Pages: [1]   Go Down

Author Topic: May MF CMOS be around the corner?  (Read 3966 times)

ErikKaffehr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 11311
    • Echophoto
May MF CMOS be around the corner?
« on: December 26, 2012, 04:46:43 am »

Hi,

We had a discussion about the future of MF, but it was closed because the discussion sort of navigated away from the original issue.

The impression I had was that there are quite a few users of technical cameras who would prefer have live view. The new Alpa FPS for instance would make a lot more sense with live view, as it is my understanding that it is not intended to be used with a sliding back. The Hartblei HCAM has a motorized sliding back on the other hand.

My guess is that CMOS may be around. Leica has a new full frame sensor developed by CMOSIS. It may make a lot of sense to develop a CMOS sensor for larger formats. The major benefits could be better live view and some stops of extra DR, with DR advantage coming mostly from reduced readout noise.

What do you think, CMOS around the door or not?

Best regards
Erik
Logged
Erik Kaffehr
 

FredBGG

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1630
Re: May MF CMOS be around the corner?
« Reply #1 on: December 26, 2012, 05:45:38 am »

MF CMOS is technically possible. However for it to be "around the corner" it has to be
financially viable.
Who would have the financial resources to bring it to market.
We currently have a new generation of backs that have the same sensors as the previous backs.
Those sensor manufacturers are focusing their resources towards more lucrative and larger medical, scientific and
military markets. Beautiful pictures have little to do with those markets and they go far beyond the visual spectrum.

I also think that if CMOS were to come to MF digital it will not be very soon and it will be over shadowed by
faster advancements in 35mm sensor technology like Fuji's new sensor and fovion type sensors with no moire issues
and other types of artifacts.

I suspect that Sigma is getting close to a full frame fovion type sensor. They certainly have come out with a few of extremely high quality full frame
lenses, one of them being the 85mm 1.4. Right now I would say it's the best 85mm 1.4 there is as far as sharpness and contrast.
Even wide open it is outstanding.

I still think that there is a place for MF and it is principally landscape with very large printing.
I think that an interesting option for live view and CMOS on tech cameras would be a fast and efficient stitch back
with a d800s type sensor in it.
A compact back with an internal mechanism for shifting the sensor quickly. Center the sensor for focusing,
than shift the sensor the two shots producing a virtual 645 sensor.
Also an architectural photographer could simply use the same setup, but only one shot, but taking advantage of the "film plane"
shift. Maybe even add sensor based tilt in a higher end version.

One would also have the possibility of shooting motion landscape using the same sensor and camera, but just adding an extra 35mm system wide angle
for non stitch work.

For composition one could still use the singe centered sensor using a rather crude add-on lens to simulate two sensor stitch.

It would be possible to make such a D800 sensor based stitch backs for less than current MF backs.... however
it would also be possible to make much less expensive MF backs. Pentax makes a complete camera for less than current backs.
« Last Edit: December 26, 2012, 05:55:30 am by FredBGG »
Logged

Gel

  • Full Member
  • ***
  • Offline Offline
  • Posts: 240
Re: May MF CMOS be around the corner?
« Reply #2 on: December 26, 2012, 06:07:48 am »

Two Canon 1DX chips sat next to each other can't be a hard thing to do.

Although Canon hasn't done it, must be a reason for that.

FredBGG

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1630
Re: May MF CMOS be around the corner?
« Reply #3 on: December 26, 2012, 06:26:12 am »

Two Canon 1DX chips sat next to each other can't be a hard thing to do.

Although Canon hasn't done it, must be a reason for that.

If you mean putting two sensors together to make one large sensor... it's not that simple.
First of all you would need to redesign the sensor completely as the standard canon or Nikon sensors
have output electronics on all four sides. You would have to get rid of that on one side to seamlessly but two sensors up against each other.

I think the main reason neither Canon or Nikon have not ventured into MF or bought an MF company is that
there is not enough of a market to make it worth their while. Both Nikon and especially Canon preferred to
venture out into motion picture. Interestingly both Nikon and Canon have opened offices right in the heart of Hollywood.
It's a different story for Phase One and Hasselblad. They are stuck with what they have to build on.
Neither have the resources to move into other formats and fields so they have to work within the confines of the cameras they have.
Logged

ErikKaffehr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 11311
    • Echophoto
Re: May MF CMOS be around the corner?
« Reply #4 on: December 26, 2012, 07:20:20 am »

Hi,

Leica has developed their sensor in cooperation with CMOSIS. Quite possible they could develop an MF sensor, too. I presume that designing an MF sensor is no more complex than designing a similar full frame sensor. Manufacture is probably much more expensive.

Best regards
Erik


If you mean putting two sensors together to make one large sensor... it's not that simple.
First of all you would need to redesign the sensor completely as the standard canon or Nikon sensors
have output electronics on all four sides. You would have to get rid of that on one side to seamlessly but two sensors up against each other.

I think the main reason neither Canon or Nikon have not ventured into MF or bought an MF company is that
there is not enough of a market to make it worth their while. Both Nikon and especially Canon preferred to
venture out into motion picture. Interestingly both Nikon and Canon have opened offices right in the heart of Hollywood.
It's a different story for Phase One and Hasselblad. They are stuck with what they have to build on.
Neither have the resources to move into other formats and fields so they have to work within the confines of the cameras they have.
Logged
Erik Kaffehr
 

Bart_van_der_Wolf

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 8913
Re: May MF CMOS be around the corner?
« Reply #5 on: December 26, 2012, 07:46:40 am »

I also think that if CMOS were to come to MF digital it will not be very soon and it will be over shadowed by faster advancements in 35mm sensor technology like Fuji's new sensor and fovion type sensors with no moire issues and other types of artifacts.

Hi Fred,

Foveon type sensors also ehibit moiré issues, just not false color moiré. Moiré is inherent in (under)sampling of fine (high spatial frequency) detail, and is most noticeable in repetitive patterns positioned at an angle to the sampling grid, and high contrast edges which will lead to stairstepping/jaggies.

Quote
I suspect that Sigma is getting close to a full frame fovion type sensor.


I suspect they are not, because the design of the sensor doesn't play well with light that strikes at oblique angles, such as near the edges of a larger sensor array, or with Tilt and Shift (lenses/cameras).

Cheers,
Bart
Logged
== If you do what you did, you'll get what you got. ==

Bart_van_der_Wolf

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 8913
Re: May MF CMOS be around the corner?
« Reply #6 on: December 26, 2012, 07:55:49 am »

You would have to get rid of that on one side to seamlessly but two sensors up against each other.

Hi Fred,

This is already the current practice on MF sized sensors. The stitching is done in photolithography, so no physical assembly is required, but it does require wafer steppers that can do that with adequate positioning accuracy.

Quote
I think the main reason neither Canon or Nikon have not ventured into MF or bought an MF company is that
there is not enough of a market to make it worth their while.


That, and coupled with the required wafer stepper equipment to do the stitching. Also, the trend towards wafer steppers with ever smaller feature capabilities also opens the path to more sensels on the existing sensor real estate (which also reduces aliasing artifacts due to the higher sampling density).

Cheers,
Bart
« Last Edit: December 26, 2012, 07:59:03 am by BartvanderWolf »
Logged
== If you do what you did, you'll get what you got. ==

FredBGG

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1630
Re: May MF CMOS be around the corner?
« Reply #7 on: December 26, 2012, 01:47:24 pm »

Fred
Quote
You would have to get rid of that on one side to seamlessly but two sensors up against each other.

Hi Fred,

This is already the current practice on MF sized sensors. The stitching is done in photolithography, so no physical assembly is required, but it does require wafer steppers that can do that with adequate positioning accuracy.
Cheers,
Bart

And this is not without problems. There are MF users sometimes getting lines in images and having to have their sensors replaced or calibrated.


Here is an example of the problem for one user. Here he turned up the saturation to show the problem more clearly online.
Logged

Gel

  • Full Member
  • ***
  • Offline Offline
  • Posts: 240
Re: May MF CMOS be around the corner?
« Reply #8 on: December 26, 2012, 03:29:13 pm »


What do you think, CMOS around the door or not?

Best regards
Erik

2 years tops. Hasselblad will be first.
Something someone at Hasselblad said to me recently strongly suggested this.

'Big changes in the next couple of years'
'Something new on the way'

BJL

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 6600
May MF CMOS be around the corner? Butting two sensors vs on-silicon stitching
« Reply #9 on: December 26, 2012, 04:50:23 pm »

This is already the current practice on MF sized sensors. The stitching is done in photolithography, so no physical assembly is required, but it does require wafer steppers that can do that with adequate positioning accuracy.
Bart,
We need to distinguish two different things: Gel and Fred are talking about first making two sensors, with no contacts along one edge, and then butting them together. This is in fact done by Teledyne-Dalsa for example to make large X-ray sensors, but it produces visible join lines that would be unacceptable in a DMF camera. What is instead done to make all camera sensors 36x24mm and up is so-called stitching on the silicon wafer: etch two or more parts of the sensor side-by-side on the silicon wafer, moving the wafer between parts. A number of fabs can do this including the one used by Leica and CMOSIS to make the new Leica M sensor, but the rejects rate is apparently high, and so makes sensor costs high, and rapidly higher for even larger sensors that require even more pieces to be joined on the wafer.
Logged

RVB

  • Jr. Member
  • **
  • Offline Offline
  • Posts: 58
Re: May MF CMOS be around the corner?
« Reply #10 on: December 27, 2012, 07:36:00 am »

At a push I would put my money on Leica being the first to use cmos in MF,the next generation of the S will likely have a sensor made by cmosis ..
Logged

design_freak

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1128
Re: May MF CMOS be around the corner?
« Reply #11 on: December 28, 2012, 07:40:45 am »

It is possible to produce, the question is for how much it would cost.
http://www.canon.com/news/2010/aug31e.html
Logged
Best regards,
DF

ErikKaffehr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 11311
    • Echophoto
Re: May MF CMOS be around the corner?
« Reply #12 on: December 28, 2012, 07:51:48 am »

Hi,

What is the crop factor on 8x10" ?

Best regards
Erik


It is possible to produce, the question is for how much it would cost.
http://www.canon.com/news/2010/aug31e.html
Logged
Erik Kaffehr
 

BJL

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 6600
Re: May MF CMOS be around the corner?
« Reply #13 on: December 28, 2012, 09:44:49 am »

It is possible to produce, the question is for how much it would cost.
http://www.canon.com/news/2010/aug31e.html
That sensor has huge photosites, allowing the use of large, low resolution fabrication equipment that could not be used to produce a sensor suitable for medium format. With large enough photosites, even the equipment used to make LCD panels can be used, allowing sensors the size of flat panel TV's! In fact one custom 10"x8" sensor has been made like this, as a "viewfinder" for a large format film photographer, but it has only a few MP.
Logged

Stefan.Steib

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 471
    • HCam - Hartblei Pro Photography solutions
Re: May MF CMOS be around the corner?
« Reply #14 on: December 28, 2012, 10:33:42 am »

The sensor alone wouldn´t cut the slice needed.
The market now demands more comfort and better wideangles, Movements and highspeed shutter and syncing.
The actual MF technology portfolios would need major overhauls to fulfill these.
Without a move to mirrorless, electronic finders and A LOT of R&D, which needs to be paid firsthand there will not be
a catch up to the fast moving "semipro/amateur" market that cannibalized from the bottom to the middle and is now moving further upwards.

With the current model of distribution this cannot be done, the whole cost/efficiency structure of the "Highend Market" dealers are in question.

This needs to be analyzed closely and  drastic steps to be taken. Otherwise.......

Greetings from Germany
Stefan
Logged
Because Photography is more than Technology and "as we have done it before".

ErikKaffehr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 11311
    • Echophoto
Re: May MF CMOS be around the corner?
« Reply #15 on: December 29, 2012, 01:34:10 am »

Stefan,

Can you elaborate on the need for innovation? The features you mentioned are those your customers are asking for? I am sort of aware that you are involved with high speed flash.

The question is in part which features are asked for, or more accurately, which features customers may be willing to pay for. One of the problems is that the customers are very different. Some need a lot of flexibility (think fully rigged Canon 5DIII for video) and some want to have film like simplicity. I guess shooting for a catalogue in studio makes very different demands than shooting a waterfall in the rain forrest.

Best regards
Erik

The sensor alone wouldn´t cut the slice needed.
The market now demands more comfort and better wideangles, Movements and highspeed shutter and syncing.
The actual MF technology portfolios would need major overhauls to fulfill these.
Without a move to mirrorless, electronic finders and A LOT of R&D, which needs to be paid firsthand there will not be
a catch up to the fast moving "semipro/amateur" market that cannibalized from the bottom to the middle and is now moving further upwards.

With the current model of distribution this cannot be done, the whole cost/efficiency structure of the "Highend Market" dealers are in question.

This needs to be analyzed closely and  drastic steps to be taken. Otherwise.......

Greetings from Germany
Stefan
Logged
Erik Kaffehr
 

Stefan.Steib

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 471
    • HCam - Hartblei Pro Photography solutions
Re: May MF CMOS be around the corner?
« Reply #16 on: December 29, 2012, 01:57:43 am »

Hi Eric

A better CMOS sensor would certainly solve a momentary gap, but the reason is more based on the economic and maker side.
As Cameras today MUST become fully electronic to fulfill the features wanted by the customers to fully utilize their digital systems
(e.g. conncetivity with either highspeed local wire - or wireless networks, internet access,precise Video preview, compactnesss and reasonable price)
the numbers of the systems must be improved by doing a programmable , universal, standardized and modular platform, that can be produced economically.
The mechanistic, small numbers with customizations done by individual modifications model will not work anymore,only very few will be able to pay for such systems
(30000 € and up) even then these need to use electronics and software, the smaller numbers will make these even more expensive and the price MUST rise further
until even the highend customers  will rethink their buying decision and the gap between R&D  and the catchup to the semipro industrial cameras will make the
race and even a try obsolete.
This is not so much about MPix, the whole workflow and the integration into a computerized customerbase will do the most pressure.
Phase/Leaf and Hasselblad are doing their own software to keep control of the image devellopment and distribution stage, but this will become more and more complex
and demanding in the future, this part alone is very costly and a real task to solve for such comparably small companies.
Highend customers alone cannot generate the necessary turnaround for all these activities- means they need products that go into a broader distribution channel
which in return means that their exclusive dealership models get into questioning.

Then the lenses: Highend lenses for MF in low numbers will also become a luxury, making it even more problematic to get the necessary turnaround.

There is more , but these are the most important facts.

Regards
Stefan
« Last Edit: December 29, 2012, 02:02:27 am by Stefan.Steib »
Logged
Because Photography is more than Technology and "as we have done it before".

ErikKaffehr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 11311
    • Echophoto
Re: May MF CMOS be around the corner?
« Reply #17 on: December 29, 2012, 02:26:45 am »

Hi,

I would suggest that MFD vendors would be wiser to fully support leading raw image processing applications like Lightroom and Capture One. In this regard I think they should do this:

- Use DNG as a raw format
- Make available spectral data for the sensor and any calibration data

With CMOS sensors I think they should need to think about usability. Add peaking, zebras and pre whitebalance raw histograms. Also they should provide a HDMI feed for external monitors. A standard bus and protocol for changing settings would be a good thing.

I guess that sensor cost is much more related to sensor surface area than to resolution, I guess that pricing is artificial. It may be better to sell more backs at a lower price. I would also suggest it would be better to standardize on a single sensor size, so lenses could be developed for a single size of sensor.

Just a few ideas.

Best regards
Erik



Hi Eric

A better CMOS sensor would certainly solve a momentary gap, but the reason is more based on the economic and maker side.
As Cameras today MUST become fully electronic to fulfill the features wanted by the customers to fully utilize their digital systems
(e.g. conncetivity with either highspeed local wire - or wireless networks, internet access,precise Video preview, compactnesss and reasonable price)
the numbers of the systems must be improved by doing a programmable , universal, standardized and modular platform, that can be produced economically.
The mechanistic, small numbers with customizations done by individual modifications model will not work anymore,only very few will be able to pay for such systems
(30000 € and up) even then these need to use electronics and software, the smaller numbers will make these even more expensive and the price MUST rise further
until even the highend customers  will rethink their buying decision and the gap between R&D  and the catchup to the semipro industrial cameras will make the
race and even a try obsolete.
This is not so much about MPix, the whole workflow and the integration into a computerized customerbase will do the most pressure.
Phase/Leaf and Hasselblad are doing their own software to keep control of the image devellopment and distribution stage, but this will become more and more complex
and demanding in the future, this part alone is very costly and a real task to solve for such comparably small companies.
Highend customers alone cannot generate the necessary turnaround for all these activities- means they need products that go into a broader distribution channel
which in return means that their exclusive dealership models get into questioning.

Then the lenses: Highend lenses for MF in low numbers will also become a luxury, making it even more problematic to get the necessary turnaround.

There is more , but these are the most important facts.

Regards
Stefan
Logged
Erik Kaffehr
 
Pages: [1]   Go Up