More pixels? larger sensor (MF)? ... .CANON??!!

[haefnerphoto]

Uprez by 40% caused my file to go beyond 9GB with layers (open in PS) !!!
This is the case for 30-40% of my production. That would fill things up quick. but I can be selective on a need be basis.

I already have close to 10TB and i'm starting to run low....I know I will be getting 4 or 8 (3-4TB in RAID+1) drives very soon, but that would be another issue with much larger files eating space very quickly.

So with Pixel not = to Pixel, (I would think even the 22MB files off the MF vs the D800 hold "better" /At some point resolution does get more information of what is in site?), I may as well stay with what isn't broken, just brief/discuss with clients more clearly as to the way they want the images, this would help in itself!

Phil, I wouldn't suggest uprezzing until you've finished and flattened the file.  I regularly scale up my images for printing and find that Photoshop (CS6) does a great job.  Jim

[BJL]

I have heard that Canon might be working ... even on a larger than FF sensor(Some time back on CanonRumors).

I have also heard such rumors of a supersized Canon system --- about every few months for the last decade or so! But so far none of them has been confirmed, and none has been backed by any persuasive evidence or arguments (hint: the arguments usually offered are no more valid now than back in the film era, and it never happened then.) Of course, a decade ago most of us saw no reason to expect that what was then "Apple Computer Inc." would transform itself into a mobile phone maker that sells a few laptops on the side and even fewer desktops, so I am not claiming that it absolutely, definitely will _not_ happen, just that so far the numerous rumors of Canon or Nikon going MF have been evidence of nothing but misguided wishful thinking, uncontaminated by consideration of economic realities.

[eronald]

I'm sure there are some prototype Canon and probably Nikon systems floating around their labs  - after all Canon has semiconductor fabs available to them so why not run up an insurance project. Just clip existing shift lenses to them, these have huge image circles ...

Edmund

I have also heard such rumors of a supersized Canon system --- about every few months for the last decade or so! But so far none of them has been confirmed, and none has been backed by any persuasive evidence or arguments (hint: the arguments usually offered are no more valid now than back in the film era, and it never happened then.) Of course, a decade ago most of us saw no reason to expect that what was then "Apple Computer Inc." would transform itself into a mobile phone maker that sells a few laptops on the side and even fewer desktops, so I am not claiming that it absolutely, definitely will _not_ happen, just that so far the numerous rumors of Canon or Nikon going MF have been evidence of nothing but misguided wishful thinking, uncontaminated by consideration of economic realities.

[telyt]

Personally, I think that OLP less sensors are snake oil. They offer false resolution, inventing non existing detail. I think that a good, correctly OLP-filtered sensor is preferable, but that is just me.

Given the a7r is selling faster than the a7 you may correct, it's just you    False detail on one hand, lost detail on the other... six of one, half-dozen of the other.  The real question is, how do the prints look?

[ErikKaffehr]

Hi,

Canon has the 7D at 18 MP, that sensor upscaled to 24x36 would yield 46 MP. Canon could probably do it, I don't know why they don't.

At 46 MP it may be possible to drop the AA-filter.

Going to a larger sensor than 24x36 doesn't make a lot of sense for a company like Canon, I guess.

Best regards
Erik

I usually try and help others in such posts, but I need to get my head out of the process to make some sense...

I post this here as the lines are ever more blurred between the two formats(MF/DSLR) with newer larger mpixel sensors in DSLR's, and most interestingly, I have heard that Canon might be working more mpixels or even on a larger than FF sensor(Some time back on CanonRumors).

I ask this question as I find myself in the switching game...

My first camera was a Canon EOS Elan Film about 20 years ago. Sometime back 7-8 years ago when I was hired to shoot for a company they supplied me with Nikon gear. The color was bad, the focus was worse. It may have been a D200 or D70? I cant remember. I used the camera as my own, and they let me do with it as I please. Around the time, I had a Oly C2500L which was like a toy, but still amazing for the time to see such a camera. This was marketed to the medical field as it can focus on objects just about touching the lens bezel.

So then at some point I got a Canon 10D, and was a happy camper for most of it in overall use. At this time there was lots of tech activity and improvements happening fast. A few models later, including the Kodak SLR/c, I got the 1Ds tank. Amazing camera to this day(purchase price $7500, Sold $700 a month ago).  Now besides the MF Phase back, I have the Canon 5DmarkII.

As I shoot both street/people/event, etc all fine with the 5Dmk2, I also shoot product/Studio with MF.

The Question I have is that I am in need of more resolution from 22. Not a lot more, but about 20+% would help in my studio work from the P25 I use.
I can't use the Canon as it too is 21mp and the A-filter is not acceptable.

So I find myself waiting for Canon to come up with something that I can use as my Studio camera and maybe as daily use camera(hoping Canon gets it right and offers a no A-filter option., and put the 5D2 as 2nd-Cam. I can't justify even a $7500 P45+ as I don't do a lot of studio work (for now). Plus now that I sold my 1Ds, I now need a 2nd-Cam to my 5Dmk2. So its either $3K for a Mark3(much better focusing), AND the Nikon for studio for $3k, or $7500 for the studio MF, and still needing a 2nd-Cam.

The D800e so far looks like it delivers. Do I wait for a Canon to answer with more mpixels? Or do I switch to Nikon just for Studio shooting to replacing Sinar/SchneiderHR( getting the DSLR adapter for the Sinar),

There it is... Another what to do? :-)

[Phil Indeblanc]

Looks like I'm buying the A7R regardless of my setup. I don't think it will change my setup, but who knows!?

Phil, I wouldn't suggest uprezzing until you've finished and flattened the file.  I regularly scale up my images for printing and find that Photoshop (CS6) does a great job.  Jim

Thanks Jim, makes sense!
I approached it thinking if an unedited file gets the uprez it will be a "better" final. Although it maybe, but likely not worth the difference. I just have to go over the final with a fine tooth comb.

[hjulenissen]

Hi,

Canon has the 7D at 18 MP, that sensor upscaled to 24x36 would yield 46 MP. Canon could probably do it, I don't know why they don't.

At 46 MP it may be possible to drop the AA-filter.

Going to a larger sensor than 24x36 doesn't make a lot of sense for a company like Canon, I guess.

Best regards
Erik

Your assumption seems to be that if you can manufacture a sensel of dimensions x in a small sensor, then you can to the same for a larger sensor. I doubt that this is strictly true.
1) The manufacture process may be entirely different for larger sensors (stitching?), thus prohibiting simple re-use of tech from smaller sensor production lines
2) Even if possible, the defect rate might make this prohibitively expensive
3) If the sensor is technically/economically possible, the supporting electronics to transport and capture the information out of the sensor might not be (at sensible frame-rates, sensible noise levels, sensible battery-life vs camera size etc)

I was disappointed with the FF sensor of the 5Dmk3, and I hope that Canon will push the state-of-the-art in the next generation.

-h

[ErikKaffehr]

Hi,

Yes, larger sensors are stitched. On the other hand Canon makes 24x36 size sensors already.

On the other hand, I just read a "sony rumor" that they have developed 54 MP full frame chip. It is supposed to be very expensive in manufacture. This may indicate that upscaling an APS-C is much more expensive than I would believe. (The 24 MP APS-C chip upscaled to 24x36 would be 54MP).

Best regards
Erik

Your assumption seems to be that if you can manufacture a sensel of dimensions x in a small sensor, then you can to the same for a larger sensor. I doubt that this is strictly true.
1) The manufacture process may be entirely different for larger sensors (stitching?), thus prohibiting simple re-use of tech from smaller sensor production lines
2) Even if possible, the defect rate might make this prohibitively expensive
3) If the sensor is technically/economically possible, the supporting electronics to transport and capture the information out of the sensor might not be (at sensible frame-rates, sensible noise levels, sensible battery-life vs camera size etc)

I was disappointed with the FF sensor of the 5Dmk3, and I hope that Canon will push the state-of-the-art in the next generation.

-h

[BJL]

Yes, larger sensors are stitched. On the other hand Canon makes 24x36 size sensors already.

Erik, I think you are missing the point: that stitching requires very precise alignment after moving to a different part of the wafer, so that the imprecisions in that could set a stricter lower limit on feature sizes in stitched sensors than "single shot" ones. Nikon and Sony also have. Distinctly larger pixel size in any of their stitched sensors than in some of their unstitched ones: their 24MP in 24x16mm would scale to 54MP.

But I can see numerous other possible reasons why Canon might have decided not to pursue higher resolution in 36X24mm format, relating to economic considerations more than technical limitations.

It could be that Canon sees a market for a higher resolution 36x24mm sensor needing a clear improvement in sharpness and detail of very large prints given by the combination of Canon sensor with Canon lenses, not just oversampling to eliminate the OLPF) and not enough of its lenses would do well enough under the pixel peeping scrutiny that many potential customers would apply. "Works great with Zeiss lenses" is not the slogan that Canon marketing wants.

It could also be that Canon expects that customers for such a camera to make close comparisons of IQ aspects like dynamic range at low ISO, where current Canon technology would leave it losing badly to the Nikon (or Sony) alternatives. If so, it could make sense for Canon to instead keep to its strengths, which in the professional area are more in the direction of fast action, low light, high ISO speed performance.

In short, just because a company could make a product, and some people would buy it, does not mean that it makes business sense to invest in developing such a product.

See also: "why Canon and Nikon could have gone into medium format cameras at any time over the last half century (even far smaller Pentax did!) but they never have."

[eronald]

Canon can do an unstitched 35mm sensor, at one point they built a stepper which could do this. I don't know whether there are any physical limitations that limit die size on a purpose-built stepper, as I would naively assume that there is a movable stage and the movements of this stage determine the max die size, where accurate stage positioning over the whole excursion would of course be a requirement. In any case as Canon seem to be an optics company first and foremost, it is not crazy to think that they could create an end-to-end process for larger sensors if they really wanted/needed to, and they might find some other uses for largeish single-step dies.

Edmund

[Paul2660]

The mention of larger sensors and stitching caught my attention.   I used to think the larger Dalsa's were stitched due to the obvious lines that you can see when you look at the sensor, however from talking to several different sources, I have been told this is not stitching.  Instead the lines are readout's for the chip so that the the data can be processed faster.  These lines are not from individual sensors butting up together.  The 60 and 80MP chips are still made from one large piece of wafer.  

No Canon, Nikon or Sony chip I have viewed shows these type of lines, only the Dalsa chips used by Phase One, Leaf etc.

The fact that the digital read out's (lines) can cause center folding seems to lead many to the idea that the center folding lines represent the edges of the sensors.  However the lines are caused by the possible differences in calibration of the various read outs causing one sector of the chip to show a bit darker.  This will show up as a hard line on an image, and will be more pronounced with heavy contrast or if the image is converted to B&W.

Phase One calibrates each back to remove centerfolding, however some backs can still show it especially if tech camera lenses are in use and shifted. (Schneider's are much worse at this than Rodenstocks).    The problem can still show up with shots taken with Mamiya/Phase One lenses and if it does, then most times the back needs to be re-calibrated.

As far as I know Phase One's 80MP is the largest sensor in use today (for traditional photographic work) however there may be larger sensors in use by other applications that may actually be combinations of smaller physical sensors.  

The older Kodak chips, that Phase One used in the P45+, other backs in that line, did not have the readout lines and thus many people who used those backs and upgraded to Dalsa saw the lines and assumed that the Dalsa chips were stitched (me included).  

Paul Caldwell

[BJL]

Canon can do an unstitched 35mm sensor, at one point they built a stepper which could do this.

Ronald, we seem to discuss this several times a year! Yes, Canon used to make one stepper with a maximum field size of about 50x50mm, introduced over ten years ago, and now discontinued. But it had a rather large minimum feature size, was advertised for tasks like the first rough stages of fabrication, and when Canon launched its 36x24mm sensors, after introducing that stepper, Canon said in several white papers that it needed to use stitching to make those 36x24mm sensors, and that the "APS-H" sensors of several 1D models were roughly at the size limit for fabrication without stitching.

That stepper might well be the one used to make the largest Kodak CCDs, of about that 50x50mm size, but those have huge 24micron pixels, good for 1.5MP in 36x24mm format.

By the way, Canon still has one stepper than can make chips far larger than 36x24mm without stitching, but its minimum feature size is far larger than the discontinued one, 1.5microns, so useless for anything under about 20 micron CMOS pixels: http://www.usa.canon.com/cusa/semiconductor/products/semiconductor_equipment/steppers/fpa_5510iv_stepper#Specifications

[eronald]

BJL,

  Thanks for jogging my recollection; the question really is not whether they have a stepper for bigger chips, it is whether they can make one for current process technology if they wish to ..
  When the 1Ds was launched, there was little demand for for 24x36 + pads chips. But now "full-frame" has become a mainstream (digital) photography consumer product, and millions of these chips need to be made each year, probably justifying dedicated fabs and production equipment.
  Stated in these terms the question becomes exactly how much a modified stepper -or some other adaptation to the current production process that would allow precisely manufactured larger chips- would cost to be put in production and what practical use it would have aside from camera chips. As you know, Canon found a zillion of uses for its chip technology, including one very strange but profitable application, turning a chip on its side which makes it into a ... printer head, aka bubblejet.
  Of course, one might ask the question differently, wondering by how much Canon is handicapped in the 35mm full frame market by not being able to scale their processes smoothly and reuse the cells of other cameras, due to the absence (if so?) of a stepper.
 BTW, I would be really interested in circuit-level documentation of some current designs if you have an accessible reference, or a PDF

Edmund

Ronald, we seem to discuss this several times a year! Yes, Canon used to make one stepper with a maximum field size of about 50x50mm, introduced over ten years ago, and now discontinued. But it had a rather large minimum feature size, was advertised for tasks like the first rough stages of fabrication, and when Canon launched its 36x24mm sensors, after introducing that stepper, Canon said in several white papers that it needed to use stitching to make those 36x24mm sensors, and that the "APS-H" sensors of several 1D models were roughly at the size limit for fabrication without stitching.

That stepper might well be the one used to make the largest Kodak CCDs, of about that 50x50mm size, but those have huge 24micron pixels, good for 1.5MP in 36x24mm format.

By the way, Canon still has one stepper than can make chips far larger than 36x24mm without stitching, but its minimum feature size is far larger than the discontinued one, 1.5microns, so useless for anything under about 20 micron CMOS pixels: http://www.usa.canon.com/cusa/semiconductor/products/semiconductor_equipment/steppers/fpa_5510iv_stepper#Specifications

[BernardLanguillier]

I have also heard such rumors of a supersized Canon system --- about every few months for the last decade or so! But so far none of them has been confirmed, and none has been backed by any persuasive evidence or arguments (hint: the arguments usually offered are no more valid now than back in the film era, and it never happened then.)

The whole marketing of Canon has been based for 20 years on leveraging the professional usage of their equipment (big white lenses at sports event) to attract casual shooters generating the bulk of their profit... against a single "enemy"... Nikon.

They are facing a problem now because Nikon has been threatening this approach since the day they released the D3 on 23-Aug-2007. It started with high ISO and has been low ISO image quality since the D3x was released on 1-Dec-2008. My understanding is that Nikon priced the D3x the way they did only to make a statement as a high end offering, they could have sold it cheaper but they wanted to release a body whose pricing would say "better than the 1Ds3".

We are at a turning point where many pros/high end amateurs are starting to really wonder whether Canon has the technology to compete in the high end segment of DSLRs. Whether they can still be seen as leading provider of camera delivery the highest possible image quality. Don't get me wrong, I know the 5DIII is an excellent camera and that Canon still produces some of the best lenses available.

If Nikon does release, as seems possible if not likely, a true 16 bits 54MP D4x early 2014, that will only further increase the gap and generate more boat shifting in the high end segment.

In that context, and assuming that Canon is still struggling to enhance DR of their sensors at the current pixel sizes, going to wider sensors may be they only technological option to remain credible in the high end segment. This is the exact same route that Phaseone selected by maintaining the same pixel size with roughly constant image quality per pixel, but increasing the sensor size to reach higher resolutions up to the P65+/IQ180/IQ280.

So it does in fact seem slightly more likely to me than it used to and the sole purpose would be to protect their image as a high end photography equipment supplier. IE, to win against Nikon. Most reasonable decision makers would IMHO have given up years ago on using inferior in-house sensors in favor of Sony parts, but they have not. That says a lot about the level of pride of some key people within Canon, and this pride may have led them towards the unlikely MF path.

Future will tell. As a photographer, I would love this to happen!

Cheers,
Bernard

[eronald]

Bernard,

I think most of here have access to both cameras from brand C and brand N, and we know both make stars and dogs.

Both companies have issues more serious, in the sense that the photo market is going through a massive change.

Both companies could buy Phase One or Hassy in 5 minutes.

'nuff said.

Edmund

The whole marketing of Canon has been based for 20 years on leveraging the professional usage of their equipment (big white lenses at sports event) to attract casual shooters generating the bulk of their profit... against a single "enemy"... Nikon.

They are facing a problem now because Nikon has been threatening this approach since the day they released the D3 on 23-Aug-2007. It started with high ISO and has been low ISO image quality since the D3x was released on 1-Dec-2008. My understanding is that Nikon priced the D3x the way they did only to make a statement as a high end offering, they could have sold it cheaper but they wanted to release a body whose pricing would say "better than the 1Ds3".

We are at a turning point where many pros/high end amateurs are starting to really wonder whether Canon has the technology to compete in the high end segment of DSLRs. Whether they can still be seen as leading provider of camera delivery the highest possible image quality. Don't get me wrong, I know the 5DIII is an excellent camera and that Canon still produces some of the best lenses available.

If Nikon does release, as seems possible if not likely, a true 16 bits 54MP D4x early 2014, that will only further increase the gap and generate more boat shifting in the high end segment.

In that context, and assuming that Canon is still struggling to enhance DR of their sensors at the current pixel sizes, going to wider sensors may be they only technological option to remain credible in the high end segment. This is the exact same route that Phaseone selected by maintaining the same pixel size with roughly constant image quality per pixel, but increasing the sensor size to reach higher resolutions up to the P65+/IQ180/IQ280.

So it does in fact seem slightly more likely to me than it used to and the sole purpose would be to protect their image as a high end photography equipment supplier. IE, to win against Nikon. Most reasonable decision makers would IMHO have given up years ago on using inferior in-house sensors in favor of Sony parts, but they have not. That says a lot about the level of pride of some key people within Canon, and this pride may have led them towards the unlikely MF path.

Future will tell. As a photographer, I would love this to happen!

Cheers,
Bernard

[Phil Indeblanc]

Bernard, If one can see this, I can see the leading MFdb's being forced to make the next step into larger sensors, so they can keep the gap wide as possible. I don't see why not.
I remember there was a photographer who had a custom made sensor fabricated to his spec that was a very larger sensor chip. I don't remember the size, yet I know the resolution was low (years back 12mp?), but the files he explained were very film like for him.

[ondebanks]

In that context, and assuming that Canon is still struggling to enhance DR of their sensors at the current pixel sizes, going to wider sensors may be they only technological option to remain credible in the high end segment. This is the exact same route that Phaseone selected by maintaining the same pixel size with roughly constant image quality per pixel, but increasing the sensor size to reach higher resolutions up to the P65+/IQ180/IQ280.

But what Phase One did - going from sub-fullframe (sub-645) sensors to [virtually] fullframe - is only like what Canon did in going from the APS-H 1D to the full-frame 1Ds. No biggy.

If Canon were to leap out of 35mm and market something which could be termed medium format, the analogy would be different: it would be like Phase One leaping out of 645 and making a full 6x9 format back. Both prospects are extremely unlikely in the near term.

Ray

[eronald]

Of course, wait 20 years and anybody will be able to buy electronics and turn his view camera into a digital

Edmund

But what Phase One did - going from sub-fullframe (sub-645) sensors to [virtually] fullframe - is only like what Canon did in going from the APS-H 1D to the full-frame 1Ds. No biggy.

If Canon were to leap out of 35mm and market something which could be termed medium format, the analogy would be different: it would be like Phase One leaping out of 645 and making a full 6x9 format back. Both prospects are extremely unlikely in the near term.

Ray

[BernardLanguillier]

Both companies have issues more serious, in the sense that the photo market is going through a massive change.

I believe both companies are in desperate need of customer thinking they are buying a mid range camera from the company making the best cameras available on the market... instead of a smart phone.

So owning the crown of image quality is IMHO extremely important for continued success in the mass market.

There is no doubt that Canon must be looking very actively for a way to get back in the race. The release of the a7R must only be accelerating their feeling of utter urgency... The 2 Canon shooters I am closest with have already bought an a7R with the matching lenses on the first day of availability... They are both just amazed by the difference a modern sensor makes and really glad they can use their excellent Canon glass.

I am not sure that buying MF brands would have much value for Canon in terms of image, it would be another admission of their inability to compete on technology. That wouldn't fix the ego thing.

Cheers,
Bernard

[BJL]

A bunch of little comments to posts above:

- I agree with Ronald that, if Canon or Nikon, or Sony did feel a need to move into MF, a prestigious but very small market, the most practical method would be to purchase or partner with an existing MF maker like Mamiya/Phase One. For one thing, this would make use of existing MF lens designs, and the installed base of such lenses, along with brand loyalties of existing MF users. These discussions tend to focus too much on sensors and not enough on lenses, and the barrier of needing to develop a whole new range of lenses for a larger format needs to be justified by a significant profit potential --- especially when one cannot sell them for Leica-sized markups. Saying "use our tilt-shift lenses" will not be enough!

- I disagree with Bernard's vision of Canon as being in some kind of desperate situation: AFAIK, Canon still lead DSLR sales overall and in the professional sectors like sports/PJ, where "speed" is far more important that getting even more than 12 steps of DR or even more than 22MP or detail. The skew in this forum towards people interested extremes of high resolution, high dynamic range, and mostly low ISO photography are not representative of where most of the camera buying money is!

- Canon is not at a disadvantage relative to any other maker of sensors 36x24mm and larger: they all must use on-wafer stitching (except for low res. sensors for X-rays and such). AMSL, which now dominates the stepper industry, pulling away from Nikon and Canon, puts a lot of emphasis on supporting stitching including 2D stitching for making large chips, and has never gone beyond 33x26mm field size in any of its steppers. This along with Canon and Nikon's lack of any recent efforts towards wider field steppers other than very low res. ones, indicates that the market for such large chips is small enough that stitching is more cost effective than designing and making a stepper of larger maximum field size.