Pages: 1 ... 12 13 [14]   Go Down

Author Topic: You wait years for CMOS backs to arrive, and then... Disappointment.  (Read 50178 times)

eronald

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 6638
    • My gallery on Instagram
Re: You wait years for CMOS backs to arrive, and then... Disappointment.
« Reply #260 on: January 30, 2014, 04:51:50 pm »


BTW:  I have a question I've never understood.  Live view on ccd still cameras is close to awful, but they're were and still are ccd eng video cameras that obviously have super live view.  What is the difference between those sensors (though small) vs. the ccd's used in medium format cameras?

IMO

BC

The interline CCDs have lines of optically masked bucket brigade circuitry between lines of sensels; the masked circuits can convey the data off the chip during the time the unmasked circuitry accumulates an exposure. So called Fullframe sensors employ all the chip real estate for sensels, but are "frozen" during the exposure. I'm not convinced that one couldn't combine the benefits of both designs, but this ship has sailed as all the research money is poured into better CMOS camera phone sensors whose technology then trickles down to compacts, dSLRs and finally MF.

Edmund
« Last Edit: January 30, 2014, 04:57:48 pm by eronald »
Logged
If you appreciate my blog posts help me by following on https://instagram.com/edmundronald

ErikKaffehr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 11284
    • Echophoto
Re: You wait years for CMOS backs to arrive, and then... Disappointment.
« Reply #261 on: January 30, 2014, 04:57:14 pm »

Hi,

Using an interline design would use half of the surface for the output buffer, so DR would be cut in half and noise increase 41%.

Best regards
Erik


The interline CCDs have lines of optically masked bucket brigade circuitry between lines of sensels; the masked circuits can convey the data off the chip during the time the unmasked circuitry accumulates an exposure. So called Fullframe sensors employ all the chip real estate for sensels, but are "frozen" during the exposure. I'm not convinced that one couldn't combine the benefits of both designs, but this ship has sailed as all the research money is poured into better CMOS camera phone sensors whose technology then trickles down to compacts, dSLRs and finally MF.

Edmund
Logged
Erik Kaffehr
 

eronald

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 6638
    • My gallery on Instagram
Re: You wait years for CMOS backs to arrive, and then... Disappointment.
« Reply #262 on: January 30, 2014, 05:00:20 pm »

Yeah, but there are microlenses ...catching the light over the whole cell and piping it over the floating gate ...if a Bayer filter is a necessity, then turning it into a microlens solves several problems at once.  Come on, Erik, we all know I'm not the sharpest knife in the box but this is a case where the applications never warranted the research effort needed to solve the apparent shortcomings of the technology within the context of consumer photography.

Edmund

Hi,

Using an interline design would use half of the surface for the output buffer, so DR would be cut in half and noise increase 41%.

Best regards
Erik


« Last Edit: January 30, 2014, 05:12:21 pm by eronald »
Logged
If you appreciate my blog posts help me by following on https://instagram.com/edmundronald

ErikKaffehr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 11284
    • Echophoto
Re: You wait years for CMOS backs to arrive, and then... Disappointment.
« Reply #263 on: January 30, 2014, 05:28:03 pm »

Hi,

Problem is that half of the sensor is used for the buffer, so full well capacity is cut in half. So the sensor can hold half the electrons compared with a frame transfer design. Half the electrons give 41% more noise, Poisson statistics you know. DR is FWC/read noise so that is cut in half.

Best regards
Erik

Yeah, but there are microlenses ...catching the light over the whole cell and piping it over the floating gate ...if a Bayer filter is a necessity, then turning it into a microlens solves several problems at once.  Come on, Erik, we all know I'm not the sharpest knife in the box but this is a case where the applications never warranted the research effort needed to solve the apparent shortcomings of the technology within the context of consumer photography.

Edmund

Logged
Erik Kaffehr
 

eronald

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 6638
    • My gallery on Instagram
Re: You wait years for CMOS backs to arrive, and then... Disappointment.
« Reply #264 on: January 30, 2014, 05:38:56 pm »

If you want to "prove" that existing interline designs are more noisy and have less DR than fullframe designs, then I will give you the point; if you want to "prove" that some CCD design with adequate liveview or better efficiency is not feasible, well, I'm not so sure of that.  If you just need liveview and not full video output, you can emplace some buffer area between several sensels one buffer for multiple real sensels, giving you reduced size LV. You could even have a Fuji-type design with small less sensitive cells tesselated between the larger sensels.

In the end, who cares? It's technology. It'll all be obsolete in 10 years, and we'll have Gigapixel 6400 ISO cameraphones with all the wonderful colors of the Nikon D4 ;)

Edmund


Hi,

Problem is that half of the sensor is used for the buffer, so full well capacity is cut in half. So the sensor can hold half the electrons compared with a frame transfer design. Half the electrons give 41% more noise, Poisson statistics you know. DR is FWC/read noise so that is cut in half.

Best regards
Erik

« Last Edit: January 31, 2014, 02:48:23 pm by eronald »
Logged
If you appreciate my blog posts help me by following on https://instagram.com/edmundronald

Christoph C. Feldhaim

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2509
  • There is no rule! No - wait ...
Re: You wait years for CMOS backs to arrive, and then... Disappointment.
« Reply #265 on: January 31, 2014, 01:15:47 am »

...
In the end, who cares? It's technology. It'll all be obsolete in 10 years, and we'll have Gigapixel 6400 ISO cameraphones with all the wonderful colors of the Nikon D4 ;)
...

We'll have that as technology integrated into contacts, controlled by a wiggle of the ears ... ;)
« Last Edit: January 31, 2014, 10:31:32 am by Christoph C. Feldhaim »
Logged

BJL

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 6600
CCDs with live view capability: interline and frame transfer
« Reply #266 on: January 31, 2014, 11:15:00 am »

Actually, there are two type of CCD that can provide video output for live view and such, but both are distinctly inferior for still image quality than the full frame type of CCD used in all DMF (until this week!).

1) Frame Transfer CCDs support a global (non-progressive) shutter by having half the area of each photosite "masked", and used to transfer the signal from the other, light receiving half.  So a frame is captured by shunting the electrons to the masked region simultaneously across the whole frame, and then these are read out while the next frame accumulates. The obvious downside is a halving of well capacity, hurting dynamic range, even if microlenses succeed is directing almost all of the incoming light to the light-sensitive region. I have never heard of this type being used in still cameras.

2) Interline Transfer CCDs instead have some wiring between the light-sensitive regions that supports more rapid read-out, good enough for progressive scan video, while taking up far less than half the total photosite area. This is what most CCDs used in still cameras are or were, including all the Sony ones formerly in SLRs.  The only exceptions I know of are the full frame CCDs used in medium format and in the early Olympus Four Thirds SLRs.

Note well that Kodak/Truesense makes interline transfer CCDs as well as full frame CCDs, and comparing the spec sheets, it seems that the interline type has two disadvantages: somewhat lower full well capacity (no surprise) and somewhat higher noise floor (which does surprise me a bit). This means that amongst Kodak CCDs, the interline ones are worse than the full frame ones both at low ISO speed and at high ISO speed.  [EDIT: I see now that the noise floor disadvantage has decreased with recent Kodak/Truesense interline CCDs but they still have both a bit more noise and less well capacity.  By the way, Truesense also offers CMOS sensors for video, and they have about the same electron well capacity per photosite area and three times lower noise floor, so within Truesense, the CMOS technology beats both CCD options at both low and high ISO speeds. And this at an operation whose history is dominated by CCD development far more than CMOS!]
« Last Edit: January 31, 2014, 11:50:03 am by BJL »
Logged

Theodoros

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2454
Re: You wait years for CMOS backs to arrive, and then... Disappointment.
« Reply #267 on: January 31, 2014, 11:36:16 am »

Yeah, but there are microlenses ...catching the light over the whole cell and piping it over the floating gate ...if a Bayer filter is a necessity, then turning it into a microlens solves several problems at once.  Come on, Erik, we all know I'm not the sharpest knife in the box but this is a case where the applications never warranted the research effort needed to solve the apparent shortcomings of the technology within the context of consumer photography.

Edmund

You can't turn a Bayer pattern element into a micro lens… what will then happen is that the number of photons that are deflected (and are entering a "wrong" pixel well) will be polluted with two colours (two different wavelengths), thus, their entering speed with respect to the others will be affected and colour information per pixel will be uncontrollable by the processor.  :)
Logged

Stefan.Steib

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 470
    • HCam - Hartblei Pro Photography solutions
Re: You wait years for CMOS backs to arrive, and then... Disappointment.
« Reply #268 on: January 31, 2014, 02:18:53 pm »

(I had posted this to GetDPI before, but it fits very well here also - so sorry for the ones who have read it already-I´ll add some more comment for this one)

Interesting thread. I have been quiet for now but here comes my analysis how I see this release:

Phase One was under pressure (as are all the other MF makers) for a new CMOS technology (lets spare out the technical details here-I want to see it from a product placement and marketing point of view). Right now the DF+ Body does not make all the users happy, but it is the base platform for the actual back portfolio. Phase One themselves have called their CCD products not ideally suited for sports or action photography, as well as for low light shots for high ISO (not to be mixed up with long exposure capability with low ISO ! I also think the Pixel Plus is  giving only a second grade solution as the resolution is only a quarter of the full one, but that´s another story).
This gap had to be filled and as you can see from the presentation on the Phase website, that is what the purpose was to reach.
I don´t think this makes much sense on a tech cam, although I understand the interest of many of my colleagues for better focusing also for their tech cams.
As I have proposed here for several occasions some time ago, the only way to get to a larger (than 24x36) CMOS chip was cooperation with other companies to increase the volume, otherwise the production would have been impossible.
The common denominator for all 4 existing customers of this Sony chip is the format that Leica and Pentax are using (which draws an interesting image of the proportions per company of the projected chips).
So this chip is tailored mainly for a more compact and SLR oriented concept. Phase has used this to scale it as an extension of the existing back portfolio.

If Phase One will make this perfect, a full body with incorporation of this chip will be needed. I am nearly sure this will happen in the not so far future.
(addition: I would give the advice to Leaf/Mamiya to incorporate this technology into a Mamiya 7 body. Keep this exact formfactor, use the lenses and earn a fortune, people woud buy this like crazy. give the body a pricepoint of under 10k, that differentiates it from the back portfolio and would earn the company some easy money -and- bring up the chip numbers to enable the development of a second bigger one to replace the highres 80 Mpix sometime soon (2015 ?).
And - I would not be surprised if they have listened to their customers very carefully and are preparing a solution that will be a market leading body , maybe showing some of the concepts already used in the industrial cameras.

I welcome this policy very much and for us as HCam producers this chip is also a godsent: it opens a much greater range of available lenses of the world market, very many cover this smaller format. And 50 Mpix is enough for most of todays working tasks( though a bit small for fine art imaging).

It is the right decision, the CCD´s will hold their stake for some more years in the traditional highres range (which they do very well).
This protects the investment of existing customers, opens the market for new ones and extends the usability of the DF+ system -Plus- gives a perspective for a future devellopment.
(addition: for all those complaining about too high prices for hardware: you are comparing apples and peas here. Phase has a dealer modell which offers support and direct sales channel by high specialist trained sales personel. You have to pay for this. If someone would want to market such a complex product in a supermarket warehouse chain, he would miserably fail). Even Pentax (now Ricoh!) may have overlooked the fact that  getting  such a product to the Promarket is a timeconsuming and cost intense work.
What is described as a "paid" Brand reputation here, has been worked on for 18 years now. It was and is expensive and even Canon and Nikon are spending comparably much more to the prosection sold products than to consumer products. Actually is is a lesser profit for more work business. But the larger companies cross finance it from marketing budgets to tell consumers how good their standard bodies are. Phase / Mamiya-Leaf and Hasselblad can´t do this. Hasselblad tries to break this scheme with their Lunar/Stellar experiment.)

Well done Phase One. Pure logic and good strategy !

Greetings from Germany
Stefan
« Last Edit: January 31, 2014, 02:33:52 pm by Stefan.Steib »
Logged
Because Photography is more than Technology and "as we have done it before".

ErikKaffehr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 11284
    • Echophoto
Re: CCDs with live view capability: interline and frame transfer
« Reply #269 on: January 31, 2014, 02:33:40 pm »

Hi,

Thanks for putting things straight.

Best regards
Erik

Actually, there are two type of CCD that can provide video output for live view and such, but both are distinctly inferior for still image quality than the full frame type of CCD used in all DMF (until this week!).

1) Frame Transfer CCDs support a global (non-progressive) shutter by having half the area of each photosite "masked", and used to transfer the signal from the other, light receiving half.  So a frame is captured by shunting the electrons to the masked region simultaneously across the whole frame, and then these are read out while the next frame accumulates. The obvious downside is a halving of well capacity, hurting dynamic range, even if microlenses succeed is directing almost all of the incoming light to the light-sensitive region. I have never heard of this type being used in still cameras.

2) Interline Transfer CCDs instead have some wiring between the light-sensitive regions that supports more rapid read-out, good enough for progressive scan video, while taking up far less than half the total photosite area. This is what most CCDs used in still cameras are or were, including all the Sony ones formerly in SLRs.  The only exceptions I know of are the full frame CCDs used in medium format and in the early Olympus Four Thirds SLRs.

Note well that Kodak/Truesense makes interline transfer CCDs as well as full frame CCDs, and comparing the spec sheets, it seems that the interline type has two disadvantages: somewhat lower full well capacity (no surprise) and somewhat higher noise floor (which does surprise me a bit). This means that amongst Kodak CCDs, the interline ones are worse than the full frame ones both at low ISO speed and at high ISO speed.  [EDIT: I see now that the noise floor disadvantage has decreased with recent Kodak/Truesense interline CCDs but they still have both a bit more noise and less well capacity.  By the way, Truesense also offers CMOS sensors for video, and they have about the same electron well capacity per photosite area and three times lower noise floor, so within Truesense, the CMOS technology beats both CCD options at both low and high ISO speeds. And this at an operation whose history is dominated by CCD development far more than CMOS!]
Logged
Erik Kaffehr
 

Christoph C. Feldhaim

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2509
  • There is no rule! No - wait ...
Re: You wait years for CMOS backs to arrive, and then... Disappointment.
« Reply #270 on: January 31, 2014, 02:36:59 pm »

We can only dream.

And if they do it (M7 digicam) replace one of the green pixels in the Bayer pattern with a yellow one ....

eronald

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 6638
    • My gallery on Instagram
Re: You wait years for CMOS backs to arrive, and then... Disappointment.
« Reply #271 on: January 31, 2014, 02:51:15 pm »

T -
 dyed material for the microlens.
Edmund

BJL -
 Thanks - very nice writeup
Edmund

You can't turn a Bayer pattern element into a micro lens… what will then happen is that the number of photons that are deflected (and are entering a "wrong" pixel well) will be polluted with two colours (two different wavelengths), thus, their entering speed with respect to the others will be affected and colour information per pixel will be uncontrollable by the processor.  :)
« Last Edit: January 31, 2014, 02:55:36 pm by eronald »
Logged
If you appreciate my blog posts help me by following on https://instagram.com/edmundronald

EricWHiss

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2610
    • Rolleiflex USA
Re: You wait years for CMOS backs to arrive, and then... Disappointment.
« Reply #272 on: January 31, 2014, 03:06:19 pm »

(addition: I would give the advice to Leaf/Mamiya to incorporate this technology into a Mamiya 7 body. Keep this exact formfactor, use the lenses and earn a fortune, people woud buy this like crazy

That's a good idea Stefan!   There are a few cameras that people work hard to adapt digital backs to where no commercial solution is available and the Mamiya rangefinder is one of them and the other is the Rolleiflex TLR (and lots ask about the SL66 too).   

I'm still behind the idea of digital backs design with universal adapter plates so that the back itself can be moved between platforms by changing the adapter.
Logged
Rolleiflex USA
Pages: 1 ... 12 13 [14]   Go Up