Why has this sensor not been made?

Started May 19, 2012 | Discussions
Great Bustard
Forum ProPosts: 21,952
Like?
Why has this sensor not been made?
May 19, 2012

The D800 sports a FF version of the D7000 sensor, which is why it's so good. How about this? A FF version of the G12 sensor?

It would have 100 MP, a 55% QE, per-pixel read noise of 4.3 electrons (throughout the ISO range, i.e., ISOless), and a per-pixel saturation of 10306 electrons.

9.5 of these pixels would cover the same area a 5D3 pixel (which has a QE of 49%). Thus, the combined read noise for 9.5 pixels would be 13.2 electrons, as compared the the 5D3 pixel which has a read noise of 33.1 electrons at base ISO to a low of 2.4 electrons at ISO 12800), and a combined saturation of 97692 electrons compared to 67531 for the 5D3.

This would put the DR at 12.9 stops at base ISO (compared to 11.0 stops for the 5D3) down to 7.5 stops at ISO 12800 (compared to 7.8 stops for the 5D3).

And all this with the old tech of the G12 pixels.

Why not?

Steen Bay
Veteran MemberPosts: 6,188
Like?
Re: Why has this sensor not been made?
In reply to Great Bustard, May 19, 2012

Great Bustard wrote:

The D800 sports a FF version of the D7000 sensor, which is why it's so good. How about this? A FF version of the G12 sensor?

It would have 100 MP, a 55% QE, per-pixel read noise of 4.3 electrons (throughout the ISO range, i.e., ISOless), and a per-pixel saturation of 10306 electrons.

The sensor would have 200+ MP, and wouldn't it be better to use Canon's own S100 pixels? The S100 has slightly lower QE, but also lower read noise, and the read noise is reduced further at higher ISOs.

9.5 of these pixels would cover the same area a 5D3 pixel (which has a QE of 49%). Thus, the combined read noise for 9.5 pixels would be 13.2 electrons, as compared the the 5D3 pixel which has a read noise of 33.1 electrons at base ISO to a low of 2.4 electrons at ISO 12800), and a combined saturation of 97692 electrons compared to 67531 for the 5D3.

This would put the DR at 12.9 stops at base ISO (compared to 11.0 stops for the 5D3) down to 7.5 stops at ISO 12800 (compared to 7.8 stops for the 5D3).

And all this with the old tech of the G12 pixels.

Why not?

Reply   Reply with quote   Complain
Great Bustard
Forum ProPosts: 21,952
Like?
Re: Why has this sensor not been made?
In reply to Steen Bay, May 19, 2012

Steen Bay wrote:

Great Bustard wrote:

The D800 sports a FF version of the D7000 sensor, which is why it's so good. How about this? A FF version of the G12 sensor?

It would have 100 MP, a 55% QE, per-pixel read noise of 4.3 electrons (throughout the ISO range, i.e., ISOless), and a per-pixel saturation of 10306 electrons.

The sensor would have 200+ MP...

Yes -- apologies for the arithmetic error.

...and wouldn't it be better to use Canon's own S100 pixels? The S100 has slightly lower QE, but also lower read noise, and the read noise is reduced further at higher ISOs.

Absolutely. And, as a bonus, even more pixels. So, why not do it? Yes, the frame rate would be abysmal, and video IQ would suffer as well due to forrendous line skipping.

But isn't there enough of a market for a FF landscape camera without the fps and video that can still deliver the goods in low light environments?

Or am I completely underestimating how difficult it is to make a FF size sensor with pixels of this size and/or the electronics required to process a 200+ MP photo?

Seems to me that it's not a matter of the tech, but a matter of fps and video IQ vs max photo IQ.

Reply   Reply with quote   Complain
hunk
Veteran MemberPosts: 3,214Gear list
Like?
Re: Why has this sensor not been made?
In reply to Great Bustard, May 19, 2012

my guess
1 - the fab line reserved for dslr's isn't capable of this newer tech
2 - marketing, the market would be too small to bother...

 hunk's gear list:hunk's gear list
Leica M9 Canon Extender EF 2x II Sony 135mm F2.8 (T4.5) STF
Reply   Reply with quote   Complain
bobn2
Forum ProPosts: 27,829
Like?
Re: Why has this sensor not been made?
In reply to hunk, May 19, 2012

hunk wrote:

my guess
1 - the fab line reserved for dslr's isn't capable of this newer tech
2 - marketing, the market would be too small to bother...

I think both of those are on the money. Offering a product in advance of the market's expectations can produce a winner. Offering a product so far in advance that the market simply doesn't understand it usually produces a loser. The world is full of failures which were successes released ten years too early.
--
Bob

Reply   Reply with quote   Complain
Mikael Risedal
Veteran MemberPosts: 3,000
Like?
Re: Why has this sensor not been made?
In reply to Great Bustard, May 19, 2012

Ask Sony why .

Very Expensive, and they have hardly any lithography equipment with such accuracy that it can handle a large 24x36mm CCD sensor without stitching.

Great Bustard wrote:

The D800 sports a FF version of the D7000 sensor, which is why it's so good. How about this? A FF version of the G12 sensor?

It would have 100 MP, a 55% QE, per-pixel read noise of 4.3 electrons (throughout the ISO range, i.e., ISOless), and a per-pixel saturation of 10306 electrons.

9.5 of these pixels would cover the same area a 5D3 pixel (which has a QE of 49%). Thus, the combined read noise for 9.5 pixels would be 13.2 electrons, as compared the the 5D3 pixel which has a read noise of 33.1 electrons at base ISO to a low of 2.4 electrons at ISO 12800), and a combined saturation of 97692 electrons compared to 67531 for the 5D3.

This would put the DR at 12.9 stops at base ISO (compared to 11.0 stops for the 5D3) down to 7.5 stops at ISO 12800 (compared to 7.8 stops for the 5D3).

And all this with the old tech of the G12 pixels.

Why not?

-- hide signature --

Life is a battle wishes aunt Titti

Reply   Reply with quote   Complain
mischivo
Contributing MemberPosts: 597
Like?
Used LinkedIn
In reply to Great Bustard, May 19, 2012

You should do a bit of LinkedIn (or other social network) investigating to find a CCD or CMOS engineer and ask them these questions. They are far more likely to provide you with a more satisfactory answer than anyone on these forums.

My own guess is that such a sensor is possible but would cater to a market so small as to make it completely not worth the manufacturers while to produce.

Nikon recommends using a tripod with the D800 to take full advantage of its resolving power. I can't think of a single tripod that would be able to hand a 200mp camera. Hell, the vibration induced by the shutter would probably be enough to create pixel-level motion smear.

-- hide signature --

portfolio :: http://pavelkounine.com
daily blog :: http://pavelkounine.com/blog

I shoot film. I shoot digital. Both are great in their own ways. The nice thing about film is that I can always have the most advanced imaging technology around packed into my 20 year old camera. That's one aspect of digital that I loathe.

Reply   Reply with quote   Complain
Mikael Risedal
Veteran MemberPosts: 3,000
Like?
Re: Why has this sensor not been made?
In reply to Steen Bay, May 19, 2012

Two different sensors, one CCD and one CMOS.
Both extremely expensive to produce in a 24x36mm area (today)

Steen Bay wrote:

Great Bustard wrote:

The D800 sports a FF version of the D7000 sensor, which is why it's so good. How about this? A FF version of the G12 sensor?

It would have 100 MP, a 55% QE, per-pixel read noise of 4.3 electrons (throughout the ISO range, i.e., ISOless), and a per-pixel saturation of 10306 electrons.

The sensor would have 200+ MP, and wouldn't it be better to use Canon's own S100 pixels? The S100 has slightly lower QE, but also lower read noise, and the read noise is reduced further at higher ISOs.

9.5 of these pixels would cover the same area a 5D3 pixel (which has a QE of 49%). Thus, the combined read noise for 9.5 pixels would be 13.2 electrons, as compared the the 5D3 pixel which has a read noise of 33.1 electrons at base ISO to a low of 2.4 electrons at ISO 12800), and a combined saturation of 97692 electrons compared to 67531 for the 5D3.

This would put the DR at 12.9 stops at base ISO (compared to 11.0 stops for the 5D3) down to 7.5 stops at ISO 12800 (compared to 7.8 stops for the 5D3).

And all this with the old tech of the G12 pixels.

Why not?

-- hide signature --

Life is a battle wishes aunt Titti

Reply   Reply with quote   Complain
Mikael Risedal
Veteran MemberPosts: 3,000
Like?
Re: Used LinkedIn
In reply to mischivo, May 20, 2012

Why, I have often a dialog with Eric Fossum by mail and there are a number of others with great knowledge here at Dpreview. BobN2 DSP,Marianne,John Sheehy, The Suede Emil etc etc etc
Start to google at - who is Eric Fossum

mischivo wrote:

You should do a bit of LinkedIn (or other social network) investigating to find a CCD or CMOS engineer and ask them these questions. They are far more likely to provide you with a more satisfactory answer than anyone on these forums.

My own guess is that such a sensor is possible but would cater to a market so small as to make it completely not worth the manufacturers while to produce.

Nikon recommends using a tripod with the D800 to take full advantage of its resolving power. I can't think of a single tripod that would be able to hand a 200mp camera. Hell, the vibration induced by the shutter would probably be enough to create pixel-level motion smear.

-- hide signature --

portfolio :: http://pavelkounine.com
daily blog :: http://pavelkounine.com/blog

I shoot film. I shoot digital. Both are great in their own ways. The nice thing about film is that I can always have the most advanced imaging technology around packed into my 20 year old camera. That's one aspect of digital that I loathe.

-- hide signature --

Life is a battle wishes aunt Titti

Reply   Reply with quote   Complain
Mikael Risedal
Veteran MemberPosts: 3,000
Like?
Re: Used LinkedIn
In reply to Mikael Risedal, May 20, 2012

Here you have a list by Daniel Browning

http://forums.dpreview.com/forums/read.asp?forum=1018&message=39452529

And Eric Fossum needs no presentation, the inventor of the active cmos

-- hide signature --

Life is a battle wishes aunt Titti

Reply   Reply with quote   Complain
Mikael Risedal
Veteran MemberPosts: 3,000
Like?
Re: Used LinkedIn
In reply to mischivo, May 20, 2012

And regarding resolving power

Tell all 7d or Sony 24mp APS owners that they need to have theirs cameras on a tripod = much higher pixel density than d800 .

mischivo wrote:

You should do a bit of LinkedIn (or other social network) investigating to find a CCD or CMOS engineer and ask them these questions. They are far more likely to provide you with a more satisfactory answer than anyone on these forums.

My own guess is that such a sensor is possible but would cater to a market so small as to make it completely not worth the manufacturers while to produce.

Nikon recommends using a tripod with the D800 to take full advantage of its resolving power. I can't think of a single tripod that would be able to hand a 200mp camera. Hell, the vibration induced by the shutter would probably be enough to create pixel-level motion smear.

-- hide signature --

portfolio :: http://pavelkounine.com
daily blog :: http://pavelkounine.com/blog

I shoot film. I shoot digital. Both are great in their own ways. The nice thing about film is that I can always have the most advanced imaging technology around packed into my 20 year old camera. That's one aspect of digital that I loathe.

-- hide signature --

Life is a battle wishes aunt Titti

Reply   Reply with quote   Complain
Glassfish
Regular MemberPosts: 319
Like?
Re: Why has this sensor not been made?
In reply to Great Bustard, May 20, 2012

Perhaps, keeping incremental product road map.

Reply   Reply with quote   Complain
Phil Hill
Senior MemberPosts: 2,536Gear list
Like?
Yield determines cost.
In reply to Great Bustard, May 20, 2012

At that density, the manufacturing issue is most like low yield. It’s one thing to discard a number of smaller chips, as long as you get a good number of keepers. Quite another when a single small defect kills an entire larger chip.

Reply   Reply with quote   Complain
Simon21
Junior MemberPosts: 36
Like?
Re: Why has this sensor not been made?
In reply to Great Bustard, May 20, 2012

Well, I think that Canon could have done exactly what Nikon (Sony) did: take the 18 MPX APS-C sensor from the 60D or even from the older 7D and make a FF sensor derived from it and with the same technology. This would lead to approximately 36-40 MPX with a similar quality/performance level of the APS-C sensor.

It has been asserted that it is what Nikon (Sony) did when using the D7000 sensor as the base to produce the D800 sensor. Notice that while it roughly increase its physical size by a factor of 2, and also multiply the MPX by approximately 2, it is always possible to make some improvements, usually minor ones in terms of development but possibly major ones in terms of improved capabilities.

The point is that in order to do that, a new line is NOT needed, the same technology which was used to produce the APS-C sensor can be used to produce the new sensor, this CMOS technology can be 45nm, 33nm, 22nm, whatever the APS-C was made of. The obvious question is the yield of the line to produce a much bigger VLSI chip.

It is not the only question of course. For example in addition to the "light capture" devices on this sensor chip, associate with each "pixel" there is some processing and some "intelligence" required, which also takes room and probably results in a not linear circuitry increase. Even though I have not been involved with this sort of things in a long time, I am sure it is relatively easy to do this "evolution" especially with the design automation tools available today. Actually it should be quite fast in terms of chip design and development. Integrating and testing the new sensor in a camera system is probably another story (on which I cannot comment.)

Why Canon did not do it, well your guess is as good as mine. But I will give you my opinion for what it is worth: it simply was a marketing mistake and probably a big one, at least for now. When Canon planning team put their requirement for their new cameras, I think they assumed that the MPX race was "passé" and they focused their efforts in improving their best selling 5D Mark II in all other areas, and the same for their flagship camera... This turns out to be a good move for many users, but cannot be summarized with a single number having a huge marketing value like "36 something" for others.

For example, look at their successful G series evolution: without going to the earlier ones you can see the increase of MPX with each model:G8 8MPX, G9 12.1 MPX, G10 14.7 Mpx and then dropping the sensors of the G11 and the G12 to 10MPX (OK these were CCD sensors but it is irrelevant.)

This was also the time when some people were exalting the merits of the 12MPX Nikon D700 in low light, high ISO as compared to the 5D II, and much less noise was made of the 21 MPX vs 12 MPX advantage of the 5DII.

This possibly took place at the time Canon made their decision for their new cameras and my guess is that they put the effort where they perceived being weaker and did not care or underestimated the value of the MPX factor.

Technically I am absolutely convinced that it would have been easily done. Unlike Nikon, Canon have their own CMOS high density fab lines and design teams and even if they do not have the very latest lines ones compared to what Sony is using (in terms of 33 or 22nm), they do not need Sony. Their 18 MPX APS-C sensors are existence proof that they can build chips with the right density to achieve 36 or more MPX, and most probably have already produced such sensors, at least at the prototype level.

Personally, my first DSLR, which followed many film SLR cameras, was a Canon 20 D. When the 1 DS II was announced I could not resist it and I acquired one. I must say that at that time if I knew that the 1 DS was an excellent camera and that the Mark II was even better, its 16+ MPX was a key factor in my decision at the time. I own a lot of Canon L lenses and for what I am doing (grand children pictures, wild life, family pictures) my old 1DS Mark II is still an excellent camera. Also I do that as a hobby and since I wrote the above (where I am/was definitely more knowledgeable than in photography) I would appreciate from photograph experts opinions, facts, on the advantages of the 5D III vs the 1DS II for me. Thanks.

Reply   Reply with quote   Complain
arthon
Regular MemberPosts: 247
Like?
Re: Why has this sensor not been made?
In reply to Simon21, May 20, 2012

Canon have already those sensors proposed here.

In 2007 they showed a 50MP CMOS (8000 x 6000) 19mm x 28mm; http://www.wired.com/gadgetlab/2007/06/canons_50_megap/

And in 2010 they showed a 120 megapixel CMOS (13280 x 9184) APS-H; http://www.pcpro.co.uk/news/360568/canon-unveils-120-megapixel-camera-sensor

Canon have always been the forefront of digital camera technology but they seems to have "slipped" lately. Either by miscalculated marked strategy or that they have lost some inventive sensor engineers.

The 5D3 is barley a new product, and most crucial, the sensor is no better than the previous model. I just has some slightly improved processing firmware.

Maybe now, when they get back their old CEO they lost in 2006, they will get back on course and release more inventive cameras again.

Reply   Reply with quote   Complain
mischivo
Contributing MemberPosts: 597
Like?
Re: Used LinkedIn
In reply to Mikael Risedal, May 20, 2012

I think that is relative to angle of view. Given an identical angle of view captured by the sensor, the D800 is more challenging. I feel that what you're describing, although correct in asserting the higher pixel density, is only true when comparing these cameras while using a lens of equal focal length. A 200 megapixel sensor is a far cry less tolerant than a hypothetical full-frame 7D with 46.6 megapixels or NEX7 with 57.28 megapixels.
--
portfolio :: http://pavelkounine.com
daily blog :: http://pavelkounine.com/blog

I shoot film. I shoot digital. Both are great in their own ways. The nice thing about film is that I can always have the most advanced imaging technology around packed into my 20 year old camera. That's one aspect of digital that I loathe.

Reply   Reply with quote   Complain
mischivo
Contributing MemberPosts: 597
Like?
OK, you win, you win! :)
In reply to Mikael Risedal, May 20, 2012

Are all of these user currently employed as CCD or CMOS sensor engineers at one of these companies though?

For all I know, perhaps we may have some anonymous users. I shouldn't have made such a sweeping statement.
--
portfolio :: http://pavelkounine.com
daily blog :: http://pavelkounine.com/blog

I shoot film. I shoot digital. Both are great in their own ways. The nice thing about film is that I can always have the most advanced imaging technology around packed into my 20 year old camera. That's one aspect of digital that I loathe.

Reply   Reply with quote   Complain
David Hull
Veteran MemberPosts: 5,330Gear list
Like?
How are they going to get the data out of it?
In reply to Great Bustard, May 20, 2012

Looking at the throughput right now:

Canon 5DIII: 22.3MPx6FPS=133.8MP/s
Nikon D800: 36.3MPx4=144Mp/s

So if such a thing were built into a single processor camera, you can expect a firing rate well under2FPS. I doubt that many would be too impressed with that or interested. It would be a very specialized camera at best.

There are other solutions -- multi processors (1Dx gets 216 MP/s) but the TP would still suck.

My guess is that while this is possible, some product line manager doesn't think it would be a big seller.

 David Hull's gear list:David Hull's gear list
Canon EOS 50D Canon EOS 5D Mark III
Reply   Reply with quote   Complain
Mikael Risedal
Veteran MemberPosts: 3,000
Like?
Re: Why has this sensor not been made?
In reply to Simon21, May 20, 2012

NOPE

The G series and fewer pixels after a while had nothing to do that the pixel race was passe or ower.

Sony started their first BSI CCD construction at 10 Mp and this sensor was used in for example s-90 , s-95 G11, G12 , Nikon P and many more cameras
Now Sony has 16Mp BSI.

Canon has no modern sensor lines like Sony,Aptina, Omnivision and others.
The latest line is the line that making 60D sensors.

a discussion about Canons sensor lines you find here
http://forums.dpreview.com/forums/read.asp?forum=1018&message=40493801

Simon21 wrote:

For example, look at their successful G series evolution: without going to the earlier ones you can see the increase of MPX with each model:G8 8MPX, G9 12.1 MPX, G10 14.7 Mpx and then dropping the sensors of the G11 and the G12 to 10MPX (OK these were CCD sensors but it is irrelevant.)

This was also the time when some people were exalting the merits of the 12MPX Nikon D700 in low light, high ISO as compared to the 5D II, and much less noise was made of the 21 MPX vs 12 MPX advantage of the 5DII.

This possibly took place at the time Canon made their decision for their new cameras and my guess is that they put the effort where they perceived being weaker and did not care or underestimated the value of the MPX factor.

Technically I am absolutely convinced that it would have been easily done. Unlike Nikon, Canon have their own CMOS high density fab lines and design teams and even if they do not have the very latest lines ones compared to what Sony is using (in terms of 33 or 22nm), they do not need Sony. Their 18 MPX APS-C sensors are existence proof that they can build chips with the right density to achieve 36 or more MPX, and most probably have already produced such sensors, at least at the prototype level.

Personally, my first DSLR, which followed many film SLR cameras, was a Canon 20 D. When the 1 DS II was announced I could not resist it and I acquired one. I must say that at that time if I knew that the 1 DS was an excellent camera and that the Mark II was even better, its 16+ MPX was a key factor in my decision at the time. I own a lot of Canon L lenses and for what I am doing (grand children pictures, wild life, family pictures) my old 1DS Mark II is still an excellent camera. Also I do that as a hobby and since I wrote the above (where I am/was definitely more knowledgeable than in photography) I would appreciate from photograph experts opinions, facts, on the advantages of the 5D III vs the 1DS II for me. Thanks.

-- hide signature --

Life is a battle wishes aunt Titti

Reply   Reply with quote   Complain
Mikael Risedal
Veteran MemberPosts: 3,000
Like?
Re: Used LinkedIn
In reply to mischivo, May 20, 2012

The problems starts when the cell / pixel size is under 3 micron and you try to use F 1,2 F1,4 at 24x36mm large sensor area, and the problems are mainly at the borders and corners.

mischivo wrote:

I think that is relative to angle of view. Given an identical angle of view captured by the sensor, the D800 is more challenging. I feel that what you're describing, although correct in asserting the higher pixel density, is only true when comparing these cameras while using a lens of equal focal length. A 200 megapixel sensor is a far cry less tolerant than a hypothetical full-frame 7D with 46.6 megapixels or NEX7 with 57.28 megapixels.
--
portfolio :: http://pavelkounine.com
daily blog :: http://pavelkounine.com/blog

I shoot film. I shoot digital. Both are great in their own ways. The nice thing about film is that I can always have the most advanced imaging technology around packed into my 20 year old camera. That's one aspect of digital that I loathe.

-- hide signature --

Life is a battle wishes aunt Titti

Reply   Reply with quote   Complain
Keyboard shortcuts:
FForum MMy threads