Canon EOS 7D Mark II in 2014 [CR2]

[Don Haines]

DSLR's have a limited lifespan anyway because of the mirrorless segment and if you don't think that mirrorless will someday outperform DSLR's, well then you are not really paying attention to the technology and what it's capable of. The technology is rapidly advancing and because people don't really like radical change, it will come, in small doses and before you know it. BAM! DSLR's are gone and mirrorless becomes the standard.

Just because Nikon and Canon have not given mirrorless a serious thought does not mean that it is useless. Companies like Sony and Panasonic, IF they continue down the road that they have started to build, will soon offer up some very serious camera bodies that Nikon and Canon will have to pay attention to, if they have not already.

I have a 16Mpixel p/s camera that does some amazing things... It can do video at 240hz... It has face recognition and tracking..... I wish my 60D could track birds that well..... I will probably have to wait for mirrorless technology to emerge before DSLR's can do that.....

 

[Marsu42]

It has face recognition and tracking..... I wish my 60D could track birds that well..... I will probably have to wait for mirrorless technology to emerge before DSLR's can do that.....

Nonono, you're getting this the wrong way: Your 60d is not *supposed* to track that well, that's what the more expensive models are for! Canon of course could have added some fw features at zero cost, not to speak that a better af model probably isn't an economic impossibility either for the former 60d price.

Meaning: even with mirrorless each camera segment (p&s, amateur/semipro-crop, enthusiast/pro-ff) will be stratified so that the cheaper models will lack some vital features, that's just the way it works.

 

[dufflover]

Out of the mirror mechanism and sensor pretty sure most of the cost is in the sensor. DSLRs getting replaced doesn't actually change much as the expensive part (artificial price levels or otherwise) of a larger format is the sensor, just the mirror.

 

[jrista]

So, your saying everything becomes a FF camera with the ability to read the FF area, APS-H area, or APS-C area of the sensor? Sure, this could be an option...but you would always be paying for the FF sensor. There are some benefits to that, but I think there is still explicit value in smaller physical sensor formats like APS-H and APS-C. I don't ever foresee the 7D line becoming a FF with selectable readout areas. It will always be an APS-C part.

What I said was a possible trend for DSLR's to go, not a "this is what's going to happen". (geesh : )

DSLR's have a limited lifespan anyway because of the mirrorless segment and if you don't think that mirrorless will someday outperform DSLR's, well then you are not really paying attention to the technology and what it's capable of. The technology is rapidly advancing and because people don't really like radical change, it will come, in small doses and before you know it. BAM! DSLR's are gone and mirrorless becomes the standard.

Just because Nikon and Canon have not given mirrorless a serious thought does not mean that it is useless. Companies like Sony and Panasonic, IF they continue down the road that they have started to build, will soon offer up some very serious camera bodies that Nikon and Canon will have to pay attention to, if they have not already.

Don't think DSLR's are going to be replaced? CRT monitors? Floppy disk? Tape players? How about DVD? oh wait they were ALL replaced by something better and by something that was a radical difference from what people ever thought was possible. Heck, CRT monitors disappeared almost overnight and you would be hard pressed to even fine one new now-a-days.

Don't think Canon and Nikon could ever fail as a company? Well Rome was a world power and fell...

And I never seriously suggested the 7D2 would be FF, I have always believed that the 7D2 was going to stay APS-C. Go back and look at my older posts and you will see.

Don't get all buggered on here because someone suggests a product or performance that YOU may not agree with. This IS Canon RUMORS right, not Canon FACTS???

Lighten up and have some fun...

D

I think your comparison of either Canon or Nikon DSLRs to "CRT Monitors" is a bit naive. CRT's were old and decrepit DECADES before they became obsolete. DSLRs, on the other hand, are currently the pinnacle of digital photography equipment. They have competition, yes, and eventually mirrorless will become dominant, sure. But that does not mean that Canon and Nikon will disappear right along with the DSLR. Both have already forayed into the mirrorless market. Both are fairly careful companies, so they are testing the waters rather than diving in head first with all the funding they've got (with the exception that Nikon seems rather desperate to develop and release as many diverse products as it possibly can just to keep itself viable.)

Where you say Canon will fail because they are currently primarily dependent upon the DSLR for their revenues, I say they will adapt and produce a phenomenal mirrorless part, much like they produced cine parts, in more than enough time to survive the changing market. Just because a company "is currently" dependent upon one particular product for it's revenues in a given market segment does not mean that company is incapable of changing which products they depend upon for revenue.

As facts would have it, 2013 is the FIRST year that DSLR sales actually missed their targets in many years. Up through last year, DSLR sales, particularly the entry-level models but also higher end models, were actually growing. Last year, DSLR growth actually topped mirrorless growth in a number of key markets. One year's market miss does not make a trend. We may see similar things over the next few years, and if we do, then a trend it definitely is...but it isn't a trend yet. In several articles I read last year and the first part of this year, the analysis of the DSLR growth gave an increase in photography amongst consumers in general, thanks to smartphone sales, as a driver for those same consumers becoming more interested and wanting better quality. To date, the DSLR has historically offered the best image quality to price ratio in the digital photography arena. That is certainly primed to change if cameras like the A7r become commonplace, and cheaper models become available.

But again...Canon and Nikon are certainly not incapable of adapting to market change. They have adapted to, and in some cases even lead, market change in the past decades...it isn't like this is the first radical shift the photography market has experienced. Last time, Canon came out on top, and Nikon lost some ground (and seems to have experienced a permanent supply chain/manufacturing capability problem as a result, which I think is more to blame for their struggles than producing intriguing products that people want to buy.) So, I don't disagree that mirrorless technology will eventually become dominant, never have...although I personally despise EVFs and do not think they will ever be able to physically achieve the kind of raw pixel resolution, dynamic range, or frame rate necessary to avoid their shortcomings relative to optical viewfinders, so I'm not a big fan of mirrorless technology. I do not, however, believe either Canon or Nikon, the two biggest players in photography for decades, are going to fail just because the market is enduring another shift. I think both companies are innovative, competitive, and often quite shrewed in their business dealings, and they will survive...probably with some particularly compelling products as well.

 

[dgatwood]

So, your saying everything becomes a FF camera with the ability to read the FF area, APS-H area, or APS-C area of the sensor? Sure, this could be an option...but you would always be paying for the FF sensor. There are some benefits to that, but I think there is still explicit value in smaller physical sensor formats like APS-H and APS-C. I don't ever foresee the 7D line becoming a FF with selectable readout areas. It will always be an APS-C part.

Assuming they can come up with a reasonably reliable mirror box design that allows the use of EF-S lenses (e.g. a flip-down mask on the OVF to show the smaller coverage area and a mirror that slides before it flips so that it doesn't hit the back of the lens), no, there isn't any real value in a smaller physical sensor format over an APS sensor with the same pixel density (pixels per unit of area).

FF sensors will always be more expensive than APS-C sensors as well. The reasons why FF sensor cost has come down (which include the use of 300mm wafers) are the same reasons APS-C will always maintain a cost lead. Reducing defect rates or whatever will result in higher APS-C yield as well. There is always a market for the cheaper version. Doesn't matter how cheap FF gets, people will still buy APS-C (and other, similarly small formats.)

One big reason for the higher cost of FF sensors is economies of scale. You don't necessarily spend the money to get your yield up if you're building a low-volume, high-end product. If you're building a mass-market product, you do, because that extra R&D cost up front pays off over the long term. Make the APS sensor the mass-market choice, and its cost will come way down.

Now there are certainly other factors (like the wasted silicon near the edges of the wafer), but there are ways of dealing with that, too (like printing DIGIC chips and other custom parts in that space).

 

[Marsu42]

Now there are certainly other factors (like the wasted silicon near the edges of the wafer), but there are ways of dealing with that, too (like printing DIGIC chips and other custom parts in that space).

... or decreasing the sensor resolution from 22mp to 20mp like 5d3->6d, I guess next to marketing that's the reason they did this no matter how bad the video moire is because it lets you get more yield from the current production setup.

The other possibility that the 20mp sensor in the 6D is actually 22mp, but they cut it a bit to still use sensors with flaws at one or two sides... I'm just guessing of course as this would be trade secrets.

 

[dgatwood]

Now there are certainly other factors (like the wasted silicon near the edges of the wafer), but there are ways of dealing with that, too (like printing DIGIC chips and other custom parts in that space).

... or decreasing the sensor resolution from 22mp to 20mp like 5d3->6d, I guess next to marketing that's the reason they did this no matter how bad the video moire is because it lets you get more yield from the current production setup.

The other possibility that the 20mp sensor in the 6D is actually 22mp, but they cut it a bit to still use sensors with flaws at one or two sides... I'm just guessing of course as this would be trade secrets.

No, the pixel density on the 6D is lower than the 5Dmk3. The flaws on a sensor would typically be equiprobable in the center versus on the edge, so there wouldn't be much of a yield advantage to doing those sorts of tricks.

I suspect, but am not certain, that yield problems on sensors are caused more by the electronic features—that is, the wiring on the back side or front side of the sensor—than by the pixels themselves. This is one place where fully back-illuminated (stacked) sensor designs like the Exmor should have a real advantage. If you don't have to worry about the amount of vignetting caused by the wiring being in front of your photo sites, you should be able to make those features larger, which means better yield.

With such a design, you could even do neat tricks like stacking a buffer behind the photo sites to eliminate rolling shutter without the loss of sensitivity that you get with interline transfer sensors or the huge surface area needed for frame transfer sensors.

 

[jrista]

Now there are certainly other factors (like the wasted silicon near the edges of the wafer), but there are ways of dealing with that, too (like printing DIGIC chips and other custom parts in that space).

... or decreasing the sensor resolution from 22mp to 20mp like 5d3->6d, I guess next to marketing that's the reason they did this no matter how bad the video moire is because it lets you get more yield from the current production setup.

The other possibility that the 20mp sensor in the 6D is actually 22mp, but they cut it a bit to still use sensors with flaws at one or two sides... I'm just guessing of course as this would be trade secrets.

No, the pixel density on the 6D is lower than the 5Dmk3. The flaws on a sensor would typically be equiprobable in the center versus on the edge, so there wouldn't be much of a yield advantage to doing those sorts of tricks.

I suspect, but am not certain, that yield problems on sensors are caused more by the electronic features—that is, the wiring on the back side or front side of the sensor—than by the pixels themselves. This is one place where fully back-illuminated (stacked) sensor designs like the Exmor should have a real advantage. If you don't have to worry about the amount of vignetting caused by the wiring being in front of your photo sites, you should be able to make those features larger, which means better yield.

With such a design, you could even do neat tricks like stacking a buffer behind the photo sites to eliminate rolling shutter without the loss of sensitivity that you get with interline transfer sensors or the huge surface area needed for frame transfer sensors.

BI designs aren't technically "stacked". Stacked designs are basiclly where a sensor and a DSP are fabricated independently, then wired together into a single package. I'd also point out that it is only Exmor RS, not Exmor, that is a BI design. Exmor, the one used in the D800 and A7r, are good old FI designs. Sony's edge is that they use a smaller transistor, 180nm vs. Canon's 500nm. Smaller circuitry features do mean less space consumed by wiring and the like, leaving more room for photodiode area.

Etching both sides of a silicon wafer, as in a BI design, has it's own challenges. Usually the back side of the wafer is untouched, giving rigidity to the whole sensor. Etching both sides results in increased fragility, which is why BI designs are currently only used in small form factors, and not ASP-C or FF sensors. There are some patents out there, including one by Canon, that aim to address these issues with BI designs and allow their use in larger form factors...who knows if/when they might actually be viable enough for mass production.

Yield issues come into play when defect count per sensor area increases beyond a reasonable threshold. As I understand it, one defect in a pixel's area can render the entire pixel unusable. This has to do with the photodiode, not necessarily the wiring. When there are only a few pixels affected by small defects, those pixels can simply be deactivated and mapped. They effectively become dead pixels, but every sensor has them, and you never know it because of the demosaicing process which basically filters over them. Sensors with larger run a higher risk of losing more pixels per sensor, and if you lose too many (not sure what the threshold is) you can't as easily gloss over the problem. There is also the issue of larger defects, which render the entire sensor unusable. Smaller sensors means that while a large defect may render a whole sensor unusable, because you can pack so many more onto each wafer, those few lost to defects don't cost you as much. The problem of lost sensors due to large defects becomes exponentially more costly, as you lose the whole area of the sensor, which could span two or more times as much area as a single smaller sensor (plus, you also have margin losses where you can't fit whole FF sensors into areas near the periphery of the wafer, so you have yield losses there as well.)

 

[Marsu42]

No, the pixel density on the 6D is lower than the 5Dmk3.

Thanks to you & jrista for the interesting posts! And you're correct, I looked it up - both sensors have the same physical size, and with the the same pixel size (6.25 micron) obviously the pixel density has to be different and the sensors cannot be from the same production line.

 

[dgatwood]

There are other ways to do multilayer chips besides etching the back side.