Is Kodak out of the sensor design business ?

[Radu Arama]

Everybody finds humor in little bits and pieces I guess, Doug, for example I am most amused about that "645 was announced in 2005 and launched in 2010" part like it was the same camera 5 years in the making (a very common thing over the net, btw). No specific knowledge I am afraid only that I know a lot about Pentax and how they operated in the past. Typically a higher end Pentax camera is active in fabrication about 18 months or less and of course this only applied to APS-C cameras in the past but I don't see them waiting over 2 years to announce a new MF camera.

The problem with 645D is that most of its internals were almost end of life when it was launched. The Pentax Prime 2 processor is a Fujitsu Milbeaut part (which Fujitsu just upgraded) and the controllers for SDXC high speeds  weren't available. Both of these parts will be available in H2 of this year and will mean much faster speeds and responsiveness for the camera and/or the ability to work with larger files. The camera body has space for a larger sensor and a larger prism and both new lenses work on full 645 film size so clearly Pentax aims for a larger sensor in the future. They need just another sensor package.

So Pentax IMO has three choices:

- Waits for a new CMOS sensor done for any customer by a semiconductor company thus pushing the new camera and risking delays;
- Uses a larger sized CCD with the new processor and fast SDXC speeds but in the process locks another cycle of 2+ years interval when the new 35mm cameras will appear;
- Keeps the current sensor, makes the camera much faster and responsive by replacing the electronics inside and they can launch it even in late 2011 or early 2012 and reduce the price even further. This will only aggravate people who are just buying the 645D now and cement the opinion of a "poor man's MF system" which is in total contradiction with their expensive and ambitious 25mm lens and I consider it very unlikely.

So my guess is that since they are able to order large quantities they already have a deal with a Japanese company to make them exclusively at very least a 44x33 mm CMOS and have two big shows in 2012 to showcase it (early the big Japanese show of the year - this year was named CP+) and in September or so Photokina. And I also guess that Pentax will have more than one MF cameras (44x33 mm and larger sensor).

Sorry for my English I am not sure I make much sense also here is 2 AM and I am quite tired.

Regards,
Radu

I love comments about the future which include the phrase "without any doubt".

Especially about when you remember the 645D was "unveiled" at PIE2005 and shipped in late 2010.

My industry read (an informed guess and not much more) is that CMOS medium format is still more than 2 years from being in people's hands. Unless you have specific knowledge you want to share...

Doug Peterson (e-mail Me)
__________________

Head of Technical Services, Capture Integration
Phase One Partner of the Year
Leaf, Leica, Cambo, Arca Swiss, Canon, Apple, Profoto, Broncolor, Eizo & More

National: 877.217.9870  |  Cell: 740.707.2183
Newsletter | RSS Feed
Buy Capture One 6 at 10% off

[Doug Peterson]

Everybody finds humor in little bits and pieces I guess, Doug, for example I am most amused about that "645 was announced in 2005 and launched in 2010" part like it was the same camera 5 years in the making (a very common thing over the net, btw). No specific knowledge I am afraid only that I know a lot about Pentax and how they operated in the past. Typically a higher end Pentax camera is active in fabrication about 18 months or less and of course this only applied to APS-C cameras in the past but I don't see them waiting over 2 years to announce a new MF camera.

The problem with 645D is that most of its internals were almost end of life when it was launched. The Pentax Prime 2 processor is a Fujitsu Milbeaut part (which Fujitsu just upgraded) and the controllers for SDXC high speeds  weren't available. Both of these parts will be available in H2 of this year and will mean much faster speeds and responsiveness for the camera and/or the ability to work with larger files. The camera body has space for a larger sensor and a larger prism and both new lenses work on full 645 film size so clearly Pentax aims for a larger sensor in the future. They need just another sensor package.

So Pentax IMO has three choices:

- Waits for a new CMOS sensor done for any customer by a semiconductor company thus pushing the new camera and risking delays;
- Uses a larger sized CCD with the new processor and fast SDXC speeds but in the process locks another cycle of 2+ years interval when the new 35mm cameras will appear;
- Keeps the current sensor, makes the camera much faster and responsive by replacing the electronics inside and they can launch it even in late 2011 or early 2012 and reduce the price even further. This will only aggravate people who are just buying the 645D now and cement the opinion of a "poor man's MF system" which is in total contradiction with their expensive and ambitious 25mm lens and I consider it very unlikely.

So my guess is that since they are able to order large quantities they already have a deal with a Japanese company to make them exclusively at very least a 44x33 mm CMOS and have two big shows in 2012 to showcase it (early the big Japanese show of the year - this year was named CP+) and in September or so Photokina. And I also guess that Pentax will have more than one MF cameras (44x33 mm and larger sensor).

Sorry for my English I am not sure I make much sense also here is 2 AM and I am quite tired.

Yeah, my words were a bit harsh. Given the language barriers and your excellent postulations above, I'm sure your use of "no doubt" phrasing could have been just as easily read as "it seems very likely to me that".

I disagree with much of your analysis but it is clearly well thought out and reasonable. My guess is they will introduce another big-brother model next year with improved electronics and a larger sensor but that it will be from an existing manufacturer, a CCD, and likely one of the designs already on the market; in other words, boring from a tech-news perspective but with proven high quality and with few hidden issues that could delay development or otherwise hold back release. Maybe they will add tethering (outside of their current target market but a way - though harder than it sounds to implement well - to open up an entire new market). They will also have to tackle the issue of multi-point AF on a larger sensor to maintain that as one of their unique selling points. Currently their AF points are relatively stuffed to the middle on the 1.3 crop sensor. On a full frame their current AF points look very centered.

Just some fun predicting. Sorry if my words earlier were a bit edgy.

[BernardLanguillier]

Having spoken several times with Pentax, I know that they consider the lack of live view on the 645D as one of its main current limitations.

My personal guess would be that the next version of the 645D will be released when they have a working 60 megapixel CMOS with a size 20% larger than the current 645D CCD. I have no idea when such a thing could happen (not before summer 2012 would be my guess) nor at what price point, but I would be surprised if the price went up much.

But this is again just a guess.

Cheers,
Bernard

[Radu Arama]

Yeah, my words were a bit harsh. Given the language barriers and your excellent postulations above, I'm sure your use of "no doubt" phrasing could have been just as easily read as "it seems very likely to me that".

I disagree with much of your analysis but it is clearly well thought out and reasonable. My guess is they will introduce another big-brother model next year with improved electronics and a larger sensor but that it will be from an existing manufacturer, a CCD, and likely one of the designs already on the market; in other words, boring from a tech-news perspective but with proven high quality and with few hidden issues that could delay development or otherwise hold back release. Maybe they will add tethering (outside of their current target market but a way - though harder than it sounds to implement well - to open up an entire new market). They will also have to tackle the issue of multi-point AF on a larger sensor to maintain that as one of their unique selling points. Currently their AF points are relatively stuffed to the middle on the 1.3 crop sensor. On a full frame their current AF points look very centered.

Just some fun predicting. Sorry if my words earlier were a bit edgy.

No reason for apologies really, I found nothing even remotely offensive!

The main differences between Pentax and the more established MF companies are IMO two:

1) I suspect that at least Phase One aims directly for a very large CMOS sensor whilst IMO Pentax will do in the first stage with a 44x33 mm one and move up as the yields improve, that could buy them time.
2) Pentax is a major client of Sony's sensor business and in general Hoya has a lot of influence in the high tech electronics of Japan. In general I believe that besides Sony there are at least a few other Japanese companies (Toshiba comes to mind) that can design and manufacture such a sensor and the only problem is that a reasonable production number makes the unit price not too expensive.

Speaking about the tethering software apparently it is coming soon.

http://www.pentax.jp/english/products/645d/feature_6.html

(although the translation is quite wrong).

Best regards,
Radu
 

[Graham Mitchell]

You can have high quality in a small pixel model.  The 5DII is not a good standard to judge from.  The Canon sensor is severely limited by read noise, unlike the new generation Sony CMOS (as in D7000/K-5).  The clean read electronics of the Sony designs have negligible read noise and give a strong advantage to high pixel-counts in this generation.  A good CMOS sensor in medium format would be something to see.

It's true that the chip's noise performance is improving but that's only part of the problem. Photon noise and limited DR become more of an issue with smaller photosites.

[ejmartin]

Naive question: All the large sensors are stitched.  What would be involved in taking a pair of FF CMOS sensors and stitching them to something of order MF size?  Is this impossible/impractical?

[hjulenissen]

All the large sensors are stitched.  

What do you mean by this?

-h

[Graham Mitchell]

Naive question: All the large sensors are stitched.  What would be involved in taking a pair of FF CMOS sensors and stitching them to something of order MF size?  Is this impossible/impractical?

The stitching happens on one piece of silicon. It is not a matter of physically joining two pieces of silicon together. Therefore you need equipment which can handle large pieces of silicon. Afaik, this is the main limitation.

[yaya]

Naive question: All the large sensors are stitched.  What would be involved in taking a pair of FF CMOS sensors and stitching them to something of order MF size?  Is this impossible/impractical?

AFAIK, none of the sensors used in professional photography are stitched. Some are made using a stepper during the exposure of the silicon but they are not madd by glueing/ welding smaller sensors.

[ejmartin]

OK thanks.

[EricWHiss]

AFAIK, none of the sensors used in professional photography are stitched. Some are made using a stepper during the exposure of the silicon but they are not madd by glueing/ welding smaller sensors.

Huh- that's weird - I just assumed that when looking at this shot from my AFi-ii-12 back (with extreme exposure and curves lift in post) that what I am looking at is 4 pieces put together.  And to be perfectly honest, I'm not sure this is normal.

[deejjjaaaa]

Naive question: All the large sensors are stitched.  What would be involved in taking a pair of FF CMOS sensors and stitching them to something of order MF size?  Is this impossible/impractical?

Emil "stitching" exists for CMOS process as well, for example = http://ericfossum.com/Publications/Papers/A%2076x77mm%2016.86%20million%20pixel%20CMOS%20APS%20image%20sensor.pdf

that is how FF CMOS sensors are done, so if there will be a market/buyer for larger CMOS sensors that can pay off the tooling they can "stitch" some more pieces...

[Graham Mitchell]

Huh- that's weird - I just assumed that when looking at this shot from my AFi-ii-12 back (with extreme exposure and curves lift in post) that what I am looking at is 4 pieces put together.  And to be perfectly honest, I'm not sure this is normal.

It is basically 4 sensors etched next to each other. The 4 'patches' have different amount of gain, it seems, but it should be matched so the stitches become invisible. How extreme was the curve? Can you notice a seam under normal conditions?

[EricWHiss]

It is basically 4 sensors etched next to each other. The 4 'patches' have different amount of gain, it seems, but it should be matched so the stitches become invisible. How extreme was the curve? Can you notice a seam under normal conditions?

At ISO 50, I really haven't noticed it, however I have seen it in properly exposed ISO 400 images though I'm not certain everyone would pick it out.  Here's a 50% crop of such an image with the line coming down vertically - you'll see it more readily if you look at the monitor from an angle. Shows up in print more readily for some reason.   This image has 0.16 stop exposure adjustment. If I had to push the image this problem could be made to look more obvious. To be fair, I have only looked at this using C1 and not LC 11.5 but I don't think the loaner aptus 12 that I tested had this issue and it certainly had less noise overall.

[Graham Mitchell]

You should probably contact your dealer about that, if you haven't already.

[ondebanks]

Huh- that's weird - I just assumed that when looking at this shot from my AFi-ii-12 back (with extreme exposure and curves lift in post) that what I am looking at is 4 pieces put together.  And to be perfectly honest, I'm not sure this is normal.

It might also be that the silicon in this case is not stitched/stepped, but rather that 4 different amplifiers read out the different quadrants, and their gains have not been perfectly matched.

Ray

[ondebanks]

You can have high quality in a small pixel model.  The 5DII is not a good standard to judge from.  The Canon sensor is severely limited by read noise, unlike the new generation Sony CMOS (as in D7000/K-5). 

Only true, of course, at low ISO. At high ISO (1600 and up) the Canon 5DII has a readnoise of ~ 2.5 electrons, which is up there with the best DSLRs, and 5 times better than the best MFD CCDs.

A good CMOS sensor in medium format would be something to see.

Amen to that! Hopefully it's just a matter of time.

Ray