MORE Canon Mega-Pixels, 250 of them, but none for you!

[privatebydesign]

Just curious. Why would Canon test this technology on an APS-H size sensor?

Is it

A) Because it's actually an APS-HDTV sensor, not the made up APS-H format: they are going after the 70mm widescreen movie market

B) Because Canon don't see so called full frame as the be all and end all? They leave that to 5D owners.

C) Because the 1D line is going to diverge again. This chip is for the 1DV. The 1DXs gets 19MP but a stop more DR.

A) No, because they list the size "APS-H-size (approx. 29.2 x 20.2 mm)"

B) They probably see 135 format as the be all and end all of the enthusiast and pro photographer market, not least because all their high end lenses are specifically made for it. How much they believe the photography market will be their future revenue earner, and what percentage of that income, who knows.

C) No, that sensor is ten to fifteen years away from being put in a production camera, if then. As Mt Spokane says it will be tethered and require power and processing capabilities not available in the size and cost needed to fit inside a small room at this point.

I hated the APS-H sensor size, no ultrawides, I suppose we have the 11-24 now, but we can use that on FF too.......

Buy Tinky, keep stirring that pot, you are good at it ;D

 

[Orangutan]

Is this Canon's only trick?
More megapixels?

No, absolutely not. Remember that Canon is a for-profit business: even if they had super alien technology for 24-stop DR with noise almost non-existent, it's still expensive to retool, and that would raise prices, take money away from other aspects of the camera, or cut into profits. As we've seen with the Sony A7 models, they work well in certain areas (and getting better). But build quality and ergonomics are not yet getting glowing reviews, and Sony's lens and accessory line is still no match for Canon or Nikon. And Sony's support is still, according to the reports I've seen, pretty bad.

Companies have a finite amount of money for R&D, retooling and support. If they pour it all into sensor design and production, it will take away from other aspects of the camera. It's just basic business economics.

 

[privatebydesign]

Is this Canon's only trick?
More megapixels?

No, absolutely not. Remember that Canon is a for-profit business: even if they had super alien technology for 24-stop DR with noise almost non-existent, it's still expensive to retool, and that would raise prices, take money away from other aspects of the camera, or cut into profits. As we've seen with the Sony A7 models, they work well in certain areas (and getting better). But build quality and ergonomics are not yet getting glowing reviews, and Sony's lens and accessory line is still no match for Canon or Nikon. And Sony's support is still, according to the reports I've seen, pretty bad.

Companies have a finite amount of money for R&D, retooling and support. If they pour it all into sensor design and production, it will take away from other aspects of the camera. It's just basic business economics.

I'd word that a bit stronger, Sony's lens and accessory line is a pathetic shadow of the big two, and their answer is to get third parties to build lens adapters that "perform faster than on native cameras" apart from all the times they don't and as evidenced quite clearly on a multitude of videos.

Maybe Sony has it fundamentally wrong. Maybe bodies are a dead end and the money is in the lenses and accessories, lets face it the bodies are mature tech at this point and few people are buying every new model on release because we are looking at incremental increases in performance and the price isn't going down. I am still shooting 1Ds MkIII's because for me Canon haven't brought out a better alternative yet, but I have been happy to throw down thousands at the 16-35 f4 IS, the 11-24 f4, the 35 f2 IS, the TS-E17, the 100L Macro, four 600EX-RT's etc since getting those old bodies.

To be honest the improvements in software and lenses have done more for my IQ than a new sensor.

 

[Aglet]

I am going w/ the positive. Lots of development, see what can make it into a DSLR or C body.

As signal noise gets lower through whatever means (materials, engineering, etc) is there a cross over point where as some else mentioned the over sampling provides a better data stream than larger pixels for DR/Low light?

The oversampling, in a case like this with ~1.4 micron sensels, would be able to provide very good color resolution if you mash 4 (RGGB) into one full-color sampled pixel with ~ 62MP total output.

The DR measurement would benefit from a reduced noise floor in a similar way as those grouped sensels would average out the baseline noise to improve DR by ~ 1 stop.
BUT - those tiny sensels would also need to have a pretty good full-well-count to provide the numerator to the averaging-reduced-denominator since DR is a ratio of maximum measureable light to minimum resolveable light(vs noise)

If those tiny sensels work as well as the ones in a recent Panasonic compact camera reviewed on DxOmark:

www.dxomark.com/Reviews/Panasonic-Lumix-DMC-ZS50-sensor-review-Few-pixels-equals-better-dynamics/Measurements-Good-but-behind-best-in-class

... which provide ~11 stops worth of DR..

Then the 4-sensel binning would provide ~ 12 stops of DR at 62MP and 13 stops of DR at ~15MP may be acheived with 16-sensel binning.

If the hardware was up to the task (no fixed pattern noise), then yes, such a high rez sensor could provide some interesting options trading resolution for DR. Could probably still maintain base ISO at 100 as well but it would be challenging to have a good performance at higher ISO as the intrinsic system noise would become a larger proportion of the signal.
Always tradeoffs.

 

[rrcphoto]

I can't believe there are EEYORES in this thread...

I find this interesting, not because of the sensor itself - but the fact that canon created a large sized sensor with 1.5um pixel pitch.

that means one of two things, they've improved the accuracy of the 180nm line - or they've now prototyped high density sensors using a new line.

unfortunately, the whitepapers for the 120MP or probably this sensor will never reach the light of day. however, even for the 50Mp sensor, canon was already experimenting with large sensors with the 180nm line.

so while this sensor means probably nothing to the ILC world, it's fascinating that canon has probably taken a geometry leap as far as sensor architecture, most likely down to between 120 to 90nm.

now before people cry about that.. keep in mind the 36MP sensor in the D800/D810 only uses 180nm geometry.

also .. why APS-H? it's the largest sensor canon can make using steppers and a single pass / no stitch process.

 

[Botts]

excerpt:

When installed in a camera, the newly developed sensor was able to capture images enabling the distinguishing of lettering on the side of an airplane flying at a distance of approximately 18 km from the shooting location.2

2. Image capture employed a combination of optical and digital zooming while distinguishing of image content was realized through the magnification of an approximately 1/40,000th-sized area of the captured image.

-

I wonder what lens...

If Canon is publishing this, you wonder how crazy the sensor and lens packages are on NRO birds.

 

[Mt Spokane Photography]

I'd suspect that a APS-H sensor was selected because using a larger sensor can not be done in one pass due to lithography limitations.

The production systems may be able to do it in one pass, but R&D folks have to get the most bang for their buck, and live with the hand me down tools currently available to them. At least, that's how it was for my lab. I could not touch production tools, they ran 24 X 7 churning out production parts.

 

[HighLowISO]

This is abviously awesome technology for people who understand technology.
Just because some don't have the vison or creativity to see how this might be used in the real world does not mean anything. So maybe it won't be part of camera aimed at photography, if it's not for you just move on.

I'm sure Canon gave up reading this forum long ago, but hats off to them for another big score!

 

[jrista]

I am going w/ the positive. Lots of development, see what can make it into a DSLR or C body.

As signal noise gets lower through whatever means (materials, engineering, etc) is there a cross over point where as some else mentioned the over sampling provides a better data stream than larger pixels for DR/Low light?

The oversampling, in a case like this with ~1.4 micron sensels, would be able to provide very good color resolution if you mash 4 (RGGB) into one full-color sampled pixel with ~ 62MP total output.

The DR measurement would benefit from a reduced noise floor in a similar way as those grouped sensels would average out the baseline noise to improve DR by ~ 1 stop.
BUT - those tiny sensels would also need to have a pretty good full-well-count to provide the numerator to the averaging-reduced-denominator since DR is a ratio of maximum measureable light to minimum resolveable light(vs noise)

If those tiny sensels work as well as the ones in a recent Panasonic compact camera reviewed on DxOmark:

www.dxomark.com/Reviews/Panasonic-Lumix-DMC-ZS50-sensor-review-Few-pixels-equals-better-dynamics/Measurements-Good-but-behind-best-in-class

... which provide ~11 stops worth of DR..

Then the 4-sensel binning would provide ~ 12 stops of DR at 62MP and 13 stops of DR at ~15MP may be acheived with 16-sensel binning.

If the hardware was up to the task (no fixed pattern noise), then yes, such a high rez sensor could provide some interesting options trading resolution for DR. Could probably still maintain base ISO at 100 as well but it would be challenging to have a good performance at higher ISO as the intrinsic system noise would become a larger proportion of the signal.
Always tradeoffs.

I think such a camera would be awesome. With the binning, readout rate should increase as well, which would make for an extremely versatile DSLR.

Sadly, this does not seem to be within Canon's repertoire. I could see Nikon making something like that, though.

 

[privatebydesign]

Sadly, this does not seem to be within Canon's repertoire. I could see Nikon making something like that, though.

Not unless Sony make the sensors for them, and even then only a year after Sony release their version.

 

[gsealy]

This kind of resolution can be used in automated quality control. The camera takes high resolution photos of parts, for example, and that data is fed into software, which does the analysis.

 

[RGF]

Could APS-H be making a comeback?

Just think if this was full frame - 1.3 x 1.3 = 1.7 or 400+ MP

 

[ritholtz]

Sadly, this does not seem to be within Canon's repertoire. I could see Nikon making something like that, though.

Not unless Sony make the sensors for them, and even then only a year after Sony release their version.

After a year without Sony sensor pdaf goodies. Looks like, this thread is also going to be DR'ed. With all those Sony inventions (stacked, bsi and exmorr), their advantage is low iso DR. Everything else, they are same as Canon. Sony is doing more in terms of on sensor pdaf. Canon is doing their own on sensor pdaf in terms of hybrid sensor and duel pixel.
Bill Claff measures actual dynamic range (photographic dynamic range). He confirmed that ML duel iso gives more than 11 stops of PDR which is same as Exmor (d750). One needs to figure out how to use ML instead of complaining about DR in every thread. Simple ML hack with Canon sensor can match Exmor DR without all those inventions. Not sure why Canon just implement duel iso in their firmware?

 

[privatebydesign]

Sadly, this does not seem to be within Canon's repertoire. I could see Nikon making something like that, though.

Not unless Sony make the sensors for them, and even then only a year after Sony release their version.

After a year without Sony sensor pdaf goodies. Looks like, this thread is also going to be DR'ed. With all those Sony inventions (stacked, bsi and exmorr), their advantage is low iso DR. Everything else, they are same as Canon. Sony is doing more in terms of on sensor pdaf. Canon is doing their own on sensor pdaf in terms of hybrid sensor and duel pixel.
Bill Claff measures actual dynamic range (photographic dynamic range). He confirmed that ML duel iso gives more than 11 stops of PDR which is same as Exmor (d750). One needs to figure out how to use ML instead of complaining about DR in every thread. Simple ML hack with Canon sensor can match Exmor DR without all those inventions. Not sure why Canon just implement duel iso in their firmware?

Check my history dude, I have never worried about Canon DR and never will. When I get more I will use it but until then I will apply my skill, experience, and knowledge to get shots like this.

http://www.canonrumors.com/forum/index.php?topic=27593.msg545205#msg545205

With a follow up here

http://www.canonrumors.com/forum/index.php?topic=27593.msg545446#msg545446

 

[DavidA]

I was at the US Open Tennis matches over the weekend. I was walking around late evening Monday and saw a guy with a strange box attached to a Canon EF 500mm. As I look at the image Canon Rumors posted with the Canon announcement, I realized that the rig I saw must have been a test camera in similar setup. My thought is that it may have been the 1dx II rather than a test camera with the 250mpx sensor. I was hurrying to get back to a match and wish I had tried to engage the guy in a conversation. If he had access to a test camera, I am sure he knew how to keep his mouth shut and wouldn't have disclosed anything useful.

Well, just trying to play on the other thread here..
http://www.canonrumors.com/forum/index.php?topic=27593.0

But Canon's latest press release goes on about an APS-H sensor with ~250MP (19,580 x 12,600 pixels)

www.canon.com/news/2015/sep07e.html

excerpt:

When installed in a camera, the newly developed sensor was able to capture images enabling the distinguishing of lettering on the side of an airplane flying at a distance of approximately 18 km from the shooting location.2

2. Image capture employed a combination of optical and digital zooming while distinguishing of image content was realized through the magnification of an approximately 1/40,000th-sized area of the captured image.

-

I wonder what lens...

 

[emko]

Sadly, this does not seem to be within Canon's repertoire. I could see Nikon making something like that, though.

Not unless Sony make the sensors for them, and even then only a year after Sony release their version.

After a year without Sony sensor pdaf goodies. Looks like, this thread is also going to be DR'ed. With all those Sony inventions (stacked, bsi and exmorr), their advantage is low iso DR. Everything else, they are same as Canon. Sony is doing more in terms of on sensor pdaf. Canon is doing their own on sensor pdaf in terms of hybrid sensor and duel pixel.
Bill Claff measures actual dynamic range (photographic dynamic range). He confirmed that ML duel iso gives more than 11 stops of PDR which is same as Exmor (d750). One needs to figure out how to use ML instead of complaining about DR in every thread. Simple ML hack with Canon sensor can match Exmor DR without all those inventions. Not sure why Canon just implement duel iso in their firmware?

It sounds like you never used duel iso with ML or you don't understand how it works. So we get lower resolution to get DR that is Exmor? when Exmor has 42mp ?

 

[LetTheRightLensIn]

Is this Canon's only trick?
More megapixels?

yes indeed dilbert:
horrid ergonomics, horrid build quality, horrid af, horrid accessory lineup, horrid menu/ui, horrid lenses (foremost), horrid free software, horrid customer service, horrid ...
only high megapixels. time to jump ship...

I don't get the hate. If people want to move on to another brand, why can't they do that quietly? Why make a big fuss about it and look like a 3rd grader who didn't get what he wanted for Christmas?

Because they are not convinced by what they do and seek confirmation from others...

Don't be ridiculous. If anything it's those who can't handle hearing that their camera is not this best at this or that given thing who seem to need confirmation of their choice.

It's because they'd rather not have to move on and get the company to wake up before they do move on so they don't have to.

 

[LetTheRightLensIn]

This is a hint at the sensor tech Canon has been developing.

I reckon this is Canon's way of saying: "be prepared to be impressed with the capabilities of the upcoming new versions of the 5D AND 1D cameras."

All good news for the upcoming models.

I would like to see the CR3 to your quote. I dare to dream, that you exactly hit the nail... 8) Especially as it says:
Additionally, despite the exceptionally high pixel count, Canon applied its sensor technologies cultivated over many years to realize an architecture adapted for miniaturized pixels that delivers high-sensitivity, low-noise imaging performance.

Hope we gonna see some of this tech in the 5DIV, or is it too late already...?

Well, as it seems, this new tech won't be implemented in upcoming bodies. Too sad... Over at theverge they say: "Don't expect this to land in your next DSLR or five, then. But the announcement is a sign that sensor technology will continue to improve; Canon announced a 120-megapixel APS-H sensor back in 2010, when this year's 50-megapixel EOS 5DS would have been unthinkable."

One problem is not a single apparent bit of that 2010 tech has appeared in any DSLR they have since released.
That one seemed to use on chip parallel ADC and so on but nothing remotely like that gets produced for their actual cameras.

 

[LetTheRightLensIn]

Hmm so should I got for the version with or without the AA filter....

 

[coldsweat]

Is it

A) Because it's actually an APS-HDTV sensor, not the made up APS-H format: they are going after the 70mm widescreen movie market

B) Because Canon don't see so called full frame as the be all and end all? They leave that to 5D owners.

C) Because the 1D line is going to diverge again. This chip is for the 1DV. The 1DXs gets 19MP but a stop more DR.

Love it!