M9 Sensor Qualities

[nightfly]

I think we should all admit that after a certain point, we don't really care how objectively good a sensor is. It's good enough. What we're paying for is something that we enjoy shooting and that we don't fight for a shot. The same way an iPhone might not have the biggest screen or the fastest processor but it works really well and feels designed by people who care about user experience.

I know when I first started shooting film and became aware of Leica, I did everything I could NOT to buy one. The reason I ended up with a used M4-P is that nothing else really worked as well for the type of shooting I wanted to do. (At that time they all had comparable full frame sensors). When I finally got one, I was like, yeah for all the money I spent on NOT getting one, I should have just bought one. It just works.

My used M9 is coming tomorrow. I've wised up (somewhat).

 

[__--]

...But as someone else elaborated in technical details, a sensor may be a sensor, but the chain of processing does differ from model to model. Somehow I doubt all those people claiming a unique color rendition from the CCD models, myself included, aren't all totally deluded...

I don't think so, having looked at this from the point of view of one of the people who prefer the M9 color rendition to that of the M240 — although I don't feel that there's a need to argue whether this is attributable to the CCD vs CMOS or to the CFA, or whatever.

Ming Thein, who seems to have good color sense, in his M240 review wrote about a range of color color rendition from the M9 (which he liked) to the D800, and placed that of the M240 between the two. I've written previously about the conclusions on Marc Williams ("fotografz"), who gave up on the M240.

The writer of the "prosophos" blog (Peter) originally bought an M240, but after two weeks also gave up on it. Now, after running into some problems with the M9 rangefinder mechanism, he once again bought an M240. This time, he says that he has given up on trying to achieve the color rendition of the M9 in his M240 processing — with this approach he feels he may be able to achieve a more satisfactory look for his color work with the M240.

All of the above photographers are are highly skilled in post processing — and all of them think that there is a difference in the color rendition of the M9 and the M240 that cannot be equalized by post-processing, and I don't think that there is any delusion among them.

MITCH ALLAND/Potomac, MD
Download links for book project pdf files
Chiang Tung Days
Tristes Tropiques
Bangkok Hysteria
Paris au rythme de Basquiat and Other Poems

 

[Corran]

blahblahblah

You still haven't countered my post from some pages ago. Namely, that regardless of the competition, the M9 sensor is still miles away better than the raw IQ measurements of any 35mm film. And of course there are many of us still shooting film anyway. 35mm film was used professionally for decades, and an innumerable number of good and successful images are forever immortalized in that format. And further, why argue about sensors in 35mm size? Are you buying the new generation of CMOS-equipped medium-format digital?

Listen: the raw sensor metrics don't matter. Get it through your skull.

 

[aeturnum]

There seem to be a lot of people in this thread who are upset that apples aren't oranges and visa-versa.

There is no question that there are sensors that capture more information than the ones used in Leica cameras. However, the apex of sensor technology is not found in other consumer cameras, but in high-end bespoke sensors for scientific and industrial designs. The fact that Leica sensors are inferior to Sony sensors is just as irrelevant to me as the fact that my hybrid sedan is slower than a ferrari. I did not buy the hybrid to race people and I did not buy my camera with the goal of "as-close-to-perfect" rendering.

I owned an M9 for a few months when I saw the opportunity to buy one cheaply. It was, in general, an awful camera and a terrible value for money. However, I (personally) found the files to be extremely flexible and the colors to be wonderful. I am sure that if I was better at manipulating images, I could get a similar look out of my current digital camera (an A7), but with the M9 there was no need. Leica (and adobe - as this was lightroom) had set the camera up to do some of that work for me. I appreciated that. The M9 was, within, "low ISO," just as capable as any other camera I've owned. If I can work around ISO 200, I would prefer it to my A7.

As an example, I submit this photo. As you can see from the next photo, the ambient lightning was at about EV 5. However, I was an idiot and let the camera expose for the windows. I was able to push the DNG by 2.5 stops to get the result you see.

The A7 is technically superior to the M9, but the majority of situations fall well within both camera's capabilities. The reductionist view of cameras ('sensors in boxes') is unhelpful. There is a great deal of processing between the sensor and the RAW output. We can use the sensor's capabilities to inform our understanding of the limits of a particular piece of hardware, but there is also a lot that a camera maker can do to tune the RAW file. Anecdotally, I like the color on high-ISO A7 photos less than other cameras I've used. Since the sensor is, "cutting-edge," I tend to credit / blame the people who designed the "sensor measurements to ARW" part of the camera hardware. I encourage anyone who doesn't believe me and is rich to try an A7r and a D800e side-by-side in the real world. Both outputs will be excellent, but you'll also see differences between the files.

As an aside, I don't really care how leica presents itself. I find its current public image to be off-putting, but that doesn't affect the quality of their cameras. I have no idea why people (on both sides) go on and on about the 'character' of leica as a company. Who cares what kind of company Leica is - use their cameras if you like them, don't if you don't.

So, to sum it up, I have used many digital cameras for my "work" (which is neither work nor productive) and the M9 is the one that has stuck with me. The others had strengths and weaknesses, but didn't seem to draw me in the same way. I also find, in retrospect, that I gravitate to the "look" of my M9 photos more than other photos. I don't know why that is, but I do. Perhaps I've been sold on the leica marketing, but I think it's much more likely that leica managed to put together a package that performs well within the constraints of the sensor.

 

[CameraQuest]

Does any site offer downloadable comparison files of the same pic taken with the same lens from both the M9 and M240 ?

Stephen

 

[YYV_146]

It's quite obvious that you cannot push an M9 file shot at ISO 5,000 two stops because the DR at this speed is zilch, nor is it desirable to push M9 files shot at ISO 1,600. Earlier I provided a link to this thread, which describes the technique of shooting the M9 at ISO 640 and pushing in LR4 or LR5. Within that thread there is a link to Jim Kasson's excellent blog work in testing and explaining why this technique works. The thread also has many examples of the color rendition one can get from the M9 using this technique — and post #31 on page 2 has the steps suggested for shooting and processing for this technique. As in shooting and developing film, technique in shooting digital (obviously) can also matter.

MITCH ALLAND/Potomac, MD
Download links for book project pdf files
Chiang Tung Days
Tristes Tropiques
Bangkok Hysteria
Paris au rythme de Basquiat and Other Poems

None of your images are pushed *that far...640 pushed by three stops is less than ISO 5000 equivalent, and I can see an incredible amount of color noise in the dark areas of the images that underwent this treatment. The ones pushed by 1-2 stops are better, but why would you need to push anything on a modern digital camera at 2500 or 3200?

And IMO this is a technique that many of us use, not only for the M9. Note that the A7 is capable of ISO 25600 and the NEX7 goes to 12800, but I only shoot at below 5000 on the A7 and below 1600 on the NEX7. By pushing at these ISO levels I get cleaner images than the OOC RAW files at higher ISO levels. In fact even if I'm shooting at base ISO I tend to underexpose by a stop, just to make sure there aren't any wild small patches of light in the image that make post processing difficult...

 

[hepcat]

Does any site offer downloadable comparison files of the same pic taken with the same lens from both the M9 and M240 ?

Stephen

Here are a few direct comparisons available for download.

http://www.farines-photo.com/wp/?p=787

http://www.farines-photo.com/wp/?p=758

http://www.farines-photo.com/wp/?p=697

 

[YYV_146]

The A7 is technically superior to the M9, but the majority of situations fall well within both camera's capabilities. The reductionist view of cameras ('sensors in boxes') is unhelpful. There is a great deal of processing between the sensor and the RAW output. We can use the sensor's capabilities to inform our understanding of the limits of a particular piece of hardware, but there is also a lot that a camera maker can do to tune the RAW file. Anecdotally, I like the color on high-ISO A7 photos less than other cameras I've used. Since the sensor is, "cutting-edge," I tend to credit / blame the people who designed the "sensor measurements to ARW" part of the camera hardware. I encourage anyone who doesn't believe me and is rich to try an A7r and a D800e side-by-side in the real world. Both outputs will be excellent, but you'll also see differences between the files.

This is just my two cents, but aside from RAW compression, the main reason why the A7r is not as good as the D800E (and the A7 the D610) is because as mirrorless bodies they deal with a higher sensor temperature and aren't as clean at most ISO levels. The M9 also has this issue compared to CCD DSLRs, because you can't fit a DSLR-sized copper sink into such a small body...

Go check out Roger's A7r teardown and note how massive the heatsink is relative to the body size. But that's still much smaller than what's insider the D800E...

 

[Takkun]

Mitch-thanks for the added input. I misspoke; I meant to rebut the "all sensors are the same" comment earlier with something about the chain of processing downstream adds some subtleties to the camera's output.

I also agree with the "who cares about sensors these days" comment. I used to shoot with a D1x for journalism, and the low light was abysmal and the files not particularly conducive to enlarging, but two generations later everything on te market was more than good enough. I just stopped caring about resolution, low light, etc improvements after that.
Even my only practical beef with the M8 is losing a bit of wideangle.

 

[YYV_146]

I hate when people try to tell me what the lighting situation is in my photographs. YOU WEREN'T THERE. The light looks good at the boxing ring, but it WASN'T. It was DARK. There was some light there, but it looks a lot brighter in photos than it was in real life. That's part of the processing. The people are static? No they are moving. A LOT. That's where technique comes into play. You have to press the shutter release at that perfect moment when they are in between two motions, which is typically less than a half-second of time.

Yes, but the fact that there are lights on the subject changes everything. Concerts are also by far not the most demanding work for a camera's ISO performance.

A while back I was in Hong Kong doing a set on the annual June 4th commemorative march. The streetlights were dimmed, and an entire street next to the Pearl River was illuminated by candlelight . That's what I mean by dark.

And however much I applaud your technique of getting people when they are between motions, a better camera would allow you to simply shoot whenever you want. Back in the 1d mk3 and 7d days I also had to do this, when I use the NEX-7 I still have to since I'm mostly at 1/30-1/60. But on an 1DX I just fire away at 1/250, and performance is admirable enough at 25,600 that I couldn't care less. That's good high ISO performance.

And since when is f/1.4 considered slow? That's one of the stupidest things I've ever heard. So you've worked with M9 files? How extensively? I routinely push my M9 files to 6400 or more. When I was young I worked at a Ferrari dealership detailing the cars. I drove hundreds of them around the block. That doesn't make me an expert on how Ferraris handle at high speed any more than you working with a few M9 files makes you an expert on what the M9 can accomplish in truly capable hands.

Which brings me to another point. How many concerts have you ACTUALLY photographed? How many with the M9? I have shot THOUSANDS of concerts in all kinds of venues and all kinds of lighting, from bars to stadiums. I've shot over a hundred concerts using an M8 or M9. Just because you took a Sony in to shoot the local bar band a few times doesn't make you qualified to lecture me on how concert photography is done with an M9.

Really. You feel the need to actually rant like this. Well, sorry to tell you, but I am pretty good, concert work or not concert work. I shot film for several years before going to a Canon system, which I sold between 2010 and 2012. Yes, I may have not shot as many concerts as you, but I used to shoot Canon/Leica, now I shoot Sony/Medium Format, the local paper uses Nikon so I work with Nikon files all the time. I've been using Photoshop since CS2, and I can tell you, if you've carefully worked on 30,000 files from different cameras, you can tell how good a sensor is by looking at a couple of files.

Please give me one single example of you pushing an M9 file to above 6,400. That's four stops beyond ISO 400. I simply don't believe that it can be done with any respectable quality of output. And F1.4 is not fast for a 50mm prime. If you are indeed working with the M9, the super-fast options are both cheaper and more plentiful than Canikon - I'd set the bar at 1.2 for a standard lens, at the very least.

Oh, and I get plenty of contract work from bands in Asia. I don't think most "bar bands" perform in soccer stadiums....

When you get as much experience shooting concerts with an M9 then come talk to me and tell me what's what. Your opinions on the subject aren't valid because you don't have the experience to back up your claims.

No, I'm not doubting the M9 as a concert tool, I am doubt it's high ISO performance. I don't think that the concert pictures you made are bad - far from it - but in 2014 there are a lot of ways to get better output.

 

[YYV_146]

You still haven't countered my post from some pages ago. Namely, that regardless of the competition, the M9 sensor is still miles away better than the raw IQ measurements of any 35mm film. And of course there are many of us still shooting film anyway. 35mm film was used professionally for decades, and an innumerable number of good and successful images are forever immortalized in that format. And further, why argue about sensors in 35mm size? Are you buying the new generation of CMOS-equipped medium-format digital?

Listen: the raw sensor metrics don't matter. Get it through your skull.

No, sensor technology hasn't caught up with the best 135 film until rather recently. I would say only the most recent 36MP FF sensors can claim a definitive resolution edge over Velvia 50 or similar. Certainly not the M9. The color edge is always there for digital because in digital white balance can be set after making the image, but your claim largely rests on what you care about in output.

And raw sensor metrics matter very much. The only reason there are professionals shooting large format film these days is because a digital sensor capable of delivering 4x5 resolution is prohibitively expensive. Do you think that people who need a H5D or D800E is simply deluded in their search for more megapixels?

 

[hepcat]

Concerts are also by far not the most demanding work for a camera's ISO performance.

A while back I was in Hong Kong doing a set on the annual June 4th commemorative march. The streetlights were dimmed, and an entire street next to the Pearl River was illuminated by candlelight . That's what I mean by dark.

But on an 1DX I just fire away at 1/250, and performance is admirable enough at 25,600 that I couldn't care less. That's good high ISO performance.

... but in 2014 there are a lot of ways to get better output.

Victor, I find your post interesting. You're obviously very experienced and have a signficant CV relating to commercial work. How much of your work is shot above 6400 ISO?

I almost never shoot above 320, and 640 is pushing what I do. I have large aperture primes, the VC 50 f/1.1 and 35 f/1.2 and for what I shoot, that combo of ISO and aperture is generally more than sufficient.

My question is that I wonder if your work has become so specialized that those high-ISO attributes are important to what you do and why you're pushing the topic so hard. I suspect that, frankly, except for those few of us (you) who shoot black cats in coal bins at night regularly, ISO 640 and f/2 at 1/30th isn't sufficient for 98% of what most photographers shoot anyway? Are there a few of us (well, you...) here who push the envelope a lot more than the rest of us, and your needs are driving the discussion?

 

[YYV_146]

Victor, I find your post interesting. You're obviously very experienced and have a signficant CV relating to commercial work. How much of your work is shot above 6400 ISO?

I almost never shoot above 320, and 640 is pushing what I do. I have large aperture primes, the VC 50 f/1.1 and 35 f/1.2 and for what I shoot, that combo of ISO and aperture is generally more than sufficient.

My question is that I wonder if your work has become so specialized that those high-ISO attributes are important to what you do and why you're pushing the topic so hard. I suspect that, frankly, except for those few of us (you) who shoot black cats in coal bins at night regularly, ISO 640 and f/2 at 1/30th isn't sufficient for 98% of what most photographers shoot anyway? Are there a few of us (well, you...) here who push the envelope a lot more than the rest of us, and your needs are driving the discussion?

I can't tell because prior to getting the A7, all of my personal work is with cameras I don't use beyond 1600 (native), and I can't see processing levels from aggregated statistics on LR. I can maybe push the X-E1's files to 10,000 equivalent given the excellent color noise control, but it's something I'd rather not do.

But I agree, in the past year or two at least, I've been mostly known as someone who specializes in available light work. I just check the statistics on my copy of LR, and indeed more than 50% of my pictures are above 800 (all personal stuff, that is). And I process my files extensively, so people who are okay with OOCs probably won't have the same problems...

In this thread there has been a lot of name-calling, and it pains me to see this, usually coming to RFF because of the lack of it. At any rate, I won't comment on the M240 or M9's IQ any more after this thread.

If I may make some points very clear:

1. The M9's OOC Jpeg rendition is nothing short of excellent. Better than the M240 and the Fuji X bodies, way better than the A7. Most likely better than any camera I've used. But the meter is unpredictable, and I find that this problems takes away much of this advantage.

2. The M9's base ISO files suffer from low DR and shadow noise. They do not suffer from a lack of color depth. Color depth is also excellent, which probably relates to the good Jpegs. Tonal range is probably good enough - I should test that some day.

3. There are glaring deficiencies in color rendition on all of these cameras. The reds are too weak on the M9, too strong and tinted on the A7 and oddly orange-ish on the M240. On X cameras the green is simply mush. But frankly, with post processing you can fix all of that...maybe not the green mush, though. I will say this, the M240 has the best base ISO color accuracy in any 135 format camera I've ever used. Not that it's really good, but better than the rest.

4. I would gladly use an M9 for ISO 100-200 work given the sensor quality. There are a few other issues holding me back from an M9 - half of the M9 bodies I used at one point or another have been back to Solmes for new sensor cover glass, and that scares me - but the sensor is no worse than the 5d mk2 or even the mk3 at base ISO. And while Leica may be justified in delivering sub-par sensor because it is a small European firm, Canon has exactly zero excuses to deliver cameras with the poor DR and color depth they have today.

5. I absolutely love modern Leica glass.

Sony A7, 135mm F2.8 (T4.5) STF, two Canon 430EX flashes and one Sony F43M

 

[hepcat]

Thank you, Victor... THAT was an informative post. I appreciate your time and your perspectives. It sounds like you do, in fact, frequently travel into territory I just don't have a need to go. There's a big difference in using an advanced feature and merely discussing it. Thanks.

Roger

 

[Corran]

No, sensor technology hasn't caught up with the best 135 film until rather recently. I would say only the most recent 36MP FF sensors can claim a definitive resolution edge over Velvia 50 or similar. Certainly not the M9. The color edge is always there for digital because in digital white balance can be set after making the image, but your claim largely rests on what you care about in output.

And raw sensor metrics matter very much. The only reason there are professionals shooting large format film these days is because a digital sensor capable of delivering 4x5 resolution is prohibitively expensive. Do you think that people who need a H5D or D800E is simply deluded in their search for more megapixels?

My D700 clearly was better than 35mm film scanned on a high-end dedicated scanner (LS-8000). My D800E approaches 4x5 chromes in resolution scanned on a good high-end flatbed (not an Epson).

Regarding Velvia, remind me about the extreme high-ISO performance of Velvia? My point is that look at any 400-speed film, even Portra 400, and tell me it's even close to even a D3300 in resolution. Look, I love film, I shoot it almost exclusively, but it's not even in the same class per unit area as modern sensors, even comparing Velvia 50. I prefer the look of the color on film though. Let's not forget too that the best resolution color film has like half the dynamic range of most sensors made today.

Raw sensor metrics don't matter, ever. The picture is not made by just the sensor. The picture isn't even made by the camera, if we want to be philosophical.

 

[mfunnell]

My D700 clearly was better than 35mm film scanned...

Sure. Why not? While we're at it, let's discuss sensor vs sensor, digital vs film, field of view, depth of field, medium format vs large format vs 35mm, FF vs APS-C vs m43, Canon vs Nikon, Apple versus Wintel, Protestant vs Catholic, cats vs dogs, the C shell vs bash (or ksh, if you prefer), Word Perfect vs MS Word, xwindows versus common sense and everything else we can think of. That'll raise the tone 😱

...Mike

 

[Aristophanes]

Raw sensor metrics don't matter, ever. The picture is not made by just the sensor. The picture isn't even made by the camera, if we want to be philosophical.

Oh, raw sensor metrics matter most of all. The picture IS the sensor from an engineer's and boardroom perspective.

Those engineering digital cameras are engaged in an abstract artifice of making a machine, a tool. When it comes to the sensor it's exactly the same as making RAM. The data the user puts in is irrelevant save in quantity. Photons care not about colour. It's how many we catch that matters. The more we catch, the more we can discard (random photon noise, Bayer, Photoshop). An M9 sensor simply catches much less than other brands. So why such a premium in price and why do people buy a photon catcher that simply catches less?

It is genius marketing when Leica can convince people to spend more to get less. That dynamic did not exist in the film era.

The picture is always made by a camera, if we want to talk marketing. It was film marketing that stressed the subject/object persons more than the machine (Kodak moments).

I concur that Canon's sensors are long in the tooth. If you graded sensor performance to price of camera body, and created a list based on price😛erformance ratio, it might look something like this (m43+ sizes):

1) Nikon
2) Sony
3) Pentax/Ricoh
4) Fuji
5) Canon
6) Olympus/Panasonic
7) Samsung
8) Leica

(Not sure about Sigma)

Nikon seems to tease a bit more out of Sony sensors than Sony, but the current Sony APS-C and FF sensors are dominant. Not only that, but Sony fab capacity since they acquired Toshiba's plants is phenomenal. Sony has an enormous lead and it even looks like they can make non-stitched medium format sensors.

Pentax/Ricoh follow Nikon and Sony but are all APS-C.

Fuji's X-trans is very good but has issues with base ISO (maybe even some loose measures).

Canon has been slipping, especially recycling their APS-C line. Their FF line is also a full generation behind Sony. I think Canon is having a very difficult series of discussions in their corporate HQ lately.

m43 is simply smaller. They make the most of what they've got. Much of their market is aimed at SOC JPEG. Convenience brand with a good flagship (but troubling financials).

Samsung struggles.

Leica's non-Sony sensors struggle, especially on the price per pixel/DR over ISO. That's just horrendous (I also suspect CMOSIS' unit cost and yield are dismal compared to Sony). It's not just a value argument; it's about how photography's flagship brand with such a history can make their top camera with sensors of lesser performance. It's like a 700 series BMW with a Honda Civic engine, priced like a Rolls Royce.

I don't think a lot of Leicaphiles here have an understanding of what the Blackstone takeover really means for Leica.

 

[Aristophanes]

Go check out Roger's A7r teardown and note how massive the heatsink is relative to the body size. But that's still much smaller than what's insider the D800E...

That heatsink is not at all large compared to some CCD video cams from 5-10 years ago, nor is it large in relative size to anything from CCD development. The bane of CCD is heat. That is the major reason why CMOS is almost 100% triumphant.

 

[goamules]

If this thread wasn't titled "M9 Sensor Qualities" I'd start to talk about the OTHER aspects of a camera. Because the experience of using it is probably what 90% of photographers care about, not pixel peeping and sensor comparisons. No, I don't want or like the M9, my Fuji XE-1 sensor is just fine. The pictures look better than my G1, the crop factor is less severe, and the experience of using it is fun. I have no clue what an APS-C or Xfactor or whatever is. And as a former engineer, I learned all about CMOS, MOSFETs, CCD, et al. - BORING! Now, back to taking pictures while yall sweat "the sensors."

 

[Corran]

Oh, raw sensor metrics matter most of all. The picture IS the sensor from an engineer's and boardroom perspective.

Those engineering digital cameras are engaged in an abstract artifice of making a machine, a tool. When it comes to the sensor it's exactly the same as making RAM. The data the user puts in is irrelevant save in quantity. Photons care not about colour. It's how many we catch that matters. The more we catch, the more we can discard (random photon noise, Bayer, Photoshop). An M9 sensor simply catches much less than other brands. So why such a premium in price and why do people buy a photon catcher that simply catches less?

A sensor's DxO score doesn't make for a very compelling art exhibit.

Again, sensor metrics don't matter.