that Canon uses that mechanism all that much).

The average time for a patent to move the entire way through the process is ~24 months. The 18 month period is from the time they actually start the process to when it's published. There can be some "queue time" as well, and on average, that time seems to be 6 months (although, a highly innovative company like Canon may get more special treatment, honestly can't say there). Canon had to have DPAF technology years ago. People just don't understand the effort and time it takes to bring research to a consumer product. It doesn't happen overnight. It takes years, many years, and a lot of effort.

People also don't seem to understand that technologies like QIS are heavily dependent upon other technological innovations occurring elsewhere in the industry to support some of their needs. I've never gathered from his papers that Fossum is a genius in data transfer concepts. He knows sensor design like the back of his hand, and he is one of the most innovative forces in the image sensor world, but a lot of his technology builds or relies on other technology. Before QIS can happen, we need to know how to transfer data at 100 terrabits per second, and do that in such a way as to not melt the sensor or the data channel or the DSP. That is a LOT of information to move around per second. You need massive processing, orders of magnitude more processing power than current DSLRs have. You need significant cooling as well, even if were not talking about a superconductive device.

To think that a QIS sensor is "in the bag", as Diko's hopes indicate, is simply naive of the realities of some of the assumptions Fossum is making.

 

[Diko]

I might have been wrong about the wafer size change Although according to one guy README the cost reduction is more a of a fuzz.
When I mentioned about the DPAF patent I was quite clear with patenting filling time... I mentioned it because for example now we can't be sure that the next 1dx won't be with foveon quatro remember that Canon has a patent? Fossum and his team would need 20 years perhaps.....

 

[Rienzphotoz]

... but it is baffling that Canon does not seem to be interested in the Lion's share of that market :-\ ... surely, with their experience as a successful camera manufacturer, Canon CAN make a great mirrorless camera ... but for some reason they seem to have deliberately crippled the EOS-M. ... yet Canon ain't interested? ... are they deliberately sabotaging the market to give mirrorless a bad name? :-\ :-\

Maybe Canon doesn't own certain key patents regarding mirrorless cameras and thus don't want to be walloped into bankruptcy by a no-account company like Olympus. ;D

I wonder why you mention sabotaging. As if it isn't their right not to put resources to mirrorless...

I do not understand what you mean by: "As if it isn't their right not to put resources to mirrorless" ... but let me explain why I used the word "sabotaging":
Over the years I've used several third party lenses (Sigma, Tamron, Tokina, Rokinon/Samyang) on Canon cameras (400D, 450D, 500D, 60D, 7D & 5D MKII) and Nikon cameras (D70, D3100, D5100, D7000, D7100 & D6100), when I first got the third party lenses, everything would work well on my Canon/Nikon cameras ... but mysteriously whenever I updated to a new firmware(s), the third party lenses would suddenly not AF as well as they used to before, or the OS/VC would become noisy or the camera batteries would drain faster (when using third party lenses) ... so it is my assumption that Canon/Nikon do some "tinkering" (sabotaging) to "encourage" Canon/Nikon camera users to buy only their lenses (I suppose I'd do the same thing if I was in their shoes ... and why not). If you notice, both Canon/Nikon got into the mirrorless business with disappointing mirrorless camera models ... since Canon/Nikon account for a major portion of camera sales across the globe, they are in a unique position to give a product a good/bad name with subtle tactics. The EOS-M camera was released in June 2012 (that is nearly 21 months ago), yet there are only 2 native lenses available in most parts of the world ... the 11-22mm UWA lens is only available in just a handful of countries ... does one honestly believe that people will not want to buy that awesome small lens, which has received very positive reviews? ... something does not add up here when the ONLY 2 major players (Canon/Nikon) produce halfhearted cameras ... I cannot believe that Canon/Nikon, with their massive resources and R&D, are not capable of producing compelling mirrorless cameras ... for me the only logical conclusion is that they are not interested in the mirrorless business (as they've got far too much invested in their highly profitable DSLR business) and they want it to stay that way ... and the best way of giving mirrorless cameras a bad name is produce halfhearted ones that do not inspire confidence in the general public ... thus "encouraging" people to buy DSLRs ... therefore, the word "sabotaging". Everything is fair in love and war ... and business is war.

 

[Don Haines]

... but it is baffling that Canon does not seem to be interested in the Lion's share of that market :-\ ... surely, with their experience as a successful camera manufacturer, Canon CAN make a great mirrorless camera ... but for some reason they seem to have deliberately crippled the EOS-M. ... yet Canon ain't interested? ... are they deliberately sabotaging the market to give mirrorless a bad name? :-\ :-\

Maybe Canon doesn't own certain key patents regarding mirrorless cameras and thus don't want to be walloped into bankruptcy by a no-account company like Olympus. ;D

I wonder why you mention sabotaging. As if it isn't their right not to put resources to mirrorless...

I do not understand what you mean by: "As if it isn't their right not to put resources to mirrorless" ... but let me explain why I used the word "sabotaging":
Over the years I've used several third party lenses (Sigma, Tamron, Tokina, Rokinon/Samyang) on Canon cameras (400D, 450D, 500D, 60D, 7D & 5D MKII) and Nikon cameras (D70, D3100, D5100, D7000, D7100 & D6100), when I first got the third party lenses, everything would work well on my Canon/Nikon cameras ... but mysteriously whenever I updated to a new firmware(s), the third party lenses would suddenly not AF as well as they used to before, or the OS/VC would become noisy or the camera batteries would drain faster (when using third party lenses) ... so it is my assumption that Canon/Nikon do some "tinkering" (sabotaging) to "encourage" Canon/Nikon camera users to buy only their lenses (I suppose I'd do the same thing if I was in their shoes ... and why not). If you notice, both Canon/Nikon got into the mirrorless business with disappointing mirrorless camera models ... since Canon/Nikon account for a major portion of camera sales across the globe, they are in a unique position to give a product a good/bad name with subtle tactics. The EOS-M camera was released in June 2012 (that is nearly 21 months ago), yet there are only 2 native lenses available in most parts of the world ... the 11-22mm UWA lens is only available in just a handful of countries ... does one honestly believe that people will not want to buy that awesome small lens, which has received very positive reviews? ... something does not add up here when the ONLY 2 major players (Canon/Nikon) produce halfhearted cameras ... I cannot believe that Canon/Nikon, with their massive resources and R&D, are not capable of producing compelling mirrorless cameras ... for me the only logical conclusion is that they are not interested in the mirrorless business (as they've got far too much invested in their highly profitable DSLR business) and they want it to stay that way ... and the best way of giving mirrorless cameras a bad name is produce halfhearted ones that do not inspire confidence in the general public ... thus "encouraging" people to buy DSLRs ... therefore, the word "sabotaging". Everything is fair in love and war ... and business is war.

I wonder if they are looking further down the road than we give them credit for, and are positioning themselves for the day the DSLRs go mirrorless and that they plan to use the same EF mount and the same EF lenses.....

 

[Rienzphotoz]

... but it is baffling that Canon does not seem to be interested in the Lion's share of that market :-\ ... surely, with their experience as a successful camera manufacturer, Canon CAN make a great mirrorless camera ... but for some reason they seem to have deliberately crippled the EOS-M. ... yet Canon ain't interested? ... are they deliberately sabotaging the market to give mirrorless a bad name? :-\ :-\

Maybe Canon doesn't own certain key patents regarding mirrorless cameras and thus don't want to be walloped into bankruptcy by a no-account company like Olympus. ;D

I wonder why you mention sabotaging. As if it isn't their right not to put resources to mirrorless...

I do not understand what you mean by: "As if it isn't their right not to put resources to mirrorless" ... but let me explain why I used the word "sabotaging":
Over the years I've used several third party lenses (Sigma, Tamron, Tokina, Rokinon/Samyang) on Canon cameras (400D, 450D, 500D, 60D, 7D & 5D MKII) and Nikon cameras (D70, D3100, D5100, D7000, D7100 & D6100), when I first got the third party lenses, everything would work well on my Canon/Nikon cameras ... but mysteriously whenever I updated to a new firmware(s), the third party lenses would suddenly not AF as well as they used to before, or the OS/VC would become noisy or the camera batteries would drain faster (when using third party lenses) ... so it is my assumption that Canon/Nikon do some "tinkering" (sabotaging) to "encourage" Canon/Nikon camera users to buy only their lenses (I suppose I'd do the same thing if I was in their shoes ... and why not). If you notice, both Canon/Nikon got into the mirrorless business with disappointing mirrorless camera models ... since Canon/Nikon account for a major portion of camera sales across the globe, they are in a unique position to give a product a good/bad name with subtle tactics. The EOS-M camera was released in June 2012 (that is nearly 21 months ago), yet there are only 2 native lenses available in most parts of the world ... the 11-22mm UWA lens is only available in just a handful of countries ... does one honestly believe that people will not want to buy that awesome small lens, which has received very positive reviews? ... something does not add up here when the ONLY 2 major players (Canon/Nikon) produce halfhearted cameras ... I cannot believe that Canon/Nikon, with their massive resources and R&D, are not capable of producing compelling mirrorless cameras ... for me the only logical conclusion is that they are not interested in the mirrorless business (as they've got far too much invested in their highly profitable DSLR business) and they want it to stay that way ... and the best way of giving mirrorless cameras a bad name is produce halfhearted ones that do not inspire confidence in the general public ... thus "encouraging" people to buy DSLRs ... therefore, the word "sabotaging". Everything is fair in love and war ... and business is war.

I wonder if they are looking further down the road than we give them credit for, and are positioning themselves for the day the DSLRs go mirrorless and that they plan to use the same EF mount and the same EF lenses.....

Quite possible ... obviously, it isn't an accident that Canon holds the numero uno position, they seem to be very clever about their business decisions. Personally, I prefer Canon cameras (with the exception of mirrorless) over any other brand ... and it'd be awesome if we can use EF lenses on Canon mirrorless cameras.

 

[Sella174]

I wonder if they are looking further down the road than we give them credit for, and are positioning themselves for the day the DSLRs go mirrorless and that they plan to use the same EF mount and the same EF lenses.....

This makes being on the receiving end of all the name-calling, put-downs and personal attacks worth it, when others start to say what I've been saying all along.

 

[Don Haines]

I wonder if they are looking further down the road than we give them credit for, and are positioning themselves for the day the DSLRs go mirrorless and that they plan to use the same EF mount and the same EF lenses.....

This makes being on the receiving end of all the name-calling, put-downs and personal attacks worth it, when others start to say what I've been saying all along.

I have believed for several years that the future is mirror less, but I have not believed for a second that the form factor would be substantially different than the current DSLRs. The big reasons for my belief are:
1st. Ergonomics....the 35mm size is the best balance.
2nd.. Lenses. They exist and continue to be improved.
3rd.. Things like dual pixel autofocus.
4th.. Improvements in electronics

Mirrorless is a highly emotional subject to some people, almost as bad as the great film/digital debate. Right now, most cameras are mirrorless, and the trend is shifting more that way. The last two big obstacles are focus systems and viewfinders. Both are getting close, and both have the potential to offer new features and improve on the old.

Imagine, no more AFMA, no more moving parts to wear out...

The EOS-M is not the future of mirrorless, it will be the 1D?, the 5D?, the 7D?, and a host of rebels.

It is possible that the delays on the 7D2 are problems getting this right.....we might be in for a real surprise when it gets released.

Just ask yourself, where are all those people who said digital will never replace film? This is the same debate, just a different component. It will happen, it's just a matter of time...

 

[Max ☢]

+1 Don ! I wouldn't have said it better.
Mirrorless as an image acquisition technology, not as a form factor, is definitely the future evolution of our current DSLR. The major roadblocks at the moment are the EVF quality and the AF, but I am confident that improvements of the dual pixel, the OLED and the overall video technologies will clear the way.

Another reason why the evolution towards mirrorless is bound to happen is the associated decrease in manufacturing and quality control costs that will result from the simpler overall camera structure (i.e. less components). This will definitely improve the profit margins for the manufacturers, an argument which is always music to the hears of big corporate managements!
In this end, it is this factor which may drive the transition from DSLR to "DSLM" rather than customer experience or any other marketting argument (although phasing out AFMA will be a very nice evolution step for the users).

 

[J.R.]

+1 Don ! I wouldn't have said it better.
Mirrorless as an image acquisition technology, not as a form factor, is definitely the future evolution of our current DSLR. The major roadblocks at the moment are the EVF quality and the AF, but I am confident that improvements of the dual pixel, the OLED and the overall video technologies will clear the way.

Another reason why the evolution towards mirrorless is bound to happen is the associated decrease in manufacturing and quality control costs that will result from the simpler overall camera structure (i.e. less components). This will definitely improve the profit margins for the manufacturers, an argument which is always music to the hears of big corporate managements!
In this end, it is this factor which may drive the transition from DSLR to "DSLM" rather than customer experience or any other marketting argument (although phasing out AFMA will be a very nice evolution step for the users).

+1 ... mirrorless still has some way to go before it replaces the current DSLRs.

Canon has the highest market share and has the most to lose if a product goes wrong, so I'm guessing that Canon is simply playing the waiting game (if it ain't broke, don't fix it) till the mirrorless market matures and it will come out with its own mirrorless which will blow the pretenders out of the water. I also expect that it will be perfectly usable with the EF lenses, so yes, mirrorless for the technological improvements, smaller form factor be damned.

It's a pity that people end up whinging on internet forums wanting their perfect (mirrorless) camera from Canon, NOW! The amount of entitlement shown in these forums is staggering.

 

[Max ☢]

Canon has the highest market share and has the most to lose if a product goes wrong, so I'm guessing that Canon is simply playing the waiting game (if it ain't broke, don't fix it) till the mirrorless market matures and it will come out with its own mirrorless which will blow the pretenders out of the water. I also expect that it will be perfectly usable with the EF lenses, so yes, mirrorless for the technological improvements, smaller form factor be damned.

Indeed - Canon is particularly known on the market to "play it safe" and target just the "good enough" point (i.e. no fancy bells and whistles, which apparently annoys quite some users here) in order to secure a leading market position with the minimum of risk taking. At the same time, I think the company also knows that its primary interest is to secure their revenue streams and maintain or increase their profit margins. If milking some sensor platforms to death is one way to achieve this goal, preparing the future with new camera platforms is also the other strategy, which is confirmed by their significant patent output.

Their habit of not releasing cameras untill the technology within is made bullet-proof explains why Sony and others have "apparently" beaten Canon to the Mirrorless game - Apparently, because I don't think those competitors will eventually win the game by pushing half-finished products to the market (see the list of complaints about de A7/A7r), and prime market innovators are not always those who remain market leaders in the long run (if customer trust is not won overnight, it can be lost very quickly).
I think Canon's strategy in the mirrorless segment is twofold: first, quietly develop the technology in the lab with the goal of reaching a good enough point for the targetted market, and secondly, observe how the competition fares with their attempts on the market. So, the day the mirrorless tech is ready to be implemented in DSLRs, Canon will know exactly how to introduce it in order to extract maximum profit, and the competition will have paid for some of the market insights!

Of course, all this process takes some time to devellop, which make some people quite unhappy, but in the mean time I'm perfectly fine with my 6D which gets the job done

 

[9VIII]

Indeed - Canon is particularly known on the market to "play it safe" and target just the "good enough" point (i.e. no fancy bells and whistles, which apparently annoys quite some users here) in order to secure a leading market position with the minimum of risk taking.

Historically, that's not how Canon became king of the hill.
They abandoned their entire library of lenses to implement the EOS system. They pioneered Image Stabilization. They were one of the first manufacturers to push heavily for full frame. More recently their supertelephoto lenses are some of the best examples of optics on the market, and as controversial as it may be here, lets not forget the video features on the 5D2.

Canon built their empire on cutting edge technology, plus their products are generally solid. I really don't think they would be where they are right now just matching the competition spec for spec.
Ever since the 90's they have been doing new and exciting things on a regular basis, that hasn't changed. What I do think has changed significantly is the way people view the industry, our definition of "regular basis" has changed. People want yearly updates now, if not monthly, whereas before waiting a few years between big announcements was "normal".
Basically people here have psyched themselves into a vortex of expectations, and everyone inside Canon is probably just doing what they have always done.
Come 2020, I'm sure a review of this decade won't be much different from the last.

Except that the last two decades have seen two of the biggest transitions in the entire history of photograhpy (AF and then digital), which may also skew our expectations a little bit.