Next generation MF sensors

[JdeV]

Do we have any indication what the next generation of MF sensors is going to look like in terms of size and resolution?

What kind of stuff might realistically get announced at Photokina 2010?

This time a couple of years ago we knew that there were going to be 50-60 Mpixel sensors coming through.
e.g. (http://luminous-landscape.com/forum/index.php?showtopic=12532)

Is there any indication from Dalsa or Kodak or anyone else where things might go?

[Christopher]

Do we have any indication what the next generation of MF sensors is going to look like in terms of size and resolution?

What kind of stuff might realistically get announced at Photokina 2010?

This time a couple of years ago we knew that there were going to be 50-60 Mpixel sensors coming through.
e.g. (http://luminous-landscape.com/forum/index.php?showtopic=12532)

Is there any indication from Dalsa or Kodak or anyone else where things might go?

^
Something between 60-80 Mp as CMOS sensor would be nice. Oh I forgot Dalsa and Kodak don't make them...

I fear the net generation will be a 80Mp sensor with more or less the same features.

ISO 50-400, no real long exposures, perhaps 0.5 more dynamic range, but overall nothing new.

Probably no live view, no real surprise. Perhaps Phase even dares to sell the same LCD again :-P

However as a P65 owner, there would need to be a big change. 60Mp are now enough for even 40x60 prints. So I think a 75 or even 90 Mp sensor would not really make a difference. What we really want and need are NEW features. Oh I forgot again, we would only need features other companies are offering for quite some years now.

[BJL]

Do we have any indication what the next generation of MF sensors is going to look like in terms of size and resolution?

What kind of stuff might realistically get announced at Photokina 2010?

Dalsa and Kodak are still rolling out their latest technology: another round of Full Frame type CCDs, this time with 6 micron cell size. Dalsa also has a new approach to microlenses that promises to eliminate the off-perpendicular sensitivity restrictions which have so far kept microlenses of the larger MF sensors. So Dalsa might be able to put microlenses on all future MF sensors, adding about one stop of sensitivity (ISO speed). RED is also talking about making a MF cameras with full 645 sized CMOS "Mysterium" sensors, and if I recall correctly also with 6 micron cell size. But RED
is the industry leader in ratio of products promised to products delivered so far.

Neither Kodak nor Dalsa is near going to a next generation beyond those 6 micron cell size products by 2010, and I have read no hints about their next generation plans. If you wish to speculate, Kodak has made one FF CCD sensor with 5.4 micron cell size, the 8MP one for Four Thirds models E-300 and E-500. Thanks to its microlenses, that sensor seems to outperform Kodak's current larger cell size sensors without microlenses in low light (anything higher than base ISO speed) but is not as good for per pixel signal to noise ratio ("per pixel dynamic range") at base ISO speed.

My predictions: as of Photokina 2010, there will be FF CCD sensors with 6 micron cell size in sensor sizes of 44x33mm, 45x30mm, 48x36mm/49x37mm, and nearly 645 (54x40mm or whatever); not much more than we have now.

[georgl]

Kodak once mentioned that they can go down to 5µm without compromising IQ - lenses that are capable to resolve such high frequencies (100lp/mm!!) with low aberration over large areas of the (quite large) frame are really rare. I don't expect any revolutions here. Interesting options to downsample to large files again or to do pixel binning, but no major breakthroughs regarding IQ.

CMOS offers interesting possibilities - but right now, when highest IQ is necessary, there's no other choice than full-frame CCDs (due to larger photosites).

Dalsa offers CMOS, RED offers marketing but they don't manufacture cameras, lenses or sensors - but they're not willing to reveal their suppliers, either. Only one thing is sure: it's 5.4µm-CMOS-sensor offers ordinary performance. Sensor-technology for 60fps creates extreme challenges and no pro would be happy to see such a quality in photography (just google-search for "red"-images with 4096x2048pixels size).

So be careful what you wish for...

[aaanorton]

Sensor-technology for 60fps creates extreme challenges and no pro would be happy to see such a quality in photography (just google-search for "red"-images with 4096x2048pixels size).

So be careful what you wish for...

I don't follow this. Would you mind clarifying for me?

[nik]

Wouldn't sensor (and A/D converter) bit depth be more useful than more mpix at this stage?

-N

[Graham Mitchell]

At the risk of sounding like a broken record... why do they insist on offering more and more megapixels? For every person here asking for more resolution there are ten asking for better high ISO performance. If Kodak and Dalsa never make the chips we want then we won't see the backs we really want either.

Still wishing for a 30MP 645 chip!

[BJL]

At the risk of sounding like a broken record... why do they insist on offering more and more megapixels?

Still wishing for a 30MP 645 chip!

At the risk of sounding like a broken record: because is nothing a 30MP 645 sensor can offer in IQ than a 60MP 645 sensor cannot, with appropriate processing (downsampling or NR processing to improve S/N ratio while reducing resolution to match that of the 30MP sensor, *if and when* the lower res. is enough), but there *is* something that a 60MP 645 sensor can offer in IQ than a 30MP 645 sensor cannot: more resolution. Within reason, a somewhat higher pixel count simply gives more options in resolution/noise level/DR trade-offs.

[Graham Mitchell]

At the risk of sounding like a broken record: because is nothing a 30MP 645 sensor can offer in IQ than a 60MP 645 sensor cannot, with appropriate processing (downsampling or NR processing to improve S/N ratio while reducing resolution to match that of the 30MP sensor, *if and when* the lower res. is enough), but there *is* something that a 60MP 645 sensor can offer in IQ than a 30MP 645 sensor cannot: more resolution. Within reason, a somewhat higher pixel count simply gives more options in resolution/noise level/DR trade-offs.

All else being equal, a larger photosite will yield a greater DR (and therefore better high ISO performance). So my point was that the sensor manufacturers should be concentrating on achieving this instead of ever more resolution.

[BernardLanguillier]

Do we have any indication what the next generation of MF sensors is going to look like in terms of size and resolution?

To my eyes, the focus for the next generation should not be size/resolution.

It should be:
- DSLR level live view in back capability,
- Reduction of silly problems like color casts,
- Price reduction.

Cheers,
Bernard

[aaanorton]

I'd like 2 frames per second.
Or for Hasselblad, 1 frame per .5 second.

[Gary Ferguson]

Do we have any indication what the next generation of MF sensors is going to look like in terms of size and resolution?

What kind of stuff might realistically get announced at Photokina 2010?

This time a couple of years ago we knew that there were going to be 50-60 Mpixel sensors coming through.
e.g. (http://luminous-landscape.com/forum/index.php?showtopic=12532)

Is there any indication from Dalsa or Kodak or anyone else where things might go?

I've been impressed with the pixel binning feature on the P65+, I'm getting clean files at ISO 800 allied to a MF look. And I'm sure there's more that could be done with this technology. If we get to a 100MP back with pixel binning delivering clean 25MP files at ISO 1600, along with a live view facility on a better screen, then I'd upgrade. Otherwise I'm more than happy to stay with the P65+.

I know this has been said many times before without ever happening, but it does seem to me that we're approaching, if not the end game, then at least a plateau in MF digital development.

[georgl]

@aaanorton
RED currently offers one camera and many people are excited about their annoncement for future cameras, even with 645-sized sensors. These cameras use fast CMOS-sensors (so they have liveview and they're basically video-cameras). But this ability has a high price regarding IQ for still photography - a camera which is capable to shot 60fps video is hardly a good MFDB-replacement... Just take a look at Kodak and their current offerings:

http://www.kodak.com/global/plugins/acroba...ductSummary.pdf
http://www.kodak.com/global/plugins/acroba...ductSummary.pdf

The first one is a state-of-the art full-frame CCD, which basically means it has a very high fill-rate (large sensitive area) for high IQ (it's theoretical DR/contrast is 1:3200) but it liveview is problematic and it's slow. The second sensor is a state-of-the-art interline CCD. It has a lower fill-rate (more circuitry besides photodiodes - lower charge capacity and about half the DR) but is capable of video. With current technology, it's always the samge game - if it's called CCD, CMOS or Mysterium... You want liveview? You want video? You want internal NR for high-iso? Say goodbye to MDFB-IQ.

Technology progresses and it will be interesting to see how Leica will implement CMOS (there are rumors that they're working on a full-frame EVIL). When their CMOS-system is prefered by photographers over their previous CCD-offerings, MDFBs will most likely follow!?

[aaanorton]

@ georgl
Yes, I'm familiar with RED's offerings and have been following their imminent release announcements. Thanks for replying, the info on Kodak chips is interesting.
I get what you're saying and really have no arguments with any of it. I just don't see RED's potential easily dismissed. Ask a fashion photographer (or any people shooter) how they would feel about a 25-65 mpxl RAW file from a camera able to capture them in or near double digits, which can store them directly to an on-board SSD allowing virtually limitless shooting, has a quality EVF and live LCD and that can mount a variety of lenses from different manufacturers. Oh, and it can also shoot better-than-cinema quality RAW video footage at 50-350 frames per second. If someone offered me that option and it provided > 90% of the IQ from current MFDB, I'd jump without hesitation. I think many would.
I posted earlier in this thread that I'd like 2 fps from MFD. A totally reasonable request. I know lots of people who want that. We want that a whole lot more than we want 20 more megapixels. And we've been asking for it for 5+ years. RED may answer this request with a totally unreasonable response and I am truly excited about it.
I've also been excited (and still am) about Leica's S2. Mostly for the speed, though 1.5 fps now seems a little timid. I like everything they are doing with that camera and look forward to what they will bring in the future (cheers the Leica for true DNG writing!). I also think Phase made some great lemonade with the Mamiya lemons they got. Good for them! But all of the discussions we are, and have been, having regarding MFD seem so stale.
This is the most exciting time in MFD I can remember since pre-P25 (able to shoot untethered!). Gonna be an interesting year.

c!

[Lawrie_Hope]

Do we have any indication what the next generation of MF sensors is going to look like in terms of size and resolution?

What kind of stuff might realistically get announced at Photokina 2010?

This time a couple of years ago we knew that there were going to be 50-60 Mpixel sensors coming through.
e.g. (http://luminous-landscape.com/forum/index.php?showtopic=12532)

Is there any indication from Dalsa or Kodak or anyone else where things might go?

Shouldn't we let lens technology catch up first?

[BJL]

All else being equal, a larger photosite will yield a greater DR (and therefore better high ISO performance).

Do we have to go over this yet again?

If you get 30 million pixels from one sensor that has 30 million photosites, and then get 30 million pixels by downsampling or such from a sensor of the same size but with 60 million photosites of similar technology, and then measure the DR of these two 30MP files, the DR is likely to be about the same. Downsampling (and other processes) can give "super pixels" with a higher S/N ratio and higher DR than the individual photosites used to form them. The simple explanation is that when the signal from several photosites is combined, the combined signal is increased about in proportion to the number of photosites uses, but the combined noise grows roughly as the square root of the noise in the individual pixels, so the S/N ratio and DR increases roughly in proportion to the square root of the number of photosites used to get the "super-pixel". A factor of 1.4 in the above example, or half a stop.

My favorite example: the "photosites" of black and white film (clumps of silver halide crystals) have pathetically low S/N ratio, but aggregation of many billions of these can give quite good DR.


Please at least stop making arguments which ignore the increase in S/N ratio and DR that downsampling (and other options like spatial averaging) can produce.


P. S. Sometimes, the downsampling or spatial averaging can be done simply by the printing technology, or by the resolution limits of the viewer's eyes. When 30MP is enough to print at a PPI figure high enough to match the viewer's resolution limits (say 300PPI for a viewing distance of 12" or more?), then 60MP give the same print size at 1.4x higher PPI, 420PPI. This now exceeds the viewer's resolving power, so the tiny pixels get blurred together: spatial averaging again, this time by the rods and cones of the viewer's eyes. This increases the S/N ratio of the signal detected by those rods and cones.

[aaanorton]

If you get 30 million pixels from one sensor that has 30 million photosites, and then get 30 million pixels by downsampling or such from a sensor of the same size but with 60 million photosites of similar technology, and then measure the DR of these two 30MP files, the DR is likely to be about the same.

How many sensor manufacturers replace a current sensor with another that has larger photosites? None, right? It's always the other way around. So to compare a 30 mpxl sensor's DR with that of a 60 mpxl upgrade to the first sensor is to ignore foto-z's qualifier of "All else being equal". The 60 mpxl back didn't just happen. People had to make it. More importantly, people were paid to make it. And since they were paid, decisions had to be made: More pixels or improve the ones we got? More pixels or better bandwidth from the chip? Instead of talking about a 30 mpxl base, let's consider a 60 mpxl position. Is the answer to IQ concerns always the doubling of resolution? Should we hope for a down sampled 120 mpxl file to deliver the 800 iso DR we want at a 60 mpxl resolution? And the 1600 iso DR we really want at 30 mpxl? And if what we really want (no kidding here) is a 30-40 mpxl file (really!), should we continue to hope for IQ to be improved only through down sampling from successively larger and larger RAW files that we have to waste storage space on and production time copying, moving, backing up and archiving? I'm sure there some people over at Seagate Technologies that like this idea!
This becomes an exercise in marketing. A smooth progression of consecutively larger numbers is easier to market and sell. Plateaus on sales brochures look like turds.

Please keep in mind that I am not arguing with your technical assessment. I'm just questioning at what cost resolution.

Thanks,
c!

[erick.boileau]

no more pixels please, stooooop . to do what ? 30 MP is enough in MF  
25 ISO will be great

there a new sensor here >> http://www.scmos.com/

[RobertJ]

Digital backs desperately need higher resolution LCD screens with live view directly on the back itself.  Then on view cameras, we don't need anymore inaccurate sliding backs or tethering.  Is the only reason they don't have live view because they use CCDs instead of CMOS?

[Graham Mitchell]

Please at least stop making arguments which ignore the increase in S/N ratio and DR that downsampling (and other options like spatial averaging) can produce.

No-one is disputing that binning helps to achieve a higher S/N ratio, but it is of limited value. My point is that the sensor makers should be concentrating on techniques to lower sources of dark current and read noise, and to increase quantum efficiency. At the end of the day, it is the S/N ratio which matters and not the technology used to achieve it.

The formula for calculating the S/N ratio in a pixel binned system is


Where
M represents the number of binned pixels
P is the incident photon flux (photons/pixel/second)
Q(e) represents the CCD quantum efficiency
t is the integration time (seconds)
D is the dark current value (electrons/pixel/second)
N® represents read noise (electrons rms/pixel)

You can see that we get between 1-2 stops more S/N ratio with 2x2 pixel binning (depending on dominance of read noise). Nice, but it's not going to make any headlines, and to get a decent resolution after binning you would need 120MP+ on the sensor. Otherwise the DSLRs are in the lead.