400 Leica photographers agree: we love CCD!

[Rick]

Advertisement (gone after registration)

Comparing the spectral response of the two sensors would give a more definitive answer with regard to color rendition of the sensor versus need for post-processing. It would be nice to compare with the M9 and M8 curves. Those are published.

 

It would be nice to quantify the IR contamination for the sake of the OP's main issue with M240 skin tones.

 

You are straying from the main point of this thread, where the OP seems to indite the CMOS sensor while ignoring everything else in the image processing chain including the CFA and the in-camera image processing. Now you are straying into the issue of the differences in the thickness of the sensor IR glass.

 

Rick

[Advertisement]

The Leica Q3 is available for pre-order at B&H Photo Video and Adorama

[Jeff S]

Peter - In light of the fact that Leica has no intention of going back to CCD technology, maybe, a more fruitful approach, would be to write a letter to Leica describing the exact differences that you like in the M9 output and request Leica to incorporate these characteristics into the image processing done in the next generation M.

 

Good idea, but that might require some samples.

 

Jeff

[jaapv]

Comparing the spectral response of the two sensors would give a more definitive answer with regard to color rendition of the sensor versus need for post-processing. It would be nice to compare with the M9 and M8 curves. Those are published.

 

It would be nice to quantify the IR contamination for the sake of the OP's main issue with M240 skin tones.

Although you are certainly right that IR contamination severely impacts skin tones which needs sophisticated Photoshop techniques to correct, if at all possible, it is sufficient to know it is happening and to make use of a filter IMO.

As it is, it has been discussed here before (often!) and consensus is that the IR suppression is about 50 % on the M8, 80 % on the M9 and 70 % on the M240.

 

If we consider the anatomy of skin in as far as it is relevant to photographic rendering we find a layered structure. A fatty base layer, opaque yellowish white, then a layer of capillary veins red to blueish, depending on oxygen saturation and with a density depending on the distribution of veins and arteries. Highly IR reflective. Then Melanin particles which are visible light absorbent, UV reflective and providing toning from Finnish nothing to the blackest African skin, covered by a translucent white layer.

 

This means a large amount of metamerism and a virtually impossible exact rendering.

Edited February 19, 2015 by jaapv

[Lenshacker]

You are straying from the main point of this thread, where the OP seems to indite the CMOS sensor while ignoring everything else in the image processing chain including the CFA and the in-camera image processing. Now you are straying into the issue of the differences in the thickness of the sensor IR glass.

 

Rick

 

The difference in IR response is a function of the sensor and of the IR filter over it. We have established that. You can read some of the links posted.

 

The OP has stated that he wants to uncover the potential of the M240, he is on his third one. The IR sensitivity may be the root of his problem. Banding may also be a problem.

 

Peter: Unless you have dumped your current M240,

 

1) Take a picture of an IR remote control (for a TV) with the M9 and M240 without using IR blocking filters.

 

2) With the M240: set it to ISO6400 and take an existing light photo with the camera on Manual Exposure. Have a scene with a lot of shadows and some bright spots. You are looking for banding issues. Then set the camera to Base ISO without changing the settings. Push the underexposed image to the point where it matches the ISO6400 shots. We are trying to determine if there is underlying banding in the Base ISO shots, or if they are an artifact of High-ISO. This has to do with where the banding is introduced within the Sensor.

[Lenshacker]

Although you are certainly right that IR contamination severely impacts skin tones which needs sophisticated Photoshop techniques to correct, if at all possible, it is sufficient to know it is happening and to make use of a filter IMO.

As it is, it has been discussed here before (often!) and consensus is that the IR suppression is about 50 % on the M8, 80 % on the M9 and 70 % on the M240.

 

These numbers are completely wrong.

 

The IR suppression is far greater than your numbers shown, these numbers are made up. You can find actual numbers for IR suppression in the datasheets for the KAF-10500 and KAF-18500. You can also find the numbers in the S8612 cover glass spectral curves.

 

1/10th Second at F3, ISO200:

 

full_spectrum2

 

Infrared modified Olympus EP2.

 

Leica M8, 1/8th second at F1.5, ISO160.

 

M8_nocut

 

According to your numbers, the M8 picture should be twice as bright as the Olympus. 50% cut would be 1 F-Stop.

Edited February 19, 2015 by Lenshacker

[jaapv]

Whatever, the 80% and 50% are from Leica publications/interviews at the time of the introduction of the M9. As I said, they hold absolutely no interest to me, other than the knowledge that I always need IR filters on the M8, rarely on the M9 and somewhat more often on the M240.

[Prosophos]

Advertisement (gone after registration)

Peter, as you know I have every sympathy for your crusade, but in this thread we have seen the most knowledgeable members of this forum, like Michael, technical writer for LFI, Johnathan, prominent beta tester for Leica, Olaf, Chief Pundit of this forum, Andy Piper, Stefan Daniel, head Honcho of the Leica M division quoted ,and many others, to explain that any differences you may see are not down to sensor technology, but a number of other things.

 

The only difference you may see is a different noise character, as the CMOS has a more even per pixel output, easily cured in Photoshop by adding noise.

As Jono explained there may be an OOC microcontrast difference caused by a lower CCD DR, again, easily changed in Photoshop.

 

All the rest like colour character is solely down to different dyes in the Bayer filter, different IR filter and different processing/interpolation algorithms in the camera.

 

Your basic assumption is wrong and has led to the wrong conclusion.

 

Here is the part I find interesting:

 

Initially, many believed there weren't any differences, that is, that M240 files could be made to look like M9 files. Now, some (finally) concede that this is false. Progress, I guess.

 

Most of you who have decided to "pile-on" the latter part of this thread, however, still believe that M240 files can be made to look like M9 files. The argument has shifted to "it's not about CCD vs CMOS" but about "different dyes in the Bayer filter, different IR filter and different processing/interpolation algorithms in the camera" and manipulations in post processing (or with IR filters) can overcome these differences and make the files look identical.

 

Well, you're wrong.

 

If you still believe that the files can be made to look the same (or behave the same during post processing) we are at an impasse because I completely disagree with you - 100%.

 

At the risk of offending many, I will honestly tell you that there is no one on the list of individuals you've cited above Jaap who I have confidence in with respect to their ability to discern the nuances of which I speak — those little differences that irritate me to no end with respect to M240 vs M9 files.

 

Those whose photographic eye I do trust, however, including many of the 550 or so signatories of my letter, do see a difference. And I stand with them.

 

Let's please be civil from here on end and avoid the smarmy winky emoticons in our subsequent posts, shall we?

[Paulus]

Here is the part I find interesting:

 

Well, you're wrong.

 

If you still believe that the files can be made to look the same (or behave the same during post processing) we are at an impasse because I completely disagree with you - 100%.

 

Let's please be civil from here on end and avoid the smarmy winky emoticons in our subsequent posts, shall we?

 

I remember an article of Erwin Puts testing the M9 about 5 years ago ( this long already? ) saying that the M9 could produce decent files already without a lot of post processing. This was the reason I liked the camera. In the time that we did not got a "lightroom" for free, I really didn't like to PP. Still don't. The only post processing I realy like is sitting in the basement and shaking my scales of developper and fix.

 

I always thought that it was a goal for Leica to make a camera that did not need PP. In the M9 they succeded in my point of view. With its downsides surely.

 

I realy believe that anybody is entiteld to ask passionately for a successor of the M9 + ccd. I hope that a " simple " camera still will exist in the future.

[jaapv]

Well, it might have been better if you had not specified your petition on the wrong assumption that the differences you see (and nobody here has denied that there are differences to be seen, at least on files with little processing) are due to any difference between CCD and CMOS.

 

Had you asked for a camera that produces a similar output as the M9 regardless of sensor technology you might have gotten a more positive response.

 

Btw I don't see any smilies in the post you quote. Maybe my computer has a different processor from yours?

Edited February 19, 2015 by jaapv

[Lenshacker]

Whatever, the 80% and 50% are from Leica publications/interviews at the time of the introduction of the M9. As I said, they hold absolutely no interest to me, other than the knowledge that I always need IR filters on the M8, rarely on the M9 and somewhat more often on the M240.

 

Do you have any links to these statements? 50% is one stop, 70% ~2 stops, and 90% ~3 stops. If that were the case, IR contamination would destroy the color rendition of all of these cameras. I will test the M8 with/without an 88a filter.

 

For what it is worth: I agree with Peter. CMOS sensors perform on-chip signal processing. CCD sensors are known for uniformity. There are differences that are salient to the design of each type of sensor. Additionally, the implementation of the CMOS sensor used in the M240 demonstrates higher levels of IR contamination and Banding than many other CMOS sensors on the market. These problems need to be solved for the next generation of Digital M camera, whatever technology is used.

[jaapv]

Do you have any links to these statements? 50% is one stop, 70% ~2 stops, and 90% ~3 stops. If that were the case, IR contamination would destroy the color rendition of all of these cameras. I will test the M8 with/without an 88a filter.

 

For what it is worth: I agree with Peter. CMOS sensors perform on-chip signal processing. CCD sensors are known for uniformity. There are differences that are salient to the design of each type of sensor. Additionally, the implementation of the CMOS sensor used in the M240 demonstrates higher levels of IR contamination and Banding than many other CMOS sensors on the market. These problems need to be solved for the next generation of Digital M camera, whatever technology is used.

Look for the interview with Stefan Daniel on the introduction of the M9. It is buried somewhere in this forum, late 2009.

 

I do agree with the last part of your statement, especially with the last few words.

It will be a hard job for Leica and CMOSIS though, because the heritage design of the M system imposes severe restrictions on the filter array and acceptance angle.

Edited February 19, 2015 by jaapv

[IWC Doppel]

Thank you for your reply. But, you didn't answer my question of whether you liked the M9 output for monitor or prints? Are you judging it by monitor or print, or both. And, what is it that you see?

 

I print a lot, as well, and sometimes big 44" and I profile my own papers and calibrate the printer often and have a color matched workflow etc. etc. and I don't see differences between prints of my M cameras, when printed.

 

I do see differences in how I PP and I suppose that if I did no PP and just printed the output of the M8, M9, and M240 there would be obvious differences. But, that is not a reality of working in PP to achieve the look I want for the print. Often I make test strips like I did when wet printing and I look at the different PP strips in the light they are going to be displayed and make choices in PP for specific prints.

 

My point being, the print is never anything like what the output of the camera hands off to me. I just can't understand what it is people see from different cameras other than color hue differences in, say, skin color and most of this can be equaled out in PP.

 

But, I will admit some of it is not possible to completely change in PP, like the differences seen in skin tone between the Sony-look and the Leica IR contamination artifacts. Even then, they are so small as to be insignificant, to me.

 

Rick

 

 

To be fair I haven't printed from an M240 but have seen a number of prints and a lot of screen shots. For me there is something about the M9 CCD files I like and differences are greater than those found in lens rendering subtleties

 

They won't make better pictures but for me I like the look n that'll keep me happy.

 

Digital is a complex thing, much more so than I certainly understand

[Rick]

To be fair I haven't printed from an M240 but have seen a number of prints and a lot of screen shots. For me there is something about the M9 CCD files I like and differences are greater than those found in lens rendering subtleties

 

They won't make better pictures but for me I like the look n that'll keep me happy.

 

And, this is where we cordially part ways. I can't believe in something that is described as "something." I respect you believe in what you see, I just don't see the difference in print, although out of camera there are differences that for the most part I can equal out if I wanted. And, the "something" I see I can directly attribute to IR contamination (see previous and numerous threads).

 

In the final analysis, the advantages of the M240 CMOS sensor outweigh any small changes between the two cameras, for me.

 

 

Digital is a complex thing, much more so than I certainly understand

 

 

Belief always steps in when understanding wanes, and this topic is almost taking on a religious tone. And, I'm sorry I just can't add my signature to something as abstract.

 

Again, I'd suggest an open letter to Leica that is specific. For example, to encourage them to cut as much IR contamination as possible, but I'm sure they are walking that delicate line between sensor glass that is too thick and causes smearing and glass that is too thin and causes color problems. I really think this is where these differences that folks see in the color are originating. Leica already understands this, though.

 

If. there is a good to this letter, it reminds Leica that color hue is important and we noticed the changes in the M240 and the thinner glass.

 

Hope I haven't offended anyone and 'll wait and see what Leica does with the new M.

 

Rick

[IWC Doppel]

Rick, it has been covered by me and others I'm not going to write a comprehensive essay, but to highlihht a few

 

In camera sharpening (I believe the CMOS loses here as the more distant objects are enhanced and lose the effects of distance, looking notably flatter in some images and notably less natural)

Colour, the density and richness is just not there I couldnt recreate in LR

The graduation out of black, not dynamic range. To me this simply looks more real

 

None of the above has anything to do with religion. The final judgement is the overall look and they do look different and I do prefer M9 files. I will try the new M when it materialises, but not expecting to fall in love, but hey you never know

[Prosophos]

And, this is where we cordially part ways. I can't believe in something that is described as "something." I respect you believe in what you see, I just don't see the difference in print, although out of camera there are differences that for the most part I can equal out if I wanted. And, the "something" I see I can directly attribute to IR contamination (see previous and numerous threads).

 

In the final analysis, the advantages of the M240 CMOS sensor outweigh any small changes between the two cameras, for me.

 

 

 

 

 

Belief always steps in when understanding wanes, and this topic is almost taking on a religious tone. And, I'm sorry I just can't add my signature to something as abstract.

 

Again, I'd suggest an open letter to Leica that is specific. For example, to encourage them to cut as much IR contamination as possible, but I'm sure they are walking that delicate line between sensor glass that is too thick and causes smearing and glass that is too thin and causes color problems. I really think this is where these differences that folks see in the color are originating. Leica already understands this, though.

 

If. there is a good to this letter, it reminds Leica that color hue is important and we noticed the changes in the M240 and the thinner glass.

 

Hope I haven't offended anyone and 'll wait and see what Leica does with the new M.

 

Rick

 

Rick, you have taken a very constructive approach in your last few messages and I want to thank you for that (I am being sincere).

[Jeff S]

FWIW, and a lot of people forget, the forum was full of complaints about the initial output of the M9….poor color results, resorting to use of IR filters after disdain for them on their M8, etc. After some Leica fixes, and some user experience, the furor subsided.

 

We'll see if future generations of the M, and some more time in the hands of users, will result in some happier customers who now prefer M9 output. I doubt, though, that this will be because of a return to CCD. But Leica has fooled us before…for better (said full frame M couldn't be done), or worse (said the R10 will be done), and so on. Most here chuckled at the potential for a production model monochrome M.

 

Jeff

[jonoslack]

Hi There Peter

 

If you still believe that the files can be made to look the same (or behave the same during post processing) we are at an impasse because I completely disagree with you - 100%..

 

Well, I agree with you - I don't think you can make the M240 files look like the M9 files (or vice versa) It's like trying to make Tri-X look like HP5

 

At the risk of offending many, I will honestly tell you that there is no one on the list of individuals you've cited above Jaap who I have confidence in with respect to their ability to discern the nuances of which I speak — those little differences that irritate me to no end with respect to M240 vs M9 files.

 

Every scene imposes a new set of variables - every image you take. I certainly don't see the nuances which you find so irritating . . . but I can certainly find irritations in the other direction. Peter - you may have no confidence in my ability to discern the nuances which irritate you . . . but I don't have much confidence in your ability to discern the subtleties which I see in the M240 images which are simply missing in the M9.

 

This is subjective stuff - it's not about CCD/CMOS it's about microlens design, bayer filter design, demosaicing etc. etc. etc.

 

Those whose photographic eye I do trust, however, including many of the 550 or so signatories of my letter, do see a difference. And I stand with them..

 

Well, we always like to trust those who agree with us

 

Let's please be civil from here on end and avoid the smarmy winky emoticons in our subsequent posts, shall we?

 

Or possibly casting aspersions on the abilities of people to discern subtleties in images?

 

Seriously Peter - I absolutely subscribe to your right to prefer the images from the M9 . .and I certainly do see a difference . but I absolutely don't subscribe to your opinion that the agreed differences are to do with the CCD/CMOS argument

[jonoslack]

Rick, it has been covered by me and others I'm not going to write a comprehensive essay, but to highlihht a few

 

In camera sharpening (I believe the CMOS loses here as the more distant objects are enhanced and lose the effects of distance, looking notably flatter in some images and notably less natural)

 

I don't know for sure but I don't think there is any sharpening in RAW files - either near or far

Colour, the density and richness is just not there I couldnt recreate in LR

The graduation out of black, not dynamic range. To me this simply looks more real)

 

Colour is to do with the Bayer filter and the demosaicing - not to say that you haven't an absolute right to prefer the M9 . . .but it's nothing to do with the CCD/CMOS argument

None of the above has anything to do with religion. The final judgement is the overall look and they do look different and I do prefer M9 files. I will try the new M when it materialises, but not expecting to fall in love, but hey you never know

 

I'm sorry - it's an act of faith to conclude that the differences between the M9 and the M240 are a function of the CCD/CMOS change. Of course, I could change my mind if you could produce empirical and reproducible evidence - but as far as I'm aware nobody has done that (and any such evidence would have to range much wider than the differences between the M9 and the M240)

 

Your right to prefer the M9 images is nothing to do with religion - and it's absolutely your prerogative and I would certainly not do anything further than disagree . . .

Edited February 19, 2015 by jonoslack

[Advertisement]

Hallo,

Look here: Leica M

[Lenshacker]

The IR filter used for the M8 transmits less than 5% of light over most of the IR region. A quick test with an 88a visible blocking filter verifies this number. At ISO160 and F3, an exposure of 1/1000th second without the 88a filter changes to 1/6th second with the filter. Figure 2 stops loss for the visible portion of the spectrum for a sensor without the IR blocking filter, "roughly" 1/4th of the sensitivity of the Kodak CCD's were in the IR region. So about 4~5 stop drop due to the Kyocera BS7 glass.

 

See page 28:

 

KAF-10500-CXA-JH-AE PDF Datasheet - Kodak Image Sensors - Datasheets360.com

 

The S8612 cover glass of the M9 cuts out 99.9% or better.

 

http://www.us.schott.com/advanced_optics/english/download/schott-bandpass-s-8612-dec-2014-en.pdf

 

Either the statements regarding 50% and 90% transmission were kind of like "Think of the M8 as being 50%, then the M9 is 90%" just as a figure of speech, or the person making the statement knew very little about IR blocking filters.

 

CMOS sensors are hybrid analog/digital devices, and modern CMOS sensors incorporate signal processing on the chip. The degree of processing varies greatly among manufacturers and devices. These signal processing algorithms are applied to the image before they leave the chip. If you like the results of these signal processing algorithms on the image, you are in luck. If you do not like the resulting image- backing out the original image is not easy.

 

CCD sensors are analog devices, apply a minimum of massaging before outputting the image as an analog signal. The M9 applies a minimum of processing to these images, what you get is close to a raw output from the sensor. The M8 was hampered by the lossy DNG-8 compression algorithm, which was done off-chip. M8 RAW mode restores the capability of the sensor. I'm going to guess that I'm the only M8 user writing code in FORTRAN to process the M8 images, but maybe I'm wrong about that.

Edited February 19, 2015 by Lenshacker

[Prosophos]

Hi There Peter

 

 

Well, I agree with you - I don't think you can make the M240 files look like the M9 files (or vice versa) It's like trying to make Tri-X look like HP5

 

 

 

Every scene imposes a new set of variables - every image you take. I certainly don't see the nuances which you find so irritating . . . but I can certainly find irritations in the other direction. Peter - you may have no confidence in my ability to discern the nuances which irritate you . . . but I don't have much confidence in your ability to discern the subtleties which I see in the M240 images which are simply missing in the M9.

 

This is subjective stuff - it's not about CCD/CMOS it's about microlens design, bayer filter design, demosaicing etc. etc. etc.

 

Hi Jono.

 

For what it's worth, I've never had any issues with you and my apologies if my comments were harsh.

 

I know you agree that there are differences in the files that cannot be resolved with post-processing, but you have to admit there are many in this forum that have gone on record as saying otherwise. Some have even claimed it's just a simple WB issue that can be resolved in a few key strokes. As you acknowledge, it's more than that.

 

I'm on my 3rd M240. Why? I love its haptics. It's truly a pleasure to use.

 

I'm learning to live with its output. I don't think I'll ever get it to where I like it, but I've accepted that.

Edited February 19, 2015 by Prosophos