Sensors, and Bit Depth, what might be next

[mjperini]

Bob,
I currently own a 1Ds mkIII which I got in November of 2007, and it is a superb camera. I did not upgrade to the 1Dx because I tend to shoot at iso 200, and make largish prints and didn't think the upgrade cost would be worth it for the work I do. I also have a 7D & 40D which I like quite a bit.  Frankly I had expected a 1Ds mk IV (1DsX??) by now..
As we look forward to the next generation of image sensors, the thing I look for most is increased dynamic range but also more rather than fewer Pixels.
Those MAY be mutually exclusive goals, but I hope not.
As I look at medium format where dynamic range is larger, I notice that all or most are 16bit files
I understand that higher bit rates only provide finer gradation to whatever dynamic range the given sensor provides, and I also understand that at any given point in time pixel size is more related to dynamic range (so in that regard higher MP counts work against it)
The fact that no high MP 1 series has appeared probably means Canon is wrestling with which way to go, but given the 36mp D800 and Sony's new 36mp entry as well as Sony's rumored 50+ MP experiments, my guess is we will see some response from canon (someday).
With that as background,  I wonder if you could discuss the technical issues facing the next step for FF sensors. Will we see 16bit FF or is that overkill? Might we see pixel binning so that a 48MP/12MP combo might be possible. Or, if dynamic range is the object, is FF done in the sense that bigger pixels on a larget than FF sensor the only way to get there today.(although the Sony 36mp seems to have done that fairly well) I've also read that Canon's new sensor FAB has to come on line for FF before they will be able to make their next step.
The more I think about this the more I realize how limited my true understanding is (especially about the sensor Fab stuff), so I immediately thought of you.

Again I am not asking about any specific Canon product, merely an informed discussion of technical options.
Thanks in advance for any light you care to shed on this subject.
Michael

[Bob Atkins]

I think one factor that might be important is the fact that Canon are using rather dated ( or you could call it "mature") fabrication technology. Canon sensors are fabricated with 0.5 micron design rules, while Nikon sensors are 0.35 or 0.25 (depending on the model) and Sony sensors are 0.25 or 0.18 microns. None of these are anywhere near the current lithography limit and for sensors they probably don't need to be. Sensors aren't that complex. However for Canon to make a major increase in pixel count they might have to go to 0.35 or 0.25 microns. I believe Canon does have 0.18 micron technology, though they haven't applied it to sensors.

I doubt there there's much to be gained by going to 16-bit readout, since I'm not sure the lower two bits would contain much more than noise with current 35mm sensors and pixel counts.

Sensor performance is also linked to the degree of integration of the processing circuitry (A/D conversion, readout electronics, noise reduction etc) onto the sensor chip. If well designed, better performance  can probably be obtained with this circuitry on the chip rather than off it.

I don't know where it is all going though, or what Canon's path will be. Sensors could certainly be built with on-chip pixel binning allowing either higher resolution or higher DR modes. Sensors could be built with pixels of different sizes and/or orientations to give improved performance (at least in theory). See, for example the Fuji sensors - http://finepix.com/exr_cmos/en/

For Canon to move from the existing 0.5 micron fabrication, which is now a mature technology with all the R&D paid off, to a new process would require a significant investment. It's hard to say if the current market would support such expenditure.

CMOSIS is making the sensor for the Leica M using a combination 0f 0.11 and 0.09 micron fabrication technology (see http://www.cmosis.com/news/press_releases/new_leica_m_uses_cmosis_24_mp_cmos_image_sensor). CMOSIS is big in the smartphone camera sensor market, where high resolution lithography is needed. Canon don't have that experience because as far as I know, they aren't in the smartphone business.

If you compare the sensor performance of the Leica sensor with, say the EOS 70D (Canon's latest), you'll see the Leica has lower noise and higher dynamic range (though only at low ISO, at high ISO the 70D is better)

Samsung might be someone to watch. They have extensive experience in all types of semiconductor and sensor fabrication. They could leverage that into an interesting APS-C or Full frame sensor.

My guess is that whatever the next Canon DSLR is (my guess is a 70D MkII) it will probably stick with 0.5 micron fabrication and it will use the dual pixel scheme of the 70D sensor to provide on chip phase sensitive AF. After that I don't know. Whether Canon will do something significant with the next EOS-1 series DSLR sensor I just can't guess. If they do, then they must be working on it now. However I suspect they will milk their current technology for as long as they can without losing too much competitive advantage.

[mjperini]

Thanks very much.
I'm now in my 7th year of using my 1DsIII. It's a superb camera and I can't  say I feel "Camera Limited" in any way.
That may be because I tend to stay as close to base iso as possible though the camera is pretty good at 1600 in a pinch.
It was a tour de force 7 years ago.
I'm sure these are better sensors available now, but I've been so in love with the 1Ds series as a total package I would find it hard to a smaller body.
I came from a 40D, (which I've kept) and in the interim bought my Wife who shoots weddings and events 2 7D's which is a superb camera.
I do much of her Post processing and when she fills the frame with a tight head shot (with full exposure) the detail from  the 7d sensor is astounding.
That pixel density in FF is well over 40 MP, so there seems to be room to grow a bit more.
So I'll just wait, and continue to make pictures.

[mjperini]

Bob,
I just came across the followinghttp://www.canonrumors.com/forum/index.php?topic=10545.0
I have no Idea how credible it is.

I will say, with the success Canon has enjoyed over the last decade using their 500 nm fab, and recognizing that most industry leaders are now using various 'smaller than 500nm" processes, It would be odd for Canon not to have recognized this trend and to have an 'answer' waiting in the wings.
I would guess that you are exactly right about their mikling the 500nm process as long as possible. Given the competition I would guess that it might be time to roll out another Flagship that is also destined to last 7 years ; -)) 

[Bob Atkins]

That's an old thread (on a site dedicated to rumors!).

There's basically no reason for anyone to assume that Canon are moving to 0.18 micron fabrication, other than someone would like it if they did. It's the "wishful thinking rumor". "I want it to be true so it might be".

I could post similar rumors, but since I don't know if it's true, I don't. It's logical to assume that eventually Canon will move to  higher resolution fabrication, but whether that will be in 10 days, 10 months or 10 years nobody outside Canon knows. It's not the sort of thing they are likely to announce ahead of time. We know for sure they have the capability of making a 120MP APS-H sensor because they've done it and shown it. That's about all we know for sure though.

Last time I spoke with the Canon technical guys (last year in NYC) I got the impression that they were far more concerned with getting the interface and performance of the 1D-X tailored to the needs of their professional customers (which is why the introduced significant new firmware for it) than chasing after MP. I didn't get the impression that they felt their pros were clamoring after higher resolution sensors.

[mjperini]

Thanks again Bob, I appreciate your taking the time.
I got the same impression from Canon at PhotoPlus on the floor and at the CPS lounge.
I took it with a grain of salt, because I have near zero clout with Canon other than being a user of a former flagship camera and a bunch of lenses.
I wouldn't expect that they would tell me anything about possible future products.
Both the 1Dx and the 5D mk III are superb cameras in every regard.
I also think that you are correct that there is probably not large demand for a 1D 'studio' / Landscape / Architecture camera, but there IS some.
And given Canon's history of offering 'high' MP Flagship cameras through the years, it seems an odd thing to abandon (1Dx 'merger' notwithstanding)
But so far at least, perceived demand hasen't been enough to change their mind.
So I will continue to wait, (and occasionally stitch) with the best camera I have ever owned.
It's aging far better than I am.
Michael