Magic of the M9 KAF-18500 Sensor

[uhoh7]

First let’s be clear: trying to understand the M9 sensor and the technical aspects of gear in general is no prerequisite for good shooting. It might be a hindrance for some since it may distract from other more central elements of making good images.

Nevertheless some of us always want to know “Why?” Hence this thread.

The KAF-18500 and its de-bayerized cousin are the sensors in the M9 and MM. Once secret, they are now available for purchase by anyone, at least the former.

Here is the data sheet circa 6/24/14
http://www.onsemi.com/pub/Collateral/KAF-18500-D.PDF

I had occasion to try the 75Lux for the first time on the Sony A7 and M9 in identical conditions, my love of the M9 look was only reinforced in a number of ways, not least color rendition. Unfortunately those are shots with privacy concerns but I hope to duplicate them in the near future and I will post examples eventually. All examples are welcome in this thread, for or against the KAF-18500 CCD Sensor.

One thing is for sure: a lot of shooters who own M9 variants and MMs are in love with the output of these CCD sensors often right out of the camera.

A certain “Lenshacker” known to many of us posted a more technical explanation than usual at L-camera for why we like the M9 ouput:

“I prefer CCD's as they output an analog signal without on-chip signal processing applied. The latter allows over-sampling, lower noise and can extend dynamic range via processing. The CCD delivers raw image data, unprocessed.”

Not that this precludes good CMOS results :

“.....With that stated, the sensor on the D4 and Df is good, and there is a monochrome version of the same sensor that is used in a Nikon microscope camera.”

But basically I’m beginning to understand that CCD is analog and CMOS is digtial, as camera sensors go. One good short primer is:
https://www.teledynedalsa.com/imaging/knowledge-center/appnotes/ccd-vs-cmos/

In short, from wikipedia:
“Each cell of a CCD image sensor is an analog device. When light strikes the chip it is held as a small electrical charge in each photo sensor. The charges are converted to voltage one pixel at a time as they are read from the chip. Additional circuitry in the camera converts the voltage into digital information.”

“A CMOS imaging chip is a type of active pixel sensor made using the CMOS semiconductor process. Extra circuitry next to each photo sensor converts the light energy to a voltage. Additional circuitry on the chip may be included to convert the voltage to digital data.”

“CMOS sensors can potentially be implemented with fewer components, use less power, and/or provide faster readout than CCD sensors. CCD is a more mature technology and is in most respects the equal of CMOS. CMOS sensors are less expensive to manufacture than CCD sensors.”

So there is there outline of the great 240 vs M9 debate, and direct comparisons are on topic in this thread. But let’s admit fantastic photography is done with both systems. That said, they are not the same.

Beyond that discussion, how good is the KAF-18500 as a CCD?

To quote NazgulKing in this forum:
"...In fact, as CCDs go by, the one in the M9 is far behind the best in CCD technology. CCDs used in scientific apparatus are far better than the one used in the M9."

The “Lenshacker” found this comment curious:
“Can you ask him to post a link to the data sheets that he is referring to? I keep up with this stuff, the CCD's from other manufacturers had substantially more dark-current than the KAF-18500, and did not have as high charge well capacity. As far as "off-the-shelf" CCD's used by the scientific industry, the KAF-18500 is as good as I've seen. I have friends that design sensors, but mostly for IR. SO- if this guy can really link to some data sheets that show numbers and not "fluff", the saturation count per pixel would have to be much higher than the 60K of the Kodak's 6.8um pixel.”

and he pointed to a comparable scientific CCD from another company:
http://www.teledynedalsa.com/imaging/products/sensors/area-scan/FTF6040C/
with the comment “Higher noise and lower saturation. Noise is about where the original M8 sensor was 8 years ago.”

I think here we are at the heart of the matter. Reliability issues aside, the M9 and MM, in many minds can produce unique digital imagery which we prefer to all alternatives today. Why? How? Why no other CCD alternative, since the sensors are fully in production and application in many technical areas today?

I’m not going to plead with users to stay on topic or refrain from bashing the M9, since that will happen no matter what. It’s OK. In fact some bashing backed up by technical knowledge or links in that direction is welcomed. Stereotypes about jewelry loving Leica users can expect a retort or two LOL

But this forum has a very wide readership with some users who really understand the hows and whys, and their opinions may inform ours

If you are like me, you will be googling all sorts of terms like "dark current" LOL

 

[jaapv]

Nazulkings comment is certainly curious as sensors for scientific ( and technical) purposes are not designed to the same parameters as CCDs for image recording. As far as I am aware there is no newer image - sensor out there.
On a sidenote, the price of many specialised scientific/technical sensors would multiply the price of any camera built with them ( if such a thing would exist)

 

[ferider]

But basically I’m beginning to understand that CCD is analog and CMOS is digtial, as camera sensors go.

May I suggest to read up on basic information theory first (a digital black box is exactly that, there is nothing analog about it), then consider the different dynamic ranges and IR sensitivities of M9 and M240, and only then look at more microscopic and much less relevant details like different sensor technologies, etc. ?

Roland.

 

[CameraQuest]

Its interesting to me that some say the M9's sensor give their favorite digital Leica results,

while others say the results are essentially the same once they became used to the M240 post processing.

Stephen

 

[YYV_146]

If you do enough processing and use your own color profiles, the only thing that matters for a camera sensor is dynamic range against iso-stops. If you always have access to controlled light, color depth is more important. But mostly DR, especially low-iso DR.

The A7's DR is, converted from log-scale, more than four times that of the M9 and about twice that of the M240. The A7r adds ~15% to each number.

 

[uhoh7]

Its interesting to me that some say the M9's sensor give their favorite digital Leica results,

while others say the results are essentially the same once they became used to the M240 post processing.

Stephen

I think the more you tweak your M9 shots the more this is true. It's very hard to make a 240 shot into a M9 OOC. I'm looking at both bodies everyday in the nice Leica images thread at FM, and the clean M9 shots are pretty distinct to everyone, including those who prefer the 240.

Some (not you) think everything can be done in photoshop. Not even close, but even if it were possible, many of us prefer to edit as little as possible.

 

[JoeV]

What his "CCD is analog" point implies to me is the possibility of a better A-to-D circuit than what might be built on-chip. As in high-end audio gear.

~Joe

 

[uhoh7]

If you do enough processing and use your own color profiles, the only thing that matters for a camera sensor is dynamic range against iso-stops. If you always have access to controlled light, color depth is more important. But mostly DR, especially low-iso DR.

The A7's DR is, converted from log-scale, more than four times that of the M9 and about twice that of the M240. The A7r adds ~15% to each number.

We each have our criteria and I respect yours, but having owned A7r, and owning a A7 today my own experience is the whole DR gap is myth to my workflow. I don't like to shoot the A7 over 1k iso ever, and it's not that great there, at least in LR. The Sony Raws make noise much faster than M9 DNG and the black M9 shadows push very well if you want that. There is no light in which i prefer output of A7 to M9. I just use faster glass in dim light.

But with a lens like the FE 35/2.8 maybe the DR and ISO performance is an advantage. You will be able to get more DOF in dimmer light. Real world I don't miss that aspect, but you well might

 

[ferider]

That just means you didn't expose right, depending on the camera and your output requirements.

 

[YYV_146]

We each have our criteria and I respect yours, but having owned A7r, and owning a A7 today my own experience is the whole DR gap is myth to my workflow. I don't like to shoot the A7 over 1k iso ever, and it's not that great there, at least in LR. The Sony Raws make noise much faster than M9 DNG and the black M9 shadows push very well if you want that. There is no light in which i prefer output of A7 to M9. I just use faster glass in dim light.

But with a lens like the FE 35/2.8 maybe the DR and ISO performance is an advantage. You will be able to get more DOF in dimmer light. Real world I don't miss that aspect, but you well might

Do you have in-body NR applied? I agree that Sony's NR can be too enthusiastic at times.

The M9 can only go so far. Having used one. I would not go over iso 800 for color critical work. For the A7 I would be willing to use 2500, and for the A7s even 8,000 is borderline acceptable.

Edit: It does occur to me that color noise is a greater concern on the A7 (less "film-like"?). I apply color noise NR and chroma noise NR separately in post.

 

[raid]

Its interesting to me that some say the M9's sensor give their favorite digital Leica results,

while others say the results are essentially the same once they became used to the M240 post processing.

Stephen

I would trust most statements by users of both cameras at the same time. They can hopefully sense and see any differences if they exist.

 

[jaapv]

We each have our criteria and I respect yours, but having owned A7r, and owning a A7 today my own experience is the whole DR gap is myth to my workflow. I don't like to shoot the A7 over 1k iso ever, and it's not that great there, at least in LR. The Sony Raws make noise much faster than M9 DNG and the black M9 shadows push very well if you want that. There is no light in which i prefer output of A7 to M9. I just use faster glass in dim light.

But with a lens like the FE 35/2.8 maybe the DR and ISO performance is an advantage. You will be able to get more DOF in dimmer light. Real world I don’t miss that aspect, but you well might

The whole DR range argument is the same marketing blah as the megapixel argument. Film had about 13 stops, the M has about 13 stops, and the Sony indeed considerably more. However, you need to be pretty good at printing to get nine stops onto paper, so the only advantage of those high dynamic ranges is that one can afford to expose sloppily....

 

[Pioneer]

The whole DR range argument is the same marketing blah as the megapixel argument. Film had about 13 stops, the M has about 13 stops, and the Sony indeed considerably more. However, you need to be pretty good at printing to get nine stops onto paper, so the only advantage of those high dynamic ranges is that one can afford to expose sloppily....

Amen. There is a great deal of marketing hype that is bandied about as being critical to photography. Dynamic range and ISO capability are just a couple. I have seen some awesome prints made from I-Phone output and no one will attempt to argue that the I-phone has top of class dynamic range or iso.

A huge part of the arguments today are attempts by the digital world to convince people that the numbers are the most important part of the equation. To them, perhaps they are. But to many others the color rendition and tonal range is critical to their work.

Subject, composition and your perception of those is the toughest thing to get into the print.

 

[noimmunity]

I had the M9 and the M8 together for a short period. For pure "sparkle", I found more magic in the output of the M8's KAF-10500 than in the M9--in color as well as B&W.

The M9 produces by default high contrast color images. The output more easily lends a dramatic effect to otherwise pedestrian landscapes.

Shooting both the M8 and the M9 side-by-side against color positive film, I found nothing in images from either one to convince me to leave the latter behind.

 

[Pioneer]

...Shooting both the M8 and the M9 side-by-side against color positive film, I found nothing in images from either one to convince me to leave the latter behind.

Goodness gracious, I hope not. Why would you want to do that?

I have shot CCD and CMOS in various formats from small sensors through medium format. They can produce great results but they certainly do not replace color positive...or most any film. Film and digital are very different technologies, both with their own strengths and weaknesses.

Why do we have to throw one out the window because we happen to enjoy the other? Last I checked this wasn't meant to be a monogamous relationship.

 

[noimmunity]

Goodness gracious, I hope not. Why would you want to do that?

I have shot CCD and CMOS in various formats from small sensors through medium format. They can produce great results but they certainly do not replace color positive...or most any film. Film and digital are very different technologies, both with their own strengths and weaknesses.

Why do we have to throw one out the window because we happen to enjoy the other? Last I checked this wasn't meant to be a monogamous relationship.

Thanks for giving me a good laugh and a friendly reminder!

When I wrote that post I was really thinking about landscape applications. I wrote a whole paragraph on why I gave up on the M8/M9 for landscape photography (a decision that was really about my own needs and parameters than anything else) before I cut that out. I guess I forgot to revise the rest of the post.

 

[jaapv]

Amen. There is a great deal of marketing hype that is bandied about as being critical to photography. Dynamic range and ISO capability are just a couple. I have seen some awesome prints made from I-Phone output and no one will attempt to argue that the I-phone has top of class dynamic range or iso.

A huge part of the arguments today are attempts by the digital world to convince people that the numbers are the most important part of the equation. To them, perhaps they are. But to many others the color rendition and tonal range is critical to their work.

Subject, composition and your perception of those is the toughest thing to get into the print.

I often wonder at those forum participants that chase the latest highest numbers and ratings in cameras. It seems that they are only able to make a decent photograph using the newest, highest ISO, highest MP, highest DR camera. They must be so marvelously good at the craft for this to make a difference. However, it makes me think that their images must have looked like sh!t when they were forced to use yesterday’s junk....

Let’s dub it “The Princess and the Pea Syndrome”

 

[ferider]

... There is a great deal of marketing hype that is bandied about as being critical to photography. Dynamic range and ISO capability are just a couple. ..... But to many others the color rendition and tonal range is critical to their work....

Let's go back on thread subject, I.e., assigning some "analogue magic" to the M9 CCD sensor when compared to more modern CMOS sensors (A7 and M240), and comments on the "better" dark tones and color rendering of the M9.

I'm sure the OP really observed these differences but I'm also convinced that there are reasons much more prominent than "magic" sensor differences. For instance:

- different IR response (color rendering)

- wrong exposure. Meaning the OP might not have used the whole capability of his CMOS sensors. 2 more bits DR correspond to a 4 times wider range of dark tones, This has nothing to do with only being able to print 9 bits or less, it has _all_ to do with tonal range.

No marketing hype for digital latest and greatest from me. Some professional semiconductor engineering baggage though.

Roland.

 

[Pioneer]

Let's go back on thread subject if we can, namely assigning some "analogue magic" to the M9 CCD sensor when compared to more modern CMOS sensors (OP mentioned A7 and M240), and specifically comments on the "better" dark tones and color rendering of the M9.

I'm sure the OP really observed these differences but I'm also convinced that there are reasons much more prominent than "magic" sensor differences. In particular, I'm guessing:

- different IR response (color rendering)

- wrong exposure. Meaning the OP might not have used the whole capability of his CMOS sensors. 2 more bits DR correspond to a 4 times wider range of dark tones, if exposed correctly. This has nothing to do with only being able to print 9 bits or less (which I agree with), it has _all_ to do with tonal range.

Different workflow components such as exposure method, ir filtering (or not), possible dng lossful compression, etc., are much more important than different sensor types, and I'm guessing the OPs workflow fits the M9 much better, without any "magic" or "analogue" behavior of the underlying sensor.

No marketing hype for digital latest and greatest from me. Some professional semiconductor engineering baggage though.

Roland.

I don't think anyone is trying to deny the numbers themselves, I am personally quite certain that those differences do exist. Neither am I trying to say that the M 240 is not a good camera. I personally used one for almost a month and found it to be a terrific camera. The question though, is how important are those small numerical differences to actual, real world photography.

Or are we really dealing with a Princess and the Pea Syndrome, as so quaintly labeled by our friend Jaapv?

For myself, I don't really know. I do know that there is a visible difference between what I have seen in my M9 files and my M 240 files. I guess it could be IR response. I think it less likely that exposure techniques provide an adequate explanation. I personally used several different methods of exposure, including the camera's own meter as well as the use of a separate, reliable, spotmeter.

In the end I was less impressed with the output of the M 240 than I was with that of the M9. I know that I was literally blown away by the color output of the M9 when it first arrived. I still am. The only other digital camera that has impressed me as much has been my DP2 Merrill.

 

[ferider]

Let me put it like this: it walks and quacks like a duck ... no need for magic

- you see the differences
- the numbers are significantly different.

In the past, when shooting slide vs B+W film, did you expose differently (highlights vs shadows) ? What did you do on M9 vs M240 ?

Or like this:

Today my wife took a red velvet cake (deep purple really) out of the oven with a red glove. I'll post a photo tomorrow, taken with the 240. On the photo, the glove is bright red, but the cake is light brown ... Note that I don't mind really. It's just part of the learning curve.

Thanks,

Roland.