M9 Sensor Qualities

[Sparrow]

Can you not at least allow it as a "gentleman's third"? After all, they seem quite popular in the political classes.

...Mike

I'm with Socrates it's all about the plot, and as so much this thread is directly related to this release we can give it a "young gentleman's first"

 

[semilog]

Consider two very different CFAs. One is made to minimize design, material and production costs. The other is made to provide spatial and frequency information that directly corresponds to the Bayer-reconstruction mathematical model. Is it unreasonable to assume a design where each micro-lens collects as much light as possible (regardless of it's physical location in the array) and filters out all frequencies except the frequency the Bayer model, costs more to design, source and manufacture than a design with less ambitious goals?

Actually, yes. In 2014, it is unreasonable to make that assumption. And we might note that (1) it was Sony who invented the micro-lens array; (2) the MLA and Beyer filter are added at the fab, so the company with the best fab tech will likely be in the lead with the Beyer and MLA filters as well.

One other point. It is not merely the MLA and Beyer filters that determine color response. The underlying photosite architecture also has intrinsic on- and off-axis spectral responses that can be tuned by the sensor designer. People who do scientific imaging know that Sony has been at the forefront in all of these areas for 15-20 years. Even apples-to-apples: in the 1990s, when Sony made mostly CCDs (vs. CMOS sensors), the Sony sensors generally outperformed Kodak CCDs, often by large margins.

The M9 is a good camera with a competent CCD sensor. Indeed, I still use and like an Olympus DSLR with a very similar Kodak CCD sensor (I keep it because the 12-60 Zuiko is a magnificent lens). But Kodak CCDs are not magic, and the cameras that use them are not special by virtue of their Kodak sensors.

 

[Aristophanes]

Let's consider a sensor has two roles.

1. convert light energy into electrical charge
2. model the spatial and frequency aspects of the light exiting the lens rear element.

The electronics are photodiodes controlled by transistors. Their transistors role is to make sure things happen in the proper order and to provide a low-level of amplification. The photodiodes convert the light waves' electrical amplitudes into electrical charge. There is no fundimental difference between the photodiodes in a CCD sensor and a CMOS sensor. The electrical charge in the photodiodes is just energy stored in a semiconductor potential well. Instead the difference between CCD and CMOS performance has to do with the manufacturing processes between the CCD and CMOS integrated circuits.

The manufacturing and design process does affect the devices' efficiency. CMOS circuits are more efficient. By any objective measure the analog signal-to-noise ratio and dynamic range of the Kodak technology used in the M8/M9 series is inferior to CMOS technology available in a parallel time line. This is not Leica hating, it is not inflammatory. It is simply a state of nature based on competent, honest measurement. I don't make the news here, I just report it.

Which brings us to the the sensor's second role.

The evaluation of the raw-data SNR and DR is trivial because it is objective. The aethestic value of the rendered image, is another story altogether. On one hand, aesthetic judgements are subjective. Can arguments about CCD vs CMOS image quality can ever be resolved? In this case we have to rely on the observations of skilled photographers with integrity. I personally evaluate subjective evaluations based on the photographers' motivations. In general, human nature intervenes, so I pay more attention to the reports on subjective aesthetics from those who only have a stake in the quality of their work. Mitch is one example, but there are many, many others. After the M8 IR contamination debacle I ignore photographers with any personal ties to or commercial dependencies on Leica.

In the case of the M8/M9 there are numerous photographers who only care about the image quality in the scope of their artistic activities. These photographers assert the M8/M9 rendered images are aesthetically superior to other platforms.

This appears to create a dilemma. The CCD technology is inferior when measured objectively, but the rendered result is considered superior by many.

This dilemma does not exist because the components in-between the lens and the photodiodes play an important role.

The advantages and disadvantages of anti-aliasing filters are no longer controversial. AA filters are not required for aesthetically acceptable image rendering. I choose to ignore the IR filter.

This leaves the color-filter, micro-lens array. The CFA is just as important as the sensor signal-to-noise ratio and dynamic range. Unfortunately the characteristics of the CFA are much more difficult to measure and evaluate.

Consider two very different CFAs. One is made to minimize design, material and production costs. The other is made to provide spatial and frequency information that directly corresponds to the Bayer-reconstruction mathematical model. Is it unreasonable to assume a design where each micro-lens collects as much light as possible (regardless of it's physical location in the array) and filters out all frequencies except the frequency the Bayer model, costs more to design, source and manufacture than a design with less ambitious goals?

The value-aded aspect of the M8/M9 design is not the photodiode-array portion of the sensor. Instead it is found in the CFA optics.

Of course other brands can employ high-quality CFA optics along efficient CMOS electronics. In fact, these brands exist. But in some cases other brands are tempted to sacrifice frequency-filtering specificity for signal level. A green micro-lens that passes 15% more light from non-green frequencies generates more charge (signal). The SNR and DR are increased. However the color rendering may be inferior because the green-frequency information the Bayer model requires is contaminated. The data is not what the Bayer model expects. The rendering is compromised. The rendering is aesthetically superior.

There is no dilemma. The subjective cannot compete with the resolutely, unequivocally, and empirically objective. Those who see "umami" in a sensor need to objectively prove it, in a blind test against other output mixed from different sensors.

Good luck with that.

Not only do other brands have arrays, it is not clear that Leica has done a better or worse job than anyone else:

http://www.dpreview.com/forums/post/52351544

I would posit that by far the most significant aspect of the Leica T is that it is a whole new mount; a mount designed to eventually do away with some of the micro lens array and tele centricity issues we see now in trying to be backwards compatible. In that, Leica is pulling a Fuji (from 3 years ago). One could say that the M mount may be getting long in the tooth for where Leica wants/needs to go. As an accessory brand for the fashionistas, they need to fit in a purse taken on a stroll to the Hermes store in Manhattan.

What appears to be going on in the engineering of sensors is Sony running circles around everyone else (A7 and A7r variants as an example). Since their (re-)acquisition of Toshiba's facilities about 2 years ago Sony Industrial has totally ramped up both their R&D and their fab output. Sony is dominant. Leica does not use Sony's sensors (possibly because Sony will not sell to them) and as a result for all of Leica's excellent engineering of bodies and glass, they are solidly behind in sensor application into their bodies. Even the 240 gets middling results compared to a D600, and is eclipsed by the D800/D4s. And the S2 gets taken to the cleaners by the D800e. The Leica FF 240 gets the same DxO score as the APS-C Nikon D5200. Ouch!

So the problem for Leica from a purely photographic and not business analysis is they are less capable cameras. They make up for that on the bottom line by having Seal promote them as bling and longish videos of someone sanding aluminum. It's like watching someone make jewellery as opposed to someone take photos. Odd.

 

[semilog]

There is no dilemma. The subjective cannot compete with the resolutely, unequivocally, and empirically objective. Those who see "umami" in a sensor need to objectively prove it, in a blind test against other output mixed from different sensors.

Forget about it, Aristophanes. It's Wetzlartown.

 

[GaryLH]

M9 Sensor Qualities

I agree Semilog.

I tend to think given the computer design aids available today, it is more deciding cost benefit of going for the last percent vs the acceptable manufacturing tolerences vs cost point.

Though the new Sigma 50f1.4 art may not be as good as the Zeiss, it is so very close and cost far less, for example.

Gary

 

[Mcary]

I agree Semilog.

I tend to think given the computer design aids available today, it is more deciding cost benefit of going for the last percent vs the acceptable manufacturing tolerences vs cost point.

Though the new Sigma 50f1.4 art may not be as good as the Zeiss, it is so very close and cost far less, for example.

Gary

Agree!

Lens like the Zeiss are for people that want that extra 1 or 2% and are willing to pay the additional 300-400%+ for it.

 

[Aristophanes]

You must be thinking of Zeiss and Nikon — the companies that design and manufacture steppers.

And who do you think makes steppers for Sony's fabs?

Nikon.

And who besides Sony has had access to all of Sony's sensors, including FF even before Sony used them and during Sony's lull from FF?

Nikon.

 

[Richard G]

Forget about it, Aristophanes. It's Wetzlartown.

"Bad for glass."

 

[YYV_146]

I agree with Aristophanes here. What you can do with an image - and using a 7d for years I know all too well - is limited by the reaches of a camera's sensor.

Of course one may have preferences, and of course an older sensor may produce more pleasant images. But let's drop all the magic talk, all the red-dot associated myths and all the personal attachments to the rangefinder ideology for just a moment: The M9 is in no ways more "special" than any other CCD full frame camera, and the M240 any other CMOS full frame. The design is well thought out, the image is sharp because of the lack of AA, maybe Leica has its charms when it comes to OC Jpegs, but that's it. Leica is limited in the digital world because it does not have access to the R&D and funding of Sony, and cannot afford to constantly go for the latest and greatest sensors. Heck, even Canon is struggling in the modern digital world - their crop bodies deliver incredibly mediocre results in 2014.

And whatever the reason someone buys and uses a Leica FF body, it should not be because of image quality. They are exceedingly beautiful, well-built machines and make you feel great when working with them, but this photographer took time to try out both the M type 240 and M9 and was disappointed. The M9 simply does not deliver the images I need for critical work, and the M240's poor EVF, intrusive LED framelines and high iso banding made it a no-go.

 

[Doug]

M9 Sensor Qualities

Above earlier posts transferred here from another thread...

 

[semilog]

Above earlier posts transferred here from another thread...

Thanks for the ninja moderating, Doug. Well done.

 

[YYV_146]

Above earlier posts transferred here from another thread...

Thanks Doug!

I stand by my point. Sensor tech has improved vastly in recent years. I remember the days when 135 sensor were far worse than good 135 film. Today the relationship has largely turned around: most full frame sensors are as good as film when it comes to print sizes, and some (D800E) approach medium format film in terms of overall resolution.

When you look back at some of the older digital solutions - the RD1 or a Nikon D70 or D2, it really is amazing how much things have evolved since then. Every time I buy a new camera I am delightfully surprised by how much extra processing space it has over the older ones; with the A7 there is so much dynamic range that I can max out most of the sliders in ACR and still produce usable images at base ISO. In 2010 I would have thought this was impossible...

Is the M9 excluded from this trend of digital technology decay? Of course not! One day we will be talking about it as we talk about the original S or RD1 today - an icon, still a fine tool for imaging, but not something people generally use for critical work.

And I am all the more happy for that. Leica has demostrated with the M240 that European FABs can still (roughly) keep up with the Japanese. Good work considering their small user base and R&D budget, and they really managed to cover a lot of ground between 2009 and today. Maybe the next gen M will actually outperform the Sony FF models - Having just dumped almost a year's savings on a Noctilux, I wish Leica the best of luck.

 

[uhoh7]

The tech is all ready for a ultra-light mirrorless M-mount, with good focus aids like the Sony, and a good sensor, like the M9.

Why no one wants to build it, I'm not sure.

Meanwhile the M9 still rules at 135mm and below, close focus excepted. All the DR and ISO tests are meaningless, outside some unusual night events--even those the M9 can do with the right glass.

Results are results, and the Sonys are not close to the M9 in daylight, except in a few circumstances--even then I don't like them better.

Better yet, if you can afford a 5D mk3, you can afford an M9 today.

It's a quirky camera with all sorts of things about it which might be better, but I like mine better than any camera I've owned.

I shoot alot with the Sony A7 too, but it's not the same.

The 240 has some nice features, but no shots I've seen convinced me it's better, and with some glass it does not seem as good.

 

[x-ray]

I wish Leica the best too.

I started thinking about the 11 digital cameras I've owned and used in my work since my first DSLR, the nikon D1. I've owned 5 CCD cameras or backs and 6 CMOS DSLRs. Let me start by saying I'm very critical of color due to client requirements. In the past couple of years I've produced custom color profiles for each of my cameras under each lighting condition I shoot under. I apply these to my files during raw processing. Currently I shoot a Nikon D800, Nikon DF, M9 and Hasselblad with a CRV39 back. The CFV39 and M9 are CCDs and as you know the Nikons are CMOS. My previous DSLRs were canon 1 series digital with CMOS and early in my digital work I has a 1D Canon, D1x nikon and D1 with CCDs.

In my experience there is a difference in color rendering between CCDs and CMOS. I find very different color outputs between the same makers different models like the D800 and Df. The same was true of Canond models I had.

Producing and apying custom profiles creates more uniform colors between cameras but there are still distinct cor differences. I find the D800 after apying the profile to have very accurate colors particularly reds and blues. Saturation and purity are excellent. The D800 by far has the best dynamic range of any digital camera I've used. Next the M9 after applying the profile has very easing color with very nice greens. The look is certainly different than the Nikon but I wouldn't say it's better, it's just different. The m9 dynamic range is poor by current standards and especially compared to the D800 and noise is terrible compared to both Nikons.

Now comes the Hasselblad CFV39 back. The essentially FF back has much better color than any of the other cameras before and after the profile is applied. It's especially fine after apying the profile. Color purity is spectacular as is color depth due to its true 16 bit color capture. It like the M9 has no AA filter which is a double edged sword. Moire can be a serious problem with both cameras. Dynamic range with the CFV39 is around 12.5 stops vs 14.5 with the D800 and just guessing from experience I'd say the M9 is around 9 stops. I seem to have read that too. The diwnside of the CFV39 is noise at high ISO as is a problem with the M9. It was a problem with all of the CCD cameras I've owned vs CMOS cameras.

One thing I noticed with my 1Ds and 1DsII and other Canon CMOS cameras I've used is a particulate bad red rendition. Even after profiling the reds were biased to the yellow side. IMO they produced a very poor red.

As to the M9 having been state of the art and smoking any CMOS, I seriously have to disagree. This sensor was never an outstanding performer I any respect at the time it came out or after. It IMO is below average in all respects but does produce a pleasing image. This is just my opinion based on my API application. To me and the reason I bought the M9 was because of wide open performance of the current crop of lenses. The current leica glass is exceptional even in the corners wide open. This is the reason I purchased the system and the only reason I tolerate it's issues.


It seems like everything In life is a compromise including cameras and sensors. You select the gear that serves you best and work around the deficiencies.
Ymmv

 

[rivercityrocker]

I'm not sure why anyone is even responding to "Aristophanes".

Here's someone who posts a thread that basically says all of the same things you read on Nikon Rumors and DPReview whenever the name Leica is brought up; "The Leica camera is a status symbol" and "Leica sensors are old and suck in low light".

Why even engage in discussion? This discussion has been done OVER AND OVER.

I've got dozens and dozens of images that I shot with an M8 and M9-P in low-light that are amazing. Yet some person on the internet insists that the camera can't handle it? I have proof to the contrary.

If I was so concerned with my camera being a status symbol then why did I opt for the model without a "Red Dot"? For all anyone in the general public knows I have a crappy old film camera or possibly a Fuji.

If you don't care for Leica products and think they are inferior then WHY are you here commenting about it? You could just ignore it and let the people that like their choice of cameras go on with their lives. Do you think that all of us M9 users are going to have some epiphany? "Oh, gee! I'm a slave to marketing! I should go and buy a different camera because I've been duped by Leica..."

Some of us don't give a **** about celebrity marketing. I for one don't pay attention to what celebrities are wearing. I've also got a camera that is the top performer in low-light, but I still prefer to use an M9 even in the dark. It's not always about what is the best or following celebrity or market trends. I like the way my Leica operates. I work within the boundaries and limitations and I make amazing images with what you deem a substandard camera.

Instead of worrying so much about the "M9 Sensor Qualities" and postulating that we all bought a Leica as a status symbol, why don't you just go take whatever camera YOU think is great and go take some photographs?

 

[hepcat]

No Leica, not even the S-series, has a truly up-to-date sensor. They are all at least 1/2 generation behind Sony's fabs right now.

Leica has an unabashed product placement system for music videos, fashion catwalks, New York gallery parties and the like. Leica is notorious for sponsoring news stories. It's no secret they subsidize and gift cameras to visible events and trendsetters; their own guerrilla marketing. That was the whole point of the Hermes investment; to make Leica a status symbol:

Please note that it is Leica's marketing that repeatedly uses the word "status" when promoting their products. Leica has entered into dozens of deals to promote only luxury placement of their product:

I've been to one of these presentations. They don't even make a pre tense of selling you a camera. Leica is entirely positioned as a luxury accessory.

The schmucks who purport to buy Leica's for their IQ, which is definitely not supported by the sensors...not even close (Ming Thein's S2 review is a killer)...are only giving street cred to the status brand effort ("See. The old school RF geeks still like them"). When the RFF crowd reach up to pay luxury prices, it only feeds the Leica marketing effort to give the stars who use Leica the impression they are continuing the workaday, photojournalist credibility that built the brand. It's the mystique and nostalgia that sells. You are helping them. They've got you doing their work.

As Mr. Schultz said last year, Leica left photojournalism behind. One could argue that Leica is working to leave the RFF crowd behind as well. Fuji's taking note.

The M3 was built tough because it had to be tough. It went to war! The T's images are pedestrian given the price. It goes to fashion shows. It is a fashion show!

Leica was never a value brand save for their outstanding durability; a rugged, enduring photographic tool. The T looks like it was made out of aluminum block so they could make a marketing video showing it made out of an aluminum block. This is engineered marketing wrapped around electronics that are as average as a sub-$700 Nikon.

What's your point? My M9P works, and works well. It's a couipled-coincident, full-frame <mostly> manual rangefinder camera. As a matter of fact, it's the only game in town. It does what I need for it to do. The T isn't an M and I don't particularly care what it is or what they do with it, as long as the income it generates allows Leica to continue to develop M bodies.

And btw, the M3 never went to war, at least in the U.S. The M series is a post-war invention. The U.S. military later used M2s and M4s. You may be spot-on with your analysis of electronics, but you need to read up on your history. 😉

 

[Aristophanes]

What's your point? My M9P works, and works well. It's a couipled-coincident, full-frame <mostly> manual rangefinder camera. As a matter of fact, it's the only game in town. It does what I need for it to do. The T isn't an M and I don't particularly care what it is or what they do with it, as long as the income it generates allows Leica to continue to develop M bodies.

And btw, the M3 never went to war, at least in the U.S. The M series is a post-war invention. The U.S. military later used M2s and M4s. You may be spot-on with your analysis of electronics, but you need to read up on your history. 😉

Oh, I know my history. So does Leica:

http://www.dailymail.co.uk/news/art...es-Vietnam-war-sells-1-2-million-auction.html

http://blog.leica-camera.com/photog...nd-the-image-that-helped-end-the-vietnam-war/

Ut shot one of the most iconic wartime photos ever with an M2.

http://www.bbc.com/culture/story/20130726-leica-shoot-to-thrill

I am glad some people here think the red dot gives their M9 low light capabilities, because the laws of physics say those shots are missing critical exposure data from an inferior sensor. The DxO marks of the M9 means it has to miss far more shots compared to even some of the APS-C's of the same generation simply because its read noise and data numbers are so dismal. That affects ALL shadow shots with the M9.

Is that the value in the product you were looking for?

As for "celebrity marketing... Leica sells to you at full cost and use the premium to pay Seal. This reverse marketing is how Leica generates revenues these days. It used to be the workday journalists would buy Leica's even when AP would issue Nikon's as the field camera. That esteem of the Leica is what earned the well-deserved reputation. Rugged and well-serviced, Leica could take it and deliver excellent results. Now the marketing trend is to "seed" celebrities with the the product at visible venues and events to generate a buzz.

But from an engineering standpoint, the Leica image quality results are...missing...just like the RAW data reads from sub-par sensors. This apples to the S2 where one spends 6x the price of a D800 and the D800 eats the S2 for lunch. All that Leica glass does nothing for the S2. It does a lot for Leica's bottom line, but photographers eventually must wonder where their data is going.

The delta between Leica's IQ engineering output and its sales price has widened to an extreme we do not see with any other brand. The T uses a 3 year-old Sony sensor. Excellent engineering on the body. Top engineering on the lenses.

In the face of poor sensors, Leica's business model is t over-engineer on the optical and mechanical attributes and swallow it big on the electronics. CMOSIS makes a substandard product that does not align with the Leica standard.

What do you do when you are a brand based on German engineering pedigree but you cannot source the right product to deliver the photos comparable to a $600 Nikon?

You hire Seal. Go to the Oscars. Change the brand to a luxury brand and walk away from the photojournalist market that made your engineering stand out. You sell the status of that engineering pedigree even if the manifestation of the engineering no longer cuts it.

Leica very much counts on the inertia of its followers to purchase regardless of relative IQ. Leica actually expects you to purchase inferior quality in the images on a value basis.

There is zero colour difference between CCDs and CMOSs. It is all in the filters be they physical or software (Bayer, IR, interpolation, algorithm). If you do not understand this you do not understand how a photodiode works. CCD and CMOS have identical monochrome responses. Funny how light works.

 

[Corran]

The DxO marks of the M9 means it has to miss far more shots

Please enumerate all the photos that might be "missed" because the Leica M9 sensor has XYZ deficiencies.

Also, while you're at it, please let me know about all the photos in the last many decades that are worthless now because they were taken on 35mm film, which certainly has less real image quality than the Leica M9 sensor.

Your argument is specious at best.

 

[rivercityrocker]

I am glad some people here think the red dot gives their M9 low light capabilities, because the laws of physics say those shots are missing critical exposure data from an inferior sensor. The DxO marks of the M9 means it has to miss far more shots compared to even some of the APS-C's of the same generation simply because its read noise and data numbers are so dismal. That affects ALL shadow shots with the M9.

Is that the value in the product you were looking for?

My M9-P doesn't have a red dot. There's more to photography than physics. When I shoot in low-light I often use timing to overcome shortcomings of the high ISO capabilities. You see some of are ACTUAL photographers. You tell ME that my M9-P can't possibly be used in low light because ****ing DxO Mark says so. But I have photographic proof. I shoot concerts for a LIVING. I use my M9-P in some of the worst possible lighting conditions and for some inexplicable reason contrary to what DxO Mark says I get perfectly usable images. Hell, I even get great low-light images with my M8.


Knockout Kings II - Figueroa vs. Arakawa by _JDT0505, on Flickr


Lucky Tubb by _JDT0505, on Flickr


Reverend Horton Heat at Emo's in Austin, TX by _JDT0505, on Flickr


Dale Watson at Emo's in Austin, TX by _JDT0505, on Flickr


Jay Z by _JDT0505, on Flickr


Megadeth by _JDT0505, on Flickr

So, c'mon. Tell me that the M9-P is worthless for shooting in low-light. I can keep pulling images out to prove the contrary all night.

 

[Duane Pandorf]

I think the OP must work for Sony and is worried about his job future:

Sony downgraded to 'junk'

Sony sells PC business, cuts 5,000 jobs

Sony's Gaming Business Is in Trouble

March 2014: Sony Worldwide Studios’ No Good, Very Bad Month

Sony Layoffs Hit Distribution — Dallas Office To Close, L.A. Operations Support Group Affected

Quote from an A7r reviewer: "On the other hand, if Sony goes under (alas, they’ve been having financial problems), or if Canon comes out with an amazing 50-megapixel, noise-free, $2000 camera next month (okay, I know, that price is out of the question), I still have my Canon lenses."

So tell me where are all of these other camera companies going to go to get their sensors if or when Sony goes bankrupt?

All I can read into this thread is the OP is jealous of all of us Leica shooters who support a company that makes the only digital RF camera to date. We really enjoy making photographs with our "inferior" cameras and willing to pay a premium over the competition for the experience.

Get over it and be happy with whatever you enjoy making photographs with.

Last time I checked this is the "Rangefinder Forum". Not the EVF forum, LCD forum or DSLR mirror slap forum. Fortunately, all of these other cameras are discussed here.

When the Sony A7(r) has a system of lenses specifically designed for that body (and I'm not talking about adaptor this or adaptor that) that encompasses what Leica has already established and has simplified the back end of the camera where I don't need a 200+ page manual to figure it out....Then maybe...

But I venture from Sony's past history and current financial difficulties that that's not going to happen as Sony will have changed it all in the very near future and all of you will have to start all over again. Or be bought out by another company. Hopefully you'll have access to adaptors that will fit whatever lenses you have to keep the system working.

JMO.

BTW, I see Sony is already offering $650 discounts. Must be selling well!