Ad
Ad
Ad
Pages: « 1 [2] 3 4 »   Bottom of Page
Print
Author Topic: A true 6x7 CMOS low light sensor camera, can it exist?  (Read 15742 times)
Gigi
Sr. Member
****
Offline Offline

Posts: 422


WWW
« Reply #20 on: August 20, 2012, 03:54:40 PM »
ReplyReply

D800 with fast f:1.4 to f:2 primes produces practically the same files that MF backs with f:2.8 and f:4 lenses do, with better DR and high ISO performance
to boot. The problem is that D800 is small, cheap, fast and has too many convenient features.

-1
Logged

Geoff
torger
Sr. Member
****
Offline Offline

Posts: 1588


« Reply #21 on: August 20, 2012, 04:12:33 PM »
ReplyReply

as to why there isn't anything really groundbreaking out of the MFD arena. The most obvious to me would be a much larger sensor.

I'm waiting for something groundbreaking in pricing strategy :-).
Logged
torger
Sr. Member
****
Offline Offline

Posts: 1588


« Reply #22 on: August 21, 2012, 01:24:00 AM »
ReplyReply

Oh, it can be interesting mentioning that bigger sensors has existed in past products; Dicomed Bigshot 4000 launched in 1996 had a 60x60mm sensor with 16.8 megapixels. The color version was $55,000:

http://www.epi-centre.com/reports/9604cs.html
Logged
hjulenissen
Sr. Member
****
Offline Offline

Posts: 1683


« Reply #23 on: August 21, 2012, 02:40:03 AM »
ReplyReply

While we are on the "far-fetched" train: What about printing a large image sensor yourself, using your printer with specialized inc? Granted, a solar cell panel is not the same as a 100 megapixel 6x7 camera sensor, quantum efficiency (?) of < 1% may still not beat the D800, and most of us dont have a vacum chamber. Still...

http://web.mit.edu/newsoffice/2011/printable-solar-cells-0711.html

My point being that if people are really willing to pay the price (that is a big if) in terms of cost, or "image quality per unit area", or "sensor robustness", I am sure that the present obstacles to large-sensor imaging (prohibitive costs before the first sensor is even manufactured, prohibitive number of flawed sensors) will someday be worked around.

-h
Logged
torger
Sr. Member
****
Offline Offline

Posts: 1588


« Reply #24 on: August 21, 2012, 03:38:30 AM »
ReplyReply

Dicomed went out of business in 1999, but he company that made the CCD (Loral Fairchild) is still in business and still makes the 60x60mm CCDs, and you can buy cameras from them, designed for scientific use of course.

And there's Spectral Instruments (http://www.specinst.com/) that also makes scientific cameras, here's one https://www.youtube.com/watch?v=6gEu3T8DcrI 112 megapixels at 95x95mm, with excellent dynamic range thanks to extreme cooling. Could be cool mounted to a 4x5" view camera :-).

In 2010 Canon made a 202x205mm CMOS sensor which was later installed in a telescope. Pixel pitch on that is 160 um though so it is quite different from a normal photographic sensor.
Logged
BJL
Sr. Member
****
Offline Offline

Posts: 5163


« Reply #25 on: August 21, 2012, 02:34:06 PM »
ReplyReply

Short answer: increasing cost penalties for large, low volume formats and decreasing IQ advantages for those larger formats have for a long been driving an overall trend of downsizing, and the transition from film to digital intensified this.

Doug Peterson had a lot of good points, including:
- lack of any 6x7 autofocus lenses
...
Have you personally shot for an extended period of time with a FF 645 sensor (e.g. IQ160, Credo 60) and a fast lens? What is it you find lacking? I'd be shocked if the answer was image quality, resolution, or the ability to limit DOF. If the answer is high ISO performance then the answer is not to go bigger.

Even before the transition from film to digital there was a down-sizing trend, in particular away from the larger medium format options like 6x7 and 6x6 to 645. The fact that AF was developed in medium format mostly for 645, with only one larger format AF system (Rollei's 6x6) and that one at best struggling, if not on its death bed, is just one indication of this. Why? I think because with modern film and even more so with digital sensors, any image quality advantages of formats larger than 645 are so small that they are of relevance to a very small proportion even of professional and fine-art amateur photography, so that the demand for such gear became very small, worsening economies of scale and leading to an ever increasing cost difference, while technological progress lead to ever decreasing image quality advantages. Together, a vicious spiral dragging down the economic viability of larger formats. (It has not been very kind to 645 or even 35mm format either, each of which has a far smaller market share in digital than it had with film.)

Also, as already explained already in this thread, low volume production of large sensors has far greater cost penalties than simply putting the same emulsion on a wider roll of film, or just making different sized images (6x7 vs 6x6 vs 6x4.5) on the same rolls of film.
« Last Edit: August 23, 2012, 10:02:16 AM by BJL » Logged
BJL
Sr. Member
****
Offline Offline

Posts: 5163


« Reply #26 on: August 21, 2012, 02:58:44 PM »
ReplyReply

hjulenissen and torger,

The big problem is not just making big sensors, but making them with photosites small enough to be of interest. There are many very big sensors available for uses like X-rays and astronomy, but they all have big pixel pitches like 15 microns and up ... often way up, like 160 microns. In 6x7 format, 15 microns gives about 4600x4000, unlikely to be competitive in IQ with what modern near 645 format sensors are giving.

The reason is that to get pixels small enough to even match film resolution requires using modern semi-conductor fabrication equipment all of which has standardized on a maximum field size of 26x33mm. Using these to make a sensor larger than about 24x31mm requires a complicated, low yield, expensive fabrication process of "on-wafer stitching". (as you probably both know, this is _not_ making smaller sensors and then joining them afterwards.)

On the other hand, for lower resolution scientific and manufacturing needs, there are cheaper options like an old, low resolution stepper by Canon with large 50x50mm field size, or butting together several sensors and ignoring the join lines (acceptable for X-rays) or using steppers designed for LCD fabrication which have very large formats but very low resolution. But the ability to make those large, very low resolution devices is not the slightest evidence that there will ever be higher resolution semi-conductor fabrication equipment capable of make a high quality MF sensor without on-chip stitching.
Logged
EricWHiss
Sr. Member
****
Offline Offline

Posts: 2427



WWW
« Reply #27 on: August 21, 2012, 10:18:16 PM »
ReplyReply

D800 with fast f:1.4 to f:2 primes produces practically the same files that MF backs with f:2.8 and f:4 lenses do, with better DR and high ISO performance
to boot.

The only part of what you wrote that I don't really wonder if its true is the part about high ISO, but if you are shooting wide open with fast glass do you need high ISO?    I do wonder if a D800 can produce a similar look and sort of doubt that it could.   I don't think the 5D2 comes all that close even with my leica 80/1.4 summilux fitted.     As far as I know there is no substitution for sensor size in terms of look.    I guess I will have to borrow a D800 and have a look at wide open shots.   I can tell you that when we compared the D800E in studio using the same exact subject framing, the D800 was getting more DOF at f/10 than I was at f/16 on my AFi-ii 12.   I was surprised, but this would lead me to conclude that with my 110/2 or 80/2, I'd be getting shallower DOF than a D800 with a 1.4 optic and maybe even with my AFi at f/2.8.   Test it yourself.  YMMV. Of course DOF is just one facet of the character of an image.
Logged

Authorized Rolleiflex Dealer:
Find product information, download user manuals, or purchase online - Rolleiflex USA
uaiomex
Sr. Member
****
Offline Offline

Posts: 1005


WWW
« Reply #28 on: August 21, 2012, 10:54:29 PM »
ReplyReply

The constant word blaming the insane cost of big sensors is YIELD.  I understand that it is impossible to manufacture a perfect wafer, meaning that you always get a wafer with several bad sections. So you start with a "big" wafer with numerous defects. Pixels at random, or row of pixels, I don't know.

Now, what if you start with a wafer with a small pixel size like the one used to create the D800 sensors. If cut to the size of 6X7, it would in theory have 163 megapixels. That is 163 million pixels counting good pixels and bad pixels. Obtaining a 6X7 sensor under current industry standards for number of bad pixels in it to be approved for sale, would make it prohibitive, if not impossible.

I understand that bad pixels are mapped out in factory, so this is not new. So, why not "mapped out" bad pixels of the "rejected" sensors by pixel binning creating a sensor with an output of half the sensors? The lower count by binning and mapping would totally disguise the bad pixels and still keep 80 million pixels of incredible tonality to spare.

Yield sensor would increase dramatically, right? So, what happens if I sell every (or most) piece of 56X70mm from this wafer as an 80 megapixel sensor?  Suddenly, TRUE medium format digital sensors would be possible at earthly prices.

I'm sure a back with a true 6X7 sensor with 80 mp's would put to shame an IQ180 which is not even true 645. A strategy like this would invigorate the whole digital medium format market and will take it out of the endangered species list. imho.

Eduardo
« Last Edit: August 21, 2012, 11:21:17 PM by uaiomex » Logged
hjulenissen
Sr. Member
****
Offline Offline

Posts: 1683


« Reply #29 on: August 22, 2012, 12:41:28 AM »
ReplyReply

If it was possible to allow any wafer defects by simply interpolating "bad pixels", then yes, I assume that the yield would increase dramatically.

I think that would necessitate a new way of reading those pixels, though (increased independence). Right now, many of the gates on the sensor seems to affect many pixels. Any error in any one of them might cause visible errors in larger parts of the output image. Other errors might cause power consumption or heat to sky-rocket (?). The success of recent Sony sensors from integrating ADC on the sensor seems to indicate that the way forward is to have _more_ logic on the image sensor itself.

-h
Logged
Petrus
Sr. Member
****
Online Online

Posts: 532


« Reply #30 on: August 22, 2012, 12:41:44 AM »
ReplyReply

  I do wonder if a D800 can produce a similar look and sort of doubt that it could. 

If we approach this from the mathematical angle, the field of descriptive geometry, it really does not matter what size system we have, we can scale it up and down and the end result (the image) is mathematically always the same in every regard (when blown up to the same print size). Getting a similar look is thus a matter of having exactly equal lenses (they are not), using exactly correct apertures to achieve same DOF, and having exactly similar sensors giving same DR, color, pixel count etc. Calculating DOF is possible, but using matching lenses and sensors is not possible in real life. Then there is also the mental thing knowing which system was used to take the picture.
Logged
hjulenissen
Sr. Member
****
Offline Offline

Posts: 1683


« Reply #31 on: August 22, 2012, 12:49:07 AM »
ReplyReply

hjulenissen and torger,

The big problem is not just making big sensors, but making them with photosites small enough to be of interest.
...
I was referring to statements such as the one below. If people are willing to pay for a large sensor, even if that means high cost _and_ low quality per unit area, then the question only is "how high cost?" and "how low quality?" before one can predict if and when such a product might appear.

Even 24x36mm sensors from the large, mainstream manufacturers seems to (usually) be made from mature technology. I.e. lower sensel density than their crop models. Later introduction of new features (such as improvements in base-ISO DR). Current MF sensors seems to be made using really "old" technology.

Using these other technologies that you suggest, how large would an image sensor have to be before resolution (and other quality aspects) was acceptable (e.g. 50 Megapixel)? I think that printable sensors sounds like an exciting idea if it allows regular people to experiment with really large sensors, or replacing the film of their grandfathers 8x10. Not because it will ever have better fidelity than a D800, but because fiddling with such thing might produce interesting end-results and enlightened practitioners.
...As far as I know there is no substitution for sensor size in terms of look....
Logged
torger
Sr. Member
****
Offline Offline

Posts: 1588


« Reply #32 on: August 22, 2012, 03:06:56 AM »
ReplyReply

hjulenissen and torger,

Thanks for the info. The spectral instruments camera in my linked video seems to be a bit different though, it has an STA 1600 CCD 95x95mm with 9um pixel size 10560x10560 pixels resolution.

http://www.sta-inc.net/update-of-sta1600-10560-x-10560-high-resolution-ccd/

It seems to me that it does not have any gaps in it. It is indeed designed for astronomy applications though.
Logged
ErikKaffehr
Sr. Member
****
Offline Offline

Posts: 7670


WWW
« Reply #33 on: August 22, 2012, 11:20:54 AM »
ReplyReply

Hi,

Michael Reichmann had a video on LLVJ on Phase One, and they said that they frequently mapped out entire columns of pixels. I also think they said rejection rate was quite high, more then ten percent.

Alex Koslov posted an image from a 39 MP Hasselblad back where I would say it was pretty obvious a column of pixels were missing.

Best regards
Erik

If it was possible to allow any wafer defects by simply interpolating "bad pixels", then yes, I assume that the yield would increase dramatically.

I think that would necessitate a new way of reading those pixels, though (increased independence). Right now, many of the gates on the sensor seems to affect many pixels. Any error in any one of them might cause visible errors in larger parts of the output image. Other errors might cause power consumption or heat to sky-rocket (?). The success of recent Sony sensors from integrating ADC on the sensor seems to indicate that the way forward is to have _more_ logic on the image sensor itself.

-h
Logged

Steve Hendrix
Sr. Member
****
Offline Offline

Posts: 1132


WWW
« Reply #34 on: August 22, 2012, 03:39:05 PM »
ReplyReply

While there's been a lot of focus on pricing in this thread, I've always seen the exercise of creating larger sensors as similar to CPU production. Why does the next generation processor only go up so much? It's become quite incremental. Where computing power has expanded is by adding cores, but for imaging sensors, this is not without challenges. So - in addition to price, I see the challenge as a technical one (not un-related to price), successfully manufacturing a large sensor that produces the same level of quality (or higher would be nice, but it cannot be lower) as today's medium format CCD sensors.

The short answer to the OP is that Nikon and Canon only have to create a 36mm x 24mm sensor size. Over the year's there has been speculation that either would look into medium format, but I've been pessimistic on that prospect because I think - despite the capabilities of Canon/Nikon/Sony - that creating a larger sensor system would not be a slamdunk technically for anyone. That said, today I could see that possibility more than ever, as smaller and more nimble cameras (including their own) cut into DSLR marketshare.

It's my hope that sensor sizes continue to increase - even if it's incrementally - because at least for now, that remains a very distinct potential difference between 35mm and medium format. And we're currently at 645 full frame.....


Steve Hendrix
Capture Integration
Logged

Steve Hendrix
Sales Manager, www.captureintegration.com (e-mail Me)
MFDB: Phase One/Leaf-Mamiya/Hasselblad/Leica/Sinar
TechCam: Alpa/Cambo/Arca Swiss/Sinar
Direct: 404.543.8475
Codger
Jr. Member
**
Offline Offline

Posts: 74


« Reply #35 on: August 22, 2012, 06:16:51 PM »
ReplyReply

"And we're currently at 645 full frame....."

Steve, who's producing this?  The largest I've read about is 54x40, or there-abouts.
Logged
Doug Peterson
Sr. Member
****
Offline Offline

Posts: 2838


WWW
« Reply #36 on: August 22, 2012, 06:27:52 PM »
ReplyReply

The p65+, iq160, iq180, credo 60, credo 80 and Aptus-ii 12 are all full frame 645 as defined by being the same size as the exposed area of film or the view through the viewfinder of a Mamiya/phase 645 body.

Perhaps you were under the impression 645 format film was 6.0x4.5cm? It was not.
Logged

DOUG PETERSON (dep@digitaltransitions.com), Digital Transitions
Dealer for Phase One, Mamiya Leaf, Arca-Swiss, Cambo, Profoto
Office: 877.367.8537
Cell: 740.707.2183
Phase One IQ250 FAQ
Steve Hendrix
Sr. Member
****
Offline Offline

Posts: 1132


WWW
« Reply #37 on: August 22, 2012, 06:41:02 PM »
ReplyReply

"And we're currently at 645 full frame....."

Steve, who's producing this?  The largest I've read about is 54x40, or there-abouts.


What Doug said. The effective capture area (minus a few microns) is as close to full frame 645 as possible. Assuming a 10% - 15% increase in the imaging area (the past 3 generations) we might expect 60mm x 44mm in the next generation, and that would exceed the digital capture dimensions for a 645 camera.


Steve Hendrix
Capture Integration
Logged

Steve Hendrix
Sales Manager, www.captureintegration.com (e-mail Me)
MFDB: Phase One/Leaf-Mamiya/Hasselblad/Leica/Sinar
TechCam: Alpa/Cambo/Arca Swiss/Sinar
Direct: 404.543.8475
FredBGG
Sr. Member
****
Offline Offline

Posts: 1651


« Reply #38 on: August 22, 2012, 07:46:36 PM »
ReplyReply


What Doug said. The effective capture area (minus a few microns) is as close to full frame 645 as possible. Assuming a 10% - 15% increase in the imaging area (the past 3 generations) we might expect 60mm x 44mm in the next generation, and that would exceed the digital capture dimensions for a 645 camera.


Steve Hendrix
Capture Integration

Would that require a new camera?  645 film frames were never 645. 120 roll film is 60mm wide and there are frame numbers and other junk pre exposed on the edges.
I think the actual long side size is 54mm. Would the lenses have enough coverage for 60mm without problems in the corners....
Logged
Steve Hendrix
Sr. Member
****
Offline Offline

Posts: 1132


WWW
« Reply #39 on: August 22, 2012, 08:06:51 PM »
ReplyReply

Would that require a new camera?  645 film frames were never 645. 120 roll film is 60mm wide and there are frame numbers and other junk pre exposed on the edges.
I think the actual long side size is 54mm. Would the lenses have enough coverage for 60mm without problems in the corners....


Might require a new camera. Nothing wrong with that. It would make Eric happy. My hope is that indeed it does require a new camera, because I hope the future holds more sensor size expansion, and in that case, a new camera is essential.


Steve Hendrix
Capture Integration
Logged

Steve Hendrix
Sales Manager, www.captureintegration.com (e-mail Me)
MFDB: Phase One/Leaf-Mamiya/Hasselblad/Leica/Sinar
TechCam: Alpa/Cambo/Arca Swiss/Sinar
Direct: 404.543.8475
Pages: « 1 [2] 3 4 »   Top of Page
Print
Jump to:  

Ad
Ad
Ad