MF CMOS -- now it gets interesting

[ErikKaffehr]

Hi,

Phase One says 15 stops of DR in their announcement and either 2 or 2.5 frames/s.

Best regards
Erik

Waiting with great interest to see the particulars on the new CMOS sensor.  A difference in readout strategies would spell a difference in expected results.

If this is an Exmor sensor (column-parallel, on-chip slow readout), then I'd expect thermal noise to be an issue at handheld shutter speeds and high gain settings, especially after the use of live view.  Full readout is relatively slow despite parallelism, and frame rate likely to top out at 2.5 fps.  But engineering dynamic range at base ISO would likely exceed 15 stops by DxO metrics.

If this resembles the D4 sensor (on-chip fast readout, multiplexed column readouts, 24 channels), then there would be some limit on dynamic range at base ISO, but that limit would not be severe.  You have the benefit of on-chip readout, but the noise issues surrounding fast readout.  Thermal noise, however, would be very low.  (13.5-14 stops DR)

If this resembles the D3/D700/D3s sensor (fast readout, off-chip high quality A-D, 12 channels), similar to the earlier Canon sensors, then dynamic range is severely limited at base ISO.  Optimal read noise levels come around ISO1600.  The A-D is high quality, but low-level signals need to be routed off the chip and quickly.  Thermal noise would be extremely low.  (13 stops of DR)

[ErikKaffehr]

Hi,

This is my source for that information:

http://www.luminous-landscape.com/forum/index.php?topic=82870.msg679019#msg679019

Best regards
Erik

P65+ doesn't have any micro lenses on Erik, it would thus surprise me if P260 has any… I don't know about the 80mp sensor, but I guess it's the same with it… Traditionally, Dalsa avoided micro lenses, it was Kodak that had such versions of their sensors and those where mostly of 33x44 size.

[LKaven]

Phase One says 15 stops of DR in their announcement and either 2 or 2.5 frames/s.

If this is Sony, selling the same sensor to PhaseOne and Hasselblad, they might make a dime off it.  In that case, I wonder if any of the big Japanese players are going to enter this market.  I wonder if Pentax will bite on this for the next generation 645D.  This might be a chip you can get for $1000-1500 in quantity after a while..

[BernardLanguillier]

If this is Sony, selling the same sensor to PhaseOne and Hasselblad, they might make a dime off it.  In that case, I wonder if any of the big Japanese players are going to enter this market.  I wonder if Pentax will bite on this for the next generation 645D.  This might be a chip you can get for $1000-1500 in quantity after a while..

I heard from a Pentax top representative in Japan more than 2 years ago that they saw the lack of live view on the 645D V1 as one of its major limitations due to the non availability of MFs sensors offering it.

I would be extremely surprised if they released a VII without live view, meaning without a CMOS.

Since they are likely to sell at a much lower price point, they will probably purchase most of the volume of the sensor if it comes from Japan and are therefore the ones most likely to drive the design.

My understanding is that the latest version of the Pentax 25mm lens does not cover full frame but I am not sure how large a sensor it can accommodate.

The answer to this question may tell us how large the 50mp CMOS is... again assuming that it is coming from Japan.

I would guess that Pentax would use the CP+ show in Yokohama in 2 weeks to make such an announcement, which may be the reason why Hassy and Phaseone have decided to go public this week.

We will know in a few hours/days I guess.  

Cheers,
Bernard

[BJL]

Dalsa now has microlenses with excellent off-perpendicular response, use for example in its 60MP sensor:
http://www.teledynedalsa.com/imaging/products/sensors/area-scan/ftf9168c/
From the graphs in the spec sheets at that link, Dalsa's approach completely eliminates the problem of "micro-lens vignetting": poor sensitivity to light striking the sensor at a highly off-perpendicular angle. QE of about 30% is not quite as good as with Kodak's micro lenses sensors (about 40%) but better than MF sensors without micro lenses (about 20%); that seems a good design compromise for MF needs.

P. S. Kodak has offered 50MP 49x37mm sensors and Dalsa has offered 48MP, 48x36mm in the FTF6080C, so sizes around there are an option I suppose, whoever makes the sensor.

P. P. S. What "Phase One announcement"?  All I have seen is a leak of the name IQ250 and the pixel count 8280 x 6208 plus rumors of other specs, at http://photorumors.com/2014/01/21/phaseone-iq250-coming-soon-with-a-new-50mp-cmos-sensor/
No solid evidence there even of it being CMOS.

[Vladimirovich]

From the graphs in the spec sheets at that link, Dalsa's approach completely eliminates the problem of "micro-lens vignetting": poor sensitivity to light striking the sensor at a highly off-perpendicular angle.

does that graph really shows that it "completely eliminates" ?

[LKaven]

P. P. S. What "Phase One announcement"?  All I have seen is a leak of the name IQ250 and the pixel count 8280 x 6208 plus rumors of other specs, at http://photorumors.com/2014/01/21/phaseone-iq250-coming-soon-with-a-new-50mp-cmos-sensor/
No solid evidence there even of it being CMOS.

According to the rumors themselves, we'll only have to wait until tomorrow for an official announcement...or none. 

In favor of this rumor is the idea that anyone willing to put up enough capital to design and manufacture a medium format CMOS sensor ought to have a number of clients signed up for it.  Surely (in this rosy scenario) if Hasselblad has signed up for a CMOS run, there must be others.  I can't imagine Hasselblad selling any more than 3-4,000 of these things in the first year, and that's a stretch, especially with a CCD version of the same camera also in the catalog. 

[eronald]

According to the rumors themselves, we'll only have to wait until tomorrow for an official announcement...or none. 

In favor of this rumor is the idea that anyone willing to put up enough capital to design and manufacture a medium format CMOS sensor ought to have a number of clients signed up for it.  Surely (in this rosy scenario) if Hasselblad has signed up for a CMOS run, there must be others.  I can't imagine Hasselblad selling any more than 3-4,000 of these things in the first year, and that's a stretch, especially with a CCD version of the same camera also in the catalog. 

Why do you assume a run? If it is Sony, then they can probably make one of these on a wafer chock full of other smaller junk.

Edmund

[LKaven]

Why do you assume a run? If it is Sony, then they can probably make one of these on a wafer chock full of other smaller junk.

I might have said that Hasselblad "subscribed" to a run.  Whichever is the case, I'd say that they have other subscribers, whoever they could get to underwrite the venture with orders.  

BTW, SonyAlphaRumors seems sure that this is a Sony sensor, for whatever that's worth.

http://www.sonyalpharumors.com/hasselblad-to-launch-50-megapixel-sony-cmos-sensor-based-medium-format-camera/

[Radu Arama]

Pentax had a 645D2 with precisely a 50 Mp CMOS sensor scheduled to launch in March 2013. Whether the sensor wasn't ready because of its complexity or the earthquake/tsunami play a part in its delay I have no idea. All I know is that under Ricoh rule the decision was made that this camera will go back to the design board for some major overhaul. Since in late 2013 Pentax switched from Prime 2 to Prime 3 as image processor, introduced new AF and new metering modules and much faster write speeds on SDXC cards probably the result will be worth the wait.

For sure it will be available in this year and judging by the Ricoh policy to not overcharge a lot we can expect a price of maximum 1M Yen, maybe less.

L.E. Of course there is also the possibility that Ricoh was forced to postpone the camera because the sensor wasn't available at the said date and took the opportunity to update it on all fronts.

[torger]

So, Phase One steals the show (for now) with the IQ250 official and all, Sony 44x33mm sensor 5.3um pixel pitch, ISO 100-6400, 14 stops DR at base ISO. I'm assuming Hasselblad will be using the same sensor.

[michael]

The sensor is the same.

Michael

[Bart_van_der_Wolf]

The sensor is the same.

Hi Michael,

That would suggest that the Bayer CFA quality is also identical for all those who source this sensor from Sony. Or do you know if there are some specifications that e.g. PhaseOne could have unique, compared to the other takers?

Cheers,
Bart

[BernardLanguillier]

Pentax had a 645D2 with precisely a 50 Mp CMOS sensor scheduled to launch in March 2013. Whether the sensor wasn't ready because of its complexity or the earthquake/tsunami play a part in its delay I have no idea. All I know is that under Ricoh rule the decision was made that this camera will go back to the design board for some major overhaul. Since in late 2013 Pentax switched from Prime 2 to Prime 3 as image processor, introduced new AF and new metering modules and much faster write speeds on SDXC cards probably the result will be worth the wait.

For sure it will be available in this year and judging by the Ricoh policy to not overcharge a lot we can expect a price of maximum 1M Yen.

Yep, considering that Pentax managed to get cleaner high ISO than anybody else with the current 645D, we can expect them to do well with the II as well.

The only questions is when it will be announced and whether they'll stick to an integrated body alone or would add a standalone back too.

We don't know today how well the CMOS sensor would perform on view cameras though.

Cheers,
Bernard

[torger]

We don't know today how well the CMOS sensor would perform on view cameras though.

The guesses I've heard is that it will perform about as good/bad as an IQ180, but be prepared for even a bit worse. Ie you need retrofocus glass for wide to moderate focal lengths.

I think the unfortunate truth is that tech cam lens design is going away, and CMOS will be what killed it for good . We'll see wider and wider aperture (for better live view performance) and more and more retrofocus. If you think Zeiss Otus is large and heavy, just wait for how future tech cam lens designs will look. And noone is going to care except nostalgic persons like me that liked the distortion-free small simple and light large format designs. Unless some company actually makes a sensor design specifically for tech cams, but that's unlikely. Leica might have some interest through their M camera and ownership of Sinar. I don't think Sony would be into it though, they're more into volume and tech cams are just too narrow.

I also suspect that there actually are no known solutions for the problem currently. As far as I know the anatomy of CMOS is that light capturing sits deeper than a CCD and thus light must come more head on. Maybe some lightpipe design could solve the problem, I don't know. Maybe it's not that hard to solve, but there need to be some interest in actually doing it. And the Phase One launch is showing a mountain bike in the air, not an interior photographer focusing a Schneider 35XL using live view

[Vladimirovich]

Yep, considering that Pentax managed to get cleaner high ISO than anybody else with the current 645D

with Sony sensors both Nikon and Pentax (Ricoh) had a good experience before... but Pentax used to start an easily detectable (see DxO graphs) NR (you can't switch that off, no) after something like ISO1600 (with APS-C sensors)

[BJL]

Finally, a wild guess: what we are about to see revealed is one or more cameras using a Sony CMOS sensor in a near 645 size like 54x40mm and so with pixel pitch slightly larger than 6 microns -- because the target for these is those MF users who want features like better low-light performance, good DR and more usable Live View, and minimal cropping of their prime lenses relative to 645 film, and for whom 50MP is more than enough. Look for DR to be hyped in the marketing!

So my punditry is about as bad as all those professional tech analysts who have predicted the failure of every new Apple product in the last few years!

Note though that this is not Phase One responding to Hasselblad's announcement; clearly Hasselblad got wind of this announcement and immediate availability (Michael and Ken knew it was coming) and so made its pre-emptive announcement to avoid losing all the spotlight to Phase One.


Anyway, it is good news that a competent CMOS sensor maker is now willing and able to supply all MF makers: CMOS vs CCD can now fight it out on a level playing field. In particular, 44x33mm is a perfect fit for a new Pentax model.

[BJL]

does that graph really shows that it "completely eliminates" ?

Compared to the graphs for Kodak/Truesense sensors without micro-lenses like the KAF-50100 , the micro-lensed Dalsa has equally low rate of falloff with angle as far as I can see: down 20% from maximum at 30 degrees off-perpendicular, for example.

The new question is what Sony has done about micro-lenses and related vignetting in its new 44x33mm CMOS sensor!

[Radu Arama]

"Never" is not a word I don't use often but I think the chances that Pentax will make a digital back ever are close to 0%. They don't have a legacy film camera that could take a back. The 645D is known to have a magnesium shell but not so many know about it's internal aluminum chassis and it's role: to keep the camera rigid enough so it won't be affected by temperature variations that could lead to errors in autofocusing; for a camera that is advertise to work in cold conditions it is a must.

Pentax doesn't believe in a modular design for that reason, their philosophy is to make cameras as compact and rugged as possible. The current model has plenty of spare room inside, three parallel pcb's between the sensor and the lcd. Combine:

1) the fact that the sensor is the same size;
2) CMOS sensor will be less prone to heat so it needs a smaller radiator and less room around it;
3) most likely Pentax engineers will reduce the size of the pcbs, or their number, or both;
4) current LCDs are thinner than 2009 design;
5) there is no grip to protect so the camera could be of any size without worrying about compatibility

therefore I expect a much smaller 645D2 camera than the current model and with a larger pentaprism.

Regards,
Radu   

Yep, considering that Pentax managed to get cleaner high ISO than anybody else with the current 645D, we can expect them to do well with the II as well.

The only questions is when it will be announced and whether they'll stick to an integrated body alone or would add a standalone back too.

We don't know today how well the CMOS sensor would perform on view cameras though.

Cheers,
Bernard

[tsjanik]

...........................
5) there is no grip to protect so the camera could be of any size without worrying about compatibility

therefore I expect a much smaller 645D2 camera than the current model and with a larger pentaprism.

Regards,
Radu   

You always have interesting insights into Pentax Radu.  Can you explain #5 in your post please?  No grip to protect?

Tom