M9

[KevinA]

Closely followed by not citing a source for the information on the non-existent camera

If I'm correct the source was a French camera dealer that put it up on their site, then quickly removed it, there has also been talk of a German site that was taking pre-orders for Sept.
The French site wether it was a deliberate mistake or genuine mistake we will never know, it did by all accounts look promising. I hope it does turn out to be true and I hope the M9 is everything a Leica "M" should be and fills a bigger niche than the M8. It will have it's compromises and knockers that don't see the point of a RF or a range of manual focus fast lenses that can be used wide open. It will be expensive for its paper spec, which to most will mean it is overpriced, others will just say it's being undervalued by those that "don't get it" and so the fors and againsts will disappear up their own backsides arguing it's worth again. No doubt others will just get on with it and not care about its weaknesses and exploit its good points.
I have a romantic notion I would buy one but I know in reality there will be other things to spend my money on.

Kevin.

[georgl]

I think live-view would be really great for a RF-camera like the M9 but it's unlikely that we will see this technology in the M9, maybe in future models (they already said that they won't come up with a regular DSLR anymore but an EVF-solution). The reason is simple: the fill-rate and overall IQ of todays full-frame CCDs (the new 6µm-generation from Kodak and Dalsa seem close)  is unsurpassed (if you don't believe it, just take a look at some P40+/P65+/H3DII-50-samples). The downside is the low speed (about 120MB/s is the maximum data output of their four channel sensor which results in 1.5fps @ 37,5MP and a little over 2fps @ 24MP) and the power consumption/heat that makes live-view really difficult.

[pschefz]

There's certainly the possibility that it could be a rather nice camera for someone like me who prefers the look of a phase back to a cmos based dslr. Whether it will pan out we will see but it's very interesting to me as I love the look I get from my P21 but I hate hate hate the weight and ergonomics of it all. (and also I always shoot equivalent of a 28mm lens which I also hate Mamiya's offering).

that is exactly what i love about the m8....the files really remind me of the P20...great back but i much prefer the much smaller size/weight and handling....and actually focus is easier on the m8 then on anything with a P20....and of course making that a P21 is even better!

[TMARK]

If an M9 has the IQ of the M8, but full frame 35mm, I'll buy it and use it in place of my RZ/Aptus 54s for digital editorial.

[John Camp]

On 1, the idea that a company would try to help one product by holding down the performance of another makes no sense in a competitive market place: if a company can make a product that takes sales from another of its products but abstains, other companies will take those sales instead by not hobbling their competing products, so the company will be even worse off. Leica has to give such a camera the best sensor it can get hold of.

On 2, you do exactly what I warned against: rehashing the misguided dogma that having slightly fewer, slightly bigger pixels on the same sensor size will give better image quality at high ISO speeds when judged sanely, by looking at equal sized prints. Anyway, if Leica stays with Kodak Full Frame type CCD's, it will not be in the hunt for high ISO performance against today's best CMOS (or MOS) sensors. It would do better to showcase the quality of its lenses by offering high sensor resolution.

P. S. Bernard's comment on 18MP being more than one can likely make much use of in hand-held rangerfinder style photography makes me think that this "street photography" will naturally move to new, smaller formats with more moderate resolutions, say 10 or 12 MP, just as Leica moved photography to the new, smaller 35mm format long ago. But Leica has lens investments to protect ...

(1) What are you talking about? Haven't you heard of the D300, the D90, the D70, the D5000? All of those Nikons (and a similar string of Canons) are helping some Nikon products by holding down the performance of others. Same with almost every expensive product you can think of, from watches to cars. You've heard that there is a Boxter and a Carrera and a Turbo? If a 24mp CCD camera selling in the neighborhood of $10,000 and using the best glass in the world was available, and challenged the quality of what is going to be the most expensive system in the world -- probably $50,000 for starters -- don't you think there might be some cannibalization, as happened when the D700 came out and D3 sales dropped?
 
(2) There's a big gap between 24mp and 18mp, and generally, full-frame lower-megapixel cameras (the 1DIII, the D3) have better high ISO response than the high mp ff products. My point being that giving up some possible resolution for higher ISO response would be a decent trade, if that trade was available.

(3) Bernard always knows what he's talking about, but he's also a guy who makes very careful, studied landscape shots. A street photographer may take 200 shots in two hours. Something a little out of focus? That's okay, it's part of the aesthetic. Some of them don't work at all? That's okay, toss them, which is what happens with most street photos. And some of them will be exquisitely focused. I mean, how many of HCBs shots do you know? He was a guy who shot for decades, one of the best  photogrpahers in the world, and I doubt that anyone but a student of his would recognize the equivalent of more than one shot a year, or so. That's street photography. You don't get a lot  of keepers, and that's fine. An 18mp M with D3 ISO and a 75mm lux would be a dream. If you like Ms.

[Slough]

Regarding point 1, companies design products to suit markets, which includes product differentiation. You don't want low end products stealing from high end ones. Years ago I worked for a UK company that made expansion cards for its computers. They had 16KB and 32KB units. Both had the same memory chip inside, but the 16KB one had some pins disabled to restrict the accessible size to 16KB. Why do that? Well, they needed a range of products to suit the market, and it was more profitable to sell two units, one priced low, one priced high, than one 32KB unit priced low, or priced high and alienating users who want 16KB. Using the same chip in both reduced costs as they could increase the purchase orders and hence bulk discounts on the memory chips, and simplified design as there was really only one chip. Nikon could easily add proper MLU to the D90. So why don't they? Product differentiation.

[BJL]

(1) What are you talking about? Haven't you heard of the D300, the D90, the D70, the D5000?
(2) ... generally, full-frame lower-megapixel cameras (the 1DIII, the D3) have better high ISO response than the high mp ff products.

On (1) Those are differentiated on production cost as well as performance, which makes perfect sense: models that cost less to make than other models in exchange for lower performance level can profitably be sold at a lower price, serving a different market segment.

What makes no sense is a design choice to impair performance of a new product without reducing the cost of making the product, at least not when a rival can make a competing product for similar cost without that performance impairment. This often actually increases costs: once there is a market for the higher resolution option, designing just it is cheaper than designing both it and the lower resolution option. And Kodak has a current 6 micron cell design.

Sensors of the same size with fewer, bigger photosites are not significantly cheaper to make, so opting for lower resolution in a new sensor design for the sake of deliberately lower IQ makes no business sense. New sensors with more, smaller pixels do sell at a higher price, as in the Sony A550 vs A500 or A380 vs A330, but that is a matter of the market putting a higher value on the new higher resolution sensor, not higher production costs. Using lower res. sensors in the A230 and A500 does make those models cheaper for Sony to make because they are old designs, with production lines already in place, so for a while it is cheaper to keep using those lines than retool for the news sensors. But the cost/benefit analysis is quite different when one talks about designing and putting into production a new sensor with lower resolution. The only time that new designs have lower resolution is for the sake of high frame rates models like the EOS-1D series, D1, etc.


Didn't we go through this before with all the claims and arguments that the Canon 5DMkII would have a new sensor of lower resolution than the 1DsMIII to avoid stealing sales from the latter?


On (2), do I have to say it a third time? I have seen no good evidence or arguments for the frequent claim that fewer, bigger pixels on the same size of sensor using the same basic sensor technology gives better IQ when images are compared fairly, displayed at equal size. Calculations of per pixel S/N ratios and related DR calculations do not predict the results of such comparisons because more pixels allows printing at higher PPI which reduces visible noise levels, potentially offsetting lower per pixel S/N measurements. I wonder how the Nikon D700 and D3X compare at equal print size and equally high ISO, as an example of two equal sized sensors using the same basic technology but different pixel counts.

[mas55101]

On (2), do I have to say it a third time? I have seen no good evidence or arguments for the frequent claim that fewer, bigger pixels on the same size of sensor using the same basic sensor technology gives better IQ when images are compared fairly, displayed at equal size. Calculations of per pixel S/N ratios and related DR calculations do not predict the results of such comparisons because more pixels allows printing at higher PPI which reduces visible noise levels, potentially offsetting lower per pixel S/N measurements. I wonder how the Nikon D700 and D3X compare at equal print size and equally high ISO, as an example of two equal sized sensors using the same basic technology but different pixel counts.
[/quote]
I wouldn't say it a third time. I wouldn't have said it the first time.  That you have seen no evidence for a proven fact does not mean it is not a fact.  I have seen the differences in different sensors.  Having used a D700, p21, and 5d Mk2, I am in a position to corroborate the current explanation of the relationship of pixel density to noise and pixel number to print resolution.

Dpreview has a good explanation as well as listings of pixel densities on new sensors.  The 700 is 1.4.  Do the math; look at 40x60 prints; shoot all at 800 and higher.  You will see.

MAS

[douglasf13]

On (2), do I have to say it a third time? I have seen no good evidence or arguments for the frequent claim that fewer, bigger pixels on the same size of sensor using the same basic sensor technology gives better IQ when images are compared fairly, displayed at equal size. Calculations of per pixel S/N ratios and related DR calculations do not predict the results of such comparisons because more pixels allows printing at higher PPI which reduces visible noise levels, potentially offsetting lower per pixel S/N measurements. I wonder how the Nikon D700 and D3X compare at equal print size and equally high ISO, as an example of two equal sized sensors using the same basic technology but different pixel counts.

I wouldn't say it a third time. I wouldn't have said it the first time.  That you have seen no evidence for a proven fact does not mean it is not a fact.  I have seen the differences in different sensors.  Having used a D700, p21, and 5d Mk2, I am in a position to corroborate the current explanation of the relationship of pixel density to noise and pixel number to print resolution.

Dpreview has a good explanation as well as listings of pixel densities on new sensors.  The 700 is 1.4.  Do the math; look at 40x60 prints; shoot all at 800 and higher.  You will see.

MAS

  Interesting, MAS. Are you saying that if you shoot the same scene with the D700 and 5Dii, and make 8x10, 20x30, and 40x60 prints of the shots, the D700 has considerably less noise in the pictures at all sizes?  How does the detail of the D700 stack up to the 5Dii at 40x60?

[mas55101]

Interesting, MAS. Are you saying that if you shoot the same scene with the D700 and 5Dii, and make 8x10, 20x30, and 40x60 prints of the shots, the D700 has considerably less noise in the pictures at all sizes?  How does the detail of the D700 stack up to the 5Dii at 40x60?

At high ISO, yes-substantially less noise with D700.  At 40x60, the 5D MK2 is noticeably more detailed.

[douglasf13]

At high ISO, yes-substantially less noise with D700.  At 40x60, the 5D MK2 is noticeably more detailed.

  Thanks, that's the first I've heard that.  Most say the D700 is slightly better, but not substantially.  Are you talking high ISO like 3200, or like 12,800?

[BJL]

That you have seen no evidence for a proven fact does not mean it is not a fact.

Of course not, but when people keep stating something as fact without any evidence, or with bogus evidence (like per pixel S/N measurements) I think skepticism is justified.

I have seen the differences in different sensors.  Having used a D700, p21, and 5d Mk2, I am in a position to corroborate the current explanation of the relationship of pixel density to noise and pixel number to print resolution.

The P21 has a far larger sensor and very different pixel technology, so the most useful comparisons I can imagine from those would be equal sized prints at equally high ISO from the D700 and 5D MkII, or maybe JPEG conversions to the same pixel count from both. And what I have read reported is that the 5DMkII does as well or better in that comparison; do you see differently?

Dpreview has a good explanation ...

DPReview has the usual bogus explanation based on per pixel performance measurements, ignoring the effect of higher PPI, or other options for equalizing resolution by trading away resolution for noise reduction n post-processing a higher pixel count image.


P. S. Correction/clarification. I realize that I have expressed myself poorly in a couple of posts in this thread. I was assuming without stating it that comparisons are made between lower pixel count and higher pixel count sensors at a size that "makes sense" for the lower pixel count option: a size at which its images look sharp, so that the extra pixels of the other sensor option do not add noticeably to resolution/detail//sharpness. Maybe 300PPI for the lower pixel count image, depending on how good your eyes are.

In that situation, the higher pixel count image is being printed at a PPI so high that its additional high spatial frequency detail and noise are not resolved by the viewer's eyes but get blurred or "dithered", rendering invisible the highest frequency noise. This is where experiments by John Sheehy and other show that the extra noise of the higher pixel count image basically disappears, roughly equalizing visible noise (and visible resolution).

So in situations like this when you do not want or need the extra resolution, you can get rid of the extra high frequency noise that comes with it. A more reliable method is a bit of low pass filtering of the image, perhaps as part of the interpolation ("demosaicing") from RAW to JPEG, perhaps by converting direcvt from RAW to JPEGs' of the same pixel count as the lower pixel count sensor.  This can discard the spatial details that are on a scale finer than the lower pixel count sensor gives and discard with it the high frequency noise that goes with it. DPReview would probably then complain that "at high ISOs, resolution is lost to NR smearing"; very strange since the solution advocated by DPReview of lower pixel counts causes an equivalent loss of resolution at all ISO speeds, not just high ones.

[BJL]

This sounds like wishful thinking, particularly the 18MP, which is too low to be credible for 24x36mm format these days. ...

But it seems I was wrong! Unless the PDF brochure http://litpixel.com/m9_brochure_english_RZ_low.1.pdf is an elaborate hoax, Leica and Kodak are using Kodak's previous generation 6.8 micron pixel technology for the M9 sensor, basically the same as in the M8 and way back to the Olympus E-1.

[telyt]

But it seems I was wrong! Unless the PDF brochure http://litpixel.com/m9_brochure_english_RZ_low.1.pdf is an elaborate hoax, Leica and Kodak are using Kodak's previous generation 6.8 micron pixel technology for the M9 sensor, basically the same as in the M8 and way back to the Olympus E-1.

It's an all-new sensor with several new technologies that Kodak developed for Leica's S2.  If the DMR is any indication, the M9's 18MP with no AA filter will be about equivalent to a 27MP CMOS sensor with AA filter.

[marcs]

I was going to weigh in here, but georgl beat me too it.  These are important considerations.

The 18 megapixels are indeed strange, because they match a 6,8µm pixel-pitch. But the new CCD-architecture by Kodak has a 6µm pixel-pitch which would result in 24MP. Always remember: we're talking about pixel-pitch, not pixel size! As I was told by Kodak, the actual sensitive area of the pixels has barely changed over the last architecture-generations (9µm, 6,8µm, 6µm), only the gaps between those areas became smaller!
The 6,8µm-architecture was introduced in 2004 and the new generation (introduced with the H3DII-50 and now S2 with microlenses) seems to be superior to the previous generation regarding noise and DR (just like DALSA).

There are lots of strange comparisons Digital vs. Film, most of them with horrible film-processing. Some claim a 35mm Velvia is similar to their 11MP 1ds, some compare it to their 6MP D100 and other claim 3MP as digital aquivalent...

Velvia has 80lp/mm at a realistic contrast of 1:1,6 which would result in about 24MP at 35mm. But film also has grain and contrast slowly decreases with higher frequencies and additionally has to be multiplied with the scanner-MTF. But from my own experience, clearly visible >10MP are not a serious problem with well-processed 35mm-scans (which would result in about 25MP in 645 and 45MP in 6x7).

I've tried it carefully with my M8, the files are clean but don't contain more information than a properly done, oversampled, denoised and sharpened professional scan. Occassionally I still like to put some Velvia/Ektar (Dynamic Range of up to 15 stops!) in my MP and pay for the scan, but mostly I stay with B&W (Imagelink HQ, only 25ASA but bloody sharp!)

Here's a properly scanned 6x7 slide (even when f16 might cause slight diffraction and Provia isn't as sharp as Velvia!): http://www.rockgarden.net/download/60MP_from_6x7/

Try to get that amount of detail with any 35mm-digital-solution...

[BJL]

It's an all-new sensor with several new technologies that Kodak developed for Leica's S2.

We know it is new, as Kodak had no previous 24x36mm, 6.8 micron pixel pitch sensor, but what specifically do we know about the use of "new technologies ... developed for Leica's S2."? Improved micro-lenses?

[telyt]

We know it is new, as Kodak had no previous 24x36mm, 6.8 micron pixel pitch sensor, but what specifically do we know about the use of "new technologies ... developed for Leica's S2."? Improved micro-lenses?

At this point it's speculation, but informed speculation.  A couple of technologies Kodak developed for the S2 and will also benefit an M camera are:

-> the placement of the IR filter between the photo sites and the microlenses.  This has the potential of reducing or eliminating the need for an additional IR filter.

-> reduced sensitivity to noise without the excessive image smoothening (the 'plastic' look) of previous noise reduction technology, giving an additional stop of high-ISO performance without a noise or 'plastic' penalty.

[teddillard]

doods, this is SO last week news.  

now THIS.  THIS is news.  




(...hope you all have a great holiday!)

[henrikfoto]

Does anybody have an idea about what the price on the M9 will be? Maybe about 15.000 US$?

[BJL]

At this point it's speculation, but informed speculation.  A couple of technologies Kodak developed for the S2 and will also benefit an M camera are:
-> the placement of the IR filter between the photo sites and the microlenses.  This has the potential of reducing or eliminating the need for an additional IR filter.

For sure: that is just fixing a problem unique to the M8 sensor and not suffered by the sensors for the E-1 or the DMR back, which do have IR filters. So zero progress over the six year old E-1 sensor on that count.

-> reduced sensitivity to noise without the excessive image smoothening (the 'plastic' look) of previous noise reduction technology, giving an additional stop of high-ISO performance without a noise or 'plastic' penalty.

I see nothing "informed" about that speculation, and anyway it is about improvement in processing after the sensor, not the sensor itself. The only place there has been a jump of one-stop in high ISO performance without a resolution loss to extra NR processing is when micro-lenses have been added to a sensor type that previously did not have them. But the E-1, M8 and R digital back all have micro-lenses already; all Kodak can do is incrementally improve them, as Kodak already did once from E-1 to DMR and M8.

P.S. Here is everything I can see from Leica itself, from page 34 of the leaked M9 brochure under "Optimized Sensor":
1. "... The special layout of the microlenses found in the M9 sensor makes it tolerant of oblique light rays impinging on its surface, thus assuring uniform exposure and extreme sharpness from corner to corner in every image.
2. A newly developed sensor filter ensures the suppression of undesirable infrared light.
3. The conscious decision to do without a moiré filter, a cause of image deterioration through loss of resolution, ensures maximum resolution of fine detail.

Item 2 is a weaselly way of acknowledging the error in omitting the IR filter from the M8; it is nothing new as earlier sensors like that for the DMR back and Olympus E-1 has such filters. The brochure's language suggesting that this is a technological innovation smells of PR hype.

Item 1 refers to the great Kodak innovation of offset microlenses ... already present on the M8 and in Kodak's 39MP 44x33mm MF sensor. So maybe the offset microlenses have been improved for the M9 to handle to more severely oblique incidence at the corners of the larger sensor, but given the hype in item 2, it also seems quite possible that this does not represent any new progress over the M8. I have to wait at least a few days more, I suppose.

Item 3 could be a good idea, but it is nothing new.


And nothing I can see anywhere refers to adopting sensor technologies developed for the S2, despite what I have read in various forum posts.