M9, M9M, Do It Yourself Comparison

[pico]

Advertisement (gone after registration)

Thanks to Jono for making the M9 and M9M images of his bicycle!

 

Below are links to an image you might want to play with and show the rest.

 

These two links of the same image (one a layered TIFF, the other a layered PSD) area of the Actual Pixels from approximately the center of both images. It has four partitions of the image - all are done with no ACR pre-processing.

 

Top row is labeled 'nativ', and is locked in PSD to help you keep from changing the wrong squares. The bottom row is of two layers, each unlocked. Feel free to manipulate either cell of the bottom row.

 

http://www.digoliardi.net/m9_m9m_comp.tif

http://www.digoliardi.net/m9_m9m_comp.psd

 

I must admit that I found the results of modifying the M9 color image to B&W by mixing channels to be far more satisfying than the M9M image regardless of processing, and the M9 image is just as good for printing, regardless of the 'math', however my mind is open, but my brains aren't falling out yet. (Alan - I'm completely rethinking the issue. Thanks for your persistence on this.)

Edited June 27, 2012 by pico

[AlanG]

(Alan - I'm completely rethinking the issue. Thanks for your persistence on this.)

 

I really appreciate that. I was not trying to argue with anyone (although that seems to become inevitable on these kinds of posts sometimes) but merely present the adjustment possibilities that are inherent in the existing color images when going to b/w. I don't do much b/w work anymore but once I started using the b/w conversion controls in C1 my eyes were opened and it was easy to see that these controls did not degrade the image quality. So all of this was a revelation to me also. I got so into it that I actually compared quite a few images adjusted in numerous ways before converting to b/w. But I hesitated in posting them to this forum as I felt I was already beating folks over the head with it.

 

I come to this from both a technical and applied photography background and could see that some concerns that people were getting hung up on might at most just amount to splitting hairs in practice.

Edited June 27, 2012 by AlanG

[Advertisement]

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

[AlanG]

I opened it in Photoshop and then did these steps:

 

Clicked on Layers and selected "m9 adjusted"

 

Clicked on Image -> Adjustments - > Black & White (This previews the color image as b&w)

 

Then I adjusted the color sliders to see the effect of moving those sliders compared to the MM image.

 

--------------- I am a bicycling enthusiast, but I had no idea that is part of a bicycle.

Edited June 28, 2012 by jaapv
consistency

[Guest malland]

I just downloaded both files and looked at them, granted, very briefly but concluded that this comparison doesn't show me what I need to see, which is two full DNG files from the M9 and M-Monochrom that I could compare in processing.

 

Using a tiny, pixilated M-Monochrom file doesn't work for me because my conclusion about the M-Monochrom is that I can manipulate the files to a much greater extent, getting the relatively high contrast and deep, rich blacks that I want in combination with maintaing the mid tones that I want. I found this particularly easy to do in Lightroom 4, using the new Black Slider (as suggested by fotofgrafz-Marc) together with lifting the shadow and mid-tones.

 

Working with the M-Monochrom files that Jono Slack provided gave me the feeling that I can get to a look that I haven't been able to achieve with any digital camera, including the M8 and the M9, although I have been there with film.

 

—Mitch/Bangkok

Scratching the Surface

Edited June 28, 2012 by jaapv
consistency

[pico]

I just downloaded both files and looked at them, granted, very briefly but concluded that this comparison doesn't show me what I need to see, which is two full DNG files from the M9 and M-Monochrom that I could compare in processing.

 

Then download the two original DNG files that Jono posted. One M9, the other M9M.

[jaapv]

A completely futile waste of time. If I want to make a photograph look like a converted M9 image I am better off using an M9. If I want a photograph looking like TriX I will use TriX. If I want to make a creative journey with photographs that look like MM photographs I will need an Mm. I have no interest in lookalikes.

When I will be publishing DNG comparisons it wil be to show the differences.

Edited June 28, 2012 by jaapv

[pico]

Advertisement (gone after registration)

I think the point of the download image has been lost. It is to offer anyone the opportunity to manipulate the MM and M9 images - the M9 to monochrome and the MM as you wish in order to compare at the actual pixel level. The images are layered to make it simple to do such. Perhaps the virtue of using M9 images rapidly converted to B&W is actually intimidating to those who think they want an MM?

Edited June 28, 2012 by pico

[jaapv]

No. I have an M9 and will have an MM I will use each as appropriate. This kind of use a screwdriver as a hammer argument holds no interest for me. I can understand that it may be useful to get close to the look of an M from an M9 file if one does not own one.

[pico]

No. I have an M9 and will have an MM I will use each as appropriate. This kind of use a screwdriver as a hammer argument holds no interest for me. I can understand that it may be useful to get close to the look of an M from an M9 file if one does not own one.

 

And I can understand how an owner of an Mm can resent how an M9 user can get equal and even superior B&W results. It goes both ways.

[Guest malland]

Pico, sorry, I am lost as to what the fight between Jaap and you is about.

 

In your response to me are you referring to the picture of the bicycle in the grass that Jono provided in M9 and M-Monichrom versions? If that is the case, then, I've looked at those two files some time ago; and they simply aren't the type of image (in terms of light, etc.) that will show me what I want to know in terms where I can "push" M-Monochrom images to that will be different from what I could do with an M9. So, if I understand correctly, I come down on Jaap's side in this discussion,

 

—Mitch/Bangkok

Pak Nam Pran: From Fishing Village to...

[jaapv]

:confused: I am and will remain an M9 user - so resentment?? I don't see any in my posts. Just lack of interest. Edited June 28, 2012 by jaapv

[pico]

Pico, sorry, I am lost as to what the fight between Jaap and you is about.

 

In your response to me are you referring to the picture of the bicycle in the grass that Jono provided in M9 and M-Monichrom versions? If that is the case, then, I've looked at those two files some time ago; and they simply aren't the type of image (in terms of light, etc.) that will show me what I want to know in terms where I can "push" M-Monochrom images to that will be different from what I could do with an M9. So, if I understand correctly, I come down on Jaap's side in this discussion,

 

—Mitch/Bangkok

 

I see. The image did not have a great deal of range - shadow to highlight. Perhaps we could use another set which is closer to our typical encounter. Another point is that we do not know what the final version of the Mm will be unless Leica does not change it from the one tested by Jono and others.

[brianv]

So I'm guessing that no one has tried making color images with the M9m yet?

 

Now that would be an interesting thread. Color with the M9 monochrome using a color wheel versus the M9 with the Bayer pattern Mosaic Filter. 54MPixels and no color artifacts from the Mosaic filter.

 

I guess the world will have to wait.

 

Just to add- I have seen some computer controlled color wheels on the market- but for monochrome telescope cameras. Kodak made one years ago for the DCS series of monochrome digital cameras.

Edited June 28, 2012 by brianv

[pico]

So I'm guessing that no one has tried making color images with the M9m yet?

 

Now that would be an interesting thread. Color with the M9 monochrome using a color wheel versus the M9 with the Bayer pattern Mosaic Filter. 54MPixels and no color artifacts from the Mosaic filter.

 

I guess the world will have to wait.

 

Just to add- I have seen some computer controlled color wheels on the market- but for monochrome telescope cameras. Kodak made one years ago for the DCS series of monochrome digital cameras.

 

We did discuss that early in an Mm thread, but with a rather tongue-in-cheek way. Later, you answered our question (perhaps Alan's question) regarding color separation filters (Wratten 25, 47b, and 58). Let us propose a Leica tri-color camera.

[AlanG]

Pico, sorry, I am lost as to what the fight between Jaap and you is about.

 

In your response to me are you referring to the picture of the bicycle in the grass that Jono provided in M9 and M-Monichrom versions? If that is the case, then, I've looked at those two files some time ago; and they simply aren't the type of image (in terms of light, etc.) that will show me what I want to know in terms where I can "push" M-Monochrom images to that will be different from what I could do with an M9. So, if I understand correctly, I come down on Jaap's side in this discussion,

 

—Mitch/Bangkok

Pak Nam Pran: From Fishing Village to...

 

Well one example cannot necessarily provide all of the info everyone may want. But you should be able to look at the files and see if one has less noise. That should give you an idea if you can "push" it... bring up detail in the shadows I presume. Although from what I've seen of MM images and M9 converted images I think they often have too much shadow detail, tending on the flat side, and need to go the other way. In which case noise would be suppressed a bit.

 

However, I don't see why the MM would have some kind of ability to record a range of greyscale tones significantly differently than an M9 can. How could it do that? It uses the same sensor and the rest is up to your tone curve and anything else you do to the file.

 

The point of this example is simply to show the ability of tonal adjustments via color controls in a convenient way that can give a side by side comparison. And also to show the resolution. If you don't have the software for this or don't take the time to adjust and study them, then you won't learn much from a quick look other than that the resolution appears similar in this shot.

 

I think if an objective person examined and adjusted a number of comparative examples (some at higher ISOs too) and determined from them that the M9 files will satisfy the results they require, then they might not see justification for buying an MM. On the other hand it might show up some differences that would sell them on the MM. Without this kind of study and analysis, how could one definitively know that the MM will perform better for one's needs?

Edited June 28, 2012 by AlanG

[AlanG]

So I'm guessing that no one has tried making color images with the M9m yet?

 

Now that would be an interesting thread. Color with the M9 monochrome using a color wheel versus the M9 with the Bayer pattern Mosaic Filter. 54MPixels and no color artifacts from the Mosaic filter....

 

 

Of course the concept of shooting three separation images is not so practical when compared to the much simpler method of shooting 3 or more color images and stitching them for much higher res. And it isn't as if a rangefinder camera that lacks live view would be anyone's top choice for high end product photography such as jewelry, watches, perfume, etc. regardless of its resolution. So for what application would you want to use such a camera? BTW, Sinar's early digital cameras were three shot cameras with a moving filter system built in. That solution faded pretty quickly once larger color chips and scan backs evolved.

 

Can you explain why a filter in the optical path would be so much better than one that is over each individual pixel of the sensor where it cannot impact resolution? Now maybe there are other issues caused by differences in what lies over the two sensors but isn't that the point of these kinds of tests... to reveal whether they exist and bother you? That is where theory meets practice.

 

I am not sure what assumptions some are using to conclude that simply by using an MM their b/w images may be better. (I guess it also depends on how you think you'll use the camera and how you define "better.")

Edited June 28, 2012 by AlanG

[pico]

A rough example of modifying the bottom two panels.

Keep in mind that these are actual pixels.

The sample below is a small part of the full image.

-- EDIT -- I am not sure the two samples were taken in the same light. Jono?

Edited June 28, 2012 by pico

[bocaburger]

Sorry for what will probably seem like infantile questions to you digital-technical experts, but I'm a bit confused about how the MM works. I understand that on color sensors, each pixel splits the recorded data down three channels representing different wavelengths (colors). Does the MM sensor do likewise, splitting wavelengths down three channels except that they are interpreted by the firmware as tones of gray rather than colors? Or does each pixel record all the wavelengths seamlessly down the same channel? If the latter, why is that better than splitting it into the three channels? Does the MM really output three times the data as an M9 (would an uncompressed DNG from an MM be 3x larger than one from an M9?), or is it just recording it in a different way?

[AlanG]

Sorry for what will probably seem like infantile questions to you digital-technical experts, but I'm a bit confused about how the MM works. I understand that on color sensors, each pixel splits the recorded data down three channels representing different wavelengths (colors). Does the MM sensor do likewise, splitting wavelengths down three channels except that they are interpreted by the firmware as tones of gray rather than colors? Or does each pixel record all the wavelengths seamlessly down the same channel? If the latter, why is that better than splitting it into the three channels? Does the MM really output three times the data as an M9 (would an uncompressed DNG from an MM be 3x larger than one from an M9?), or is it just recording it in a different way?

 

The simplest answer I can think of to your question is that the MM uses all of its 18MP to record shades of grey and detail. It cannot distinguish between colors. It records to a "color" file simply by making three identical b/w images - one for each channel - red, green, and blue. So there is no color data captured or stored. The M9 divides its pixels into recording three b/w images that represent the colors needed to simulate a full color scene. It does that by having a red, green, or blue filter over each pixel receptor on the sensor. The color scene can then be re-created via the colored pixels on a monitor or via dyes or inks when making prints.

 

A more detailed explanation and my opinions...

 

The M9 has 18MP but uses 9 million of them to record green, 4.5 million of them to record red, and 4.5 million of them to record blue. It then uses all of this data to simulate what it would be like to have 18 million pixels with each one of them recording a separate red, green, and blue measurement from each specific spot. But since the red, green, and blue pixels are not in exactly the same locations, there is some "interpolation" or intelligent guesswork of what those values should be. However since most things in a scene require a mixture of these three colors, there still is 18MP of detail at play. But there is the possibility that this detail may not be as pure and exact as that from the MM's 18MP of detail. Some of this loss of detail is simply from the fact that the pixels are offset and some is due to the way the software tries to interpolate them into a larger, more detailed file. (Where its guesswork fails.)

 

The advantage in the M9 approach for b/w is that after the shot is taken, one can use the color values to adjust the tones of the image... e.g making an object lighter or darker based on its color.

 

The advantage to the MM approach is greater light sensitivity due to lack of the color filters and the potential for finer detail based on the lack of interpolation or "guess work."

 

It is clear to me that there is an advantage to converting from a color file to make a b/w image as there is a lot of control and it can be done after the fact. The MM would have to justify its less than optimal color to b/w control by having more detail, better high ISO capability or something else substantial (filters over the lens will not come close to what can be done in b/w conversion from a color file.)

 

When it comes to resolution and high ISO noise, it should be pretty easy to compare the two systems and see if the difference is great enough to justify working with a dedicated b/w camera and the limitations in post processing. IMHO for many typical photographs and applications the difference between these two approaches may not be that great. (E.g. some people used soft focus filters over the "too sharp" Hasselblad lenses when shooting portraits.) It may require pretty large prints and careful examination to see differences in detail - assuming you are shooting in a manner that will get the most out of your lens and camera in the first place. Higher ISO shooting is likely to be the exception... which will only exist if you don't use color filters over the lens when shooting with the MM.

Edited June 28, 2012 by AlanG

[Advertisement]

Hallo,

Look here: Leica M

[jaapv]

Good summary, Alan. The only thing you left out is that the CA effects of the filter in front of the sensor of the M9 (or any other bayer filtered camera for that matter) are not negligible. Light spill on adjacent pixels and refraction effects on the borders between the minifilters do make quite a difference, experts tell me.