M Color

[wlaidlaw]

Advertisement (gone after registration)

Hi Wilson, Alan's pics above bring up the ACR plug-in of my old CS3 for some reason. Are your DNG files still available?

 

LCT if you scroll up this thread a bit, you should find the link to the Dropbox page. They are still there.

 

Wilson

[Advertisement]

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

[mwilliamsphotography]

Thank you for posting these charts Alan, and for making them Wilson!. They make it much easier to eliminate variables of different skin, level of light and temp of light.

 

While doing a grab of all the charts for a comparison may introduce some differences from the original files (including differences in monitor calibration), it still is all relative enough to decipher the differences in color rendering by using the eye-dropper to study the color mix percentages found in the Info dialog box from PS.

 

Of particular interest was the read out from the Red X-Rite Square:

 

M240 RED .... C-1%, M-90%, Y-55%, K-1%

 

M9 RED ....... C-1%, M-76%, Y-68%, K-1%

 

A99 RED ...... C-1%, M-74%, Y-67%, K-1%

 

Then when you read the Dark Grey Square (the closest match to a standard grey card), the read out is almost the same for all three cameras.

 

The Green Square (foliage), Dark Blue & Light Blue (sky) squares are also all very close to equal for all three cameras.

 

This directionally seems to indicate that if you did a manual white balance off a grey card, most all the colors would be relatively close for all three cameras ... except the M reds which are definitely skewed to the Magenta and have less percentage of Yellow ... the two colors that primarily make up Caucasian skin tone, or play a strong underlying role in any other skin tone type.

 

That the remaining colors seem to be in line with one another indicates that a global adjustment isn't the answer because it will then effect the other colors that have magenta and yellow in their make-up. This may account for some M240 Images that have the skin tones corrected while the other colors start shifting.

 

Since most post software can isolate a specific color/hue/density one would think that could be done with a firmware version, a profile adjustment, or both. Leica managed to adjust the magenta skew of the S-2 without losing the other color fidelity ... the question is whether the S2 skew was global, which I cannot remember?

 

As it stands, one thing seems clear to me ... in post, trying to manually adjust for this skew in different lighting scenarios, especially in less than idea light, would be labor intense.

 

- Marc

[Guest malland]

Marc, thanks. Now, this makes we wonder why the M240 has this color issue, considering that Leica, I assume, could have used it's S2 experience, as well as using the M9 color rendition as a goal, to rectify this. That the color issue is there makes it seems that there was either faulty implementation or that there is a deeper problem that they were not yet able to rectify. Your thoughts?

 

—Mitch/Paris

My flickr site

[wlaidlaw]

Thank you for posting these charts Alan, and for making them Wilson!. They make it much easier to eliminate variables of different skin, level of light and temp of light.

 

While doing a grab of all the charts for a comparison may introduce some differences from the original files (including differences in monitor calibration), it still is all relative enough to decipher the differences in color rendering by using the eye-dropper to study the color mix percentages found in the Info dialog box from PS.

 

Of particular interest was the read out from the Red X-Rite Square:

 

M240 RED .... C-1%, M-90%, Y-55%, K-1%

 

M9 RED ....... C-1%, M-76%, Y-68%, K-1%

 

A99 RED ...... C-1%, M-74%, Y-67%, K-1%

 

Then when you read the Dark Grey Square (the closest match to a standard grey card), the read out is almost the same for all three cameras.

 

The Green Square (foliage), Dark Blue & Light Blue (sky) squares are also all very close to equal for all three cameras.

 

This directionally seems to indicate that if you did a manual white balance off a grey card, most all the colors would be relatively close for all three cameras ... except the M reds which are definitely skewed to the Magenta and have less percentage of Yellow ... the two colors that primarily make up Caucasian skin tone, or play a strong underlying role in any other skin tone type.

 

That the remaining colors seem to be in line with one another indicates that a global adjustment isn't the answer because it will then effect the other colors that have magenta and yellow in their make-up. This may account for some M240 Images that have the skin tones corrected while the other colors start shifting.

 

Since most post software can isolate a specific color/hue/density one would think that could be done with a firmware version, a profile adjustment, or both. Leica managed to adjust the magenta skew of the S-2 without losing the other color fidelity ... the question is whether the S2 skew was global, which I cannot remember?

 

As it stands, one thing seems clear to me ... in post, trying to manually adjust for this skew in different lighting scenarios, especially in less than idea light, would be labor intense.

 

- Marc

 

Marc,

 

Very much in line with my thinking. My M240 is all stripped down today, waiting for a DHL pick up to go on its holidays to Solms, so I cannot check but I am wondering if the IR filtration might be on the thin side. On its return, I will retake those CC24 pictures with a manual WB using a grey card, then with and without IR filters.

 

For the moment, I am happy enough with the dual illuminance profile corrected output for general use on LR5 and C1's skin tone corrected for portraiture. I am guessing that all the kerfuffle over lugs and whatever the production hold up cause was, has held up the firmware update. My spies told me that this was due in early to mid April but here we are 2/3rds of the way through May and still no update.

 

The lens correction for wider non-retrofocal/telecentric lenses desperately needs improvement, as everything from the 28/2 ASPH Summicron downwards is suffering red edges. My CV15 is unusable on the M240 without Cornerfix, in comparison to on the M9, where with the 21/2.8 correction selected it is just fine.

 

Wilson

[glenerrolrd]

Thank you for posting these charts Alan, and for making them Wilson!. They make it much easier to eliminate variables of different skin, level of light and temp of light.

 

While doing a grab of all the charts for a comparison may introduce some differences from the original files (including differences in monitor calibration), it still is all relative enough to decipher the differences in color rendering by using the eye-dropper to study the color mix percentages found in the Info dialog box from PS.

 

Of particular interest was the read out from the Red X-Rite Square:

 

M240 RED .... C-1%, M-90%, Y-55%, K-1%

 

M9 RED ....... C-1%, M-76%, Y-68%, K-1%

 

A99 RED ...... C-1%, M-74%, Y-67%, K-1%

 

Then when you read the Dark Grey Square (the closest match to a standard grey card), the read out is almost the same for all three cameras.

 

The Green Square (foliage), Dark Blue & Light Blue (sky) squares are also all very close to equal for all three cameras.

 

This directionally seems to indicate that if you did a manual white balance off a grey card, most all the colors would be relatively close for all three cameras ... except the M reds which are definitely skewed to the Magenta and have less percentage of Yellow ... the two colors that primarily make up Caucasian skin tone, or play a strong underlying role in any other skin tone type.

 

That the remaining colors seem to be in line with one another indicates that a global adjustment isn't the answer because it will then effect the other colors that have magenta and yellow in their make-up. This may account for some M240 Images that have the skin tones corrected while the other colors start shifting.

 

Since most post software can isolate a specific color/hue/density one would think that could be done with a firmware version, a profile adjustment, or both. Leica managed to adjust the magenta skew of the S-2 without losing the other color fidelity ... the question is whether the S2 skew was global, which I cannot remember?

 

As it stands, one thing seems clear to me ... in post, trying to manually adjust for this skew in different lighting scenarios, especially in less than idea light, would be labor intense.

 

- Marc

 

Wow this is a real surprise . I measured the same charts but after going thru LR4 with various color profiles and I would never guess that the RED was higher in magenta in the original DNG of the M . Thanks for doing the analysis .

[wlaidlaw]

Marc, thanks. Now, this makes we wonder why the M240 has this color issue, considering that Leica, I assume, could have used it's S2 experience, as well as using the M9 color rendition as a goal, to rectify this. That the color issue is there makes it seems that there was either faulty implementation or that there is a deeper problem that they were not yet able to rectify. Your thoughts?

 

—Mitch/Paris

My flickr site

 

Mitch,

 

I am still far from convinced there is a real colour issue. I think there is a WB issue and as all we can set is the colour temperature and not the tint, that is not a total cure. Nearly all professional level cameras have less than ideal WB to begin with and this gets refined on later FW updates. I am totally used to this, being an early adopter of both the M8 and M9 and don't mind terribly, as long as the improvements don't take TOO long.

 

What I think is poor is the standard profiles from both LR and C1 (don't know about DxO). After the WB issue is sorted and these profiles get corrected or you use your own profiles on LR (created with either of the free programs Adobe DNG Profile Maker or X-Rite Color-Checker Passport), I think the colours will just sort themselves out. I just wish there was a free program to make ICC profiles for C1, other than Argyll, for which you need a masters in computer science.

 

Wilson

[wlaidlaw]

Advertisement (gone after registration)

Has anyone here successfully created an ICC profile from a CC24 Gretag Macbeth chart DNG image using the Argyll command line software. I have been looking through the documentation and I cannot see any way to direct convert the DNG output into .ti3 CGATS format to apply Argyll to it. Yes you convert a JPEG to a ti3 file but that defeats the object of making a conversion profile for the M240 DNG's, as you have either first applied the in-camera conversion profile or a profile within a conversion program.

 

Alternatively does anyone know of a UK agency who will make you a few ICC profiles from CC24 images. The forum member Ronald used to do it in Paris but he doesn't seem to be around these days.

 

Wilson

[AlanG]

Alan,

 

It is always nice to acknowledge the person who took the trouble to post DNG's of CC24 charts on their Dropbox for the benefit of other Leica users - Me! (checked the EXIF camera serial number).

 

Wilson

 

Sorry and thanks for the charts but I was looking through this thread trying to find DNGs. I downloaded some that I found and then when I later opened them I had no idea of who posted them. I was tired as it was late when I posted, and it was quite a few pages back so I lost track of the link... despite looking back a few times, I couldn't find it again.

 

A key advantage of having this color chart is that one can separate what is going on in the color profile being used to convert the dng without regard to anything else the camera does regarding color. (e.g. white balance.) That tungsten chart you posted still is very warm - I don't know if you set the Kelvin value or used auto. So the next step would be to look at the tungsten chart you posted, click balance on the same square and then see how much the other colors deviate from those in the sunlit chart - processed identically. However these days tungsten lighting is getting pretty rare... a chart shot under various energy efficient bulbs would be more revealing of how well colors can be reproduced when the illumination may lack a full spectrum to work with.

 

Also keep in mind that this chart represents a very small range of colors in the real world but is good for having a standard. Color can be produced in other ways than just by reflected light off of pigment.

 

Welcome to the world of color sensitometry! Consider that color negative film uses a very heavy orange base color for a reason.

[AlanG]

Since most post software can isolate a specific color/hue/density one would think that could be done with a firmware version, a profile adjustment, or both. Leica managed to adjust the magenta skew of the S-2 without losing the other color fidelity ... the question is whether the S2 skew was global, which I cannot remember?

 

- Marc

 

It actually is very simple when you consider that is what DXO is doing. If you are willing to use DXO, you just have to pick the camera profile you like for the M files to simulate. Surely Leica could adjust the in-camera conversion any way they want also and other raw conversion companies could give you more choices in profiles such as neutral, portrait, high contrast, etc that better match what most other cameras are set for.

 

I should be able to shoot the same chart with a different camera and DXO should make the Leica M's simulated version look like the one that was actually shot with an M.

[AlanG]

Wow this is a real surprise . I measured the same charts but after going thru LR4 with various color profiles and I would never guess that the RED was higher in magenta in the original DNG of the M . Thanks for doing the analysis .

 

I don't think I know what is going on in the original dng. Consider that these are just a bunch of numbers not colors. All we know is what happens after a profile is applied and we can see that DXO and other profiles can make it look any way they want... depending on which is used.

 

I also used C1 and went through various profiles (there isn't one for the M - although generic dng - linear response may be the way to start) and they produce pretty different results.

[wlaidlaw]

Alan,

 

As you say, we don't have a tungsten bulb in the house so I used halogen as the nearest. I therefore set the colour temp in the camera at 3000ºK. Now it may be that those are very yellow halogens (50W GU10's) but I agree it seems very yellow to me as well. For the purpose of making a DNG dual illuminant profile using Color Checker Passport, that really does not matter, as I believe the program corrects any WB imbalance before it does the colour profiles.

 

Wilson

 

PS the latest version of C1 which is 7.1.1 definitely has an M240 profile.

 

PPS C1 7.1.2 has come out today, hopefully with an improved M240 profile, actually made with a camera rather than the Leica SDK this time!

[AlanG]

Alan,

 

As you say, we don't have a tungsten bulb in the house so I used halogen as the nearest. I therefore set the colour temp in the camera at 3000ºK. Now it may be that those are very yellow halogens (50W GU10's) but I agree it seems very yellow to me as well. For the purpose of making a DNG dual illuminant profile using Color Checker Passport, that really does not matter, as I believe the program corrects any WB imbalance before it does the colour profiles.

 

Wilson

 

PS the latest version of C1 which is 7.1.1 definitely has an M240 profile.

 

PPS C1 7.1.2 has come out today, hopefully with an improved M240 profile, actually made with a camera rather than the Leica SDK this time!

 

I'll update my C1 and check it out although spending a day doing sensitometry is like being back in school.

 

What I was getting at regarding bulbs is not halogen which are similar to tungsten and have a continuous spectrum. The bulbs I run across are the various compact fluorescent that now take the place of standard incandescent bulbs. Various LEDs are coming into homes also. Commercial buildings have had a wide range of lighting for many years that are various types of discharge bulbs. many of these do not comprise a full range of color and thus photos under that lighting may not have the color values we expect or desire.

 

At least with digital photography this can be compensated for somewhat or altered on a local level if necessary. But one cannot expect to get the same color reproduction on all subjects with all lighting.

[lct]

Interesting discussion indeed.

Here's the chart i've been using to tweak my first pics out of the M240: http://tinyurl.com/puw5y7s (48MB file).

A little comparo between LR and C1 shows that a bit of PP work might be advised with both, especially LR IMHO. Based on Wilson's "Sunlit Planar" dng file with many thanks to him.

http://tinyurl.com/q3talta

(5MB file)

[wlaidlaw]

It may be wishful thinking, but I thing Phase One have improved the M240 profile. It seems to have calmer reds to me and less magenta in faces. I think greens of grass and foliage are more natural as well.

 

I attach a photo of zero merit but it has a red, blue and green on my grandson's jacket and hat, that we can match to the original objects. They are pretty much spot on with my Spyder 4 Pro calibrated screen (calibrated earlier this week) on Photoshop although Safari manages the colours slightly differently, making them more muted.

 

Wilson

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

[AlanG]

I don't know if I could judge much even subjectively from that photo as there are too many possible variables.

 

I briefly looked at the charts in C1 and the M profile there really boosts the colors. I'd think the Generic Dng would be better but I haven't measured any of these. I also looked at the two other charts a bit. The problem is the exposure on the tungsten one does not match the sunlight one. I adjusted (quickly by eye only) a bit to standardize the neutral points and the gray scale of all the charts and using the M 240 profile in C1 the sunlit one is much more saturated than the others.

 

I think the C1 profile for the M looks pretty terrible to me and will have to go back and see if I made any mistakes although I can't see how since I've been working with C1 for 10 years now. I'll have to check that the output color profiles are the same and that nothing is getting changed when I bring them into another program and paste them. But the image looks exactly like this in C1 and it was obvious that it happened when I clicked on the M240 profile. I am trying to understand how a company like C1 ended up with this as its profile. I am wondering if they are relying on a profile supplied by Leica instead of making their own. I don't believe that C1 uses the WCS color codecs that are part of Windows.

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

[AlanG]

...While doing a grab of all the charts for a comparison may introduce some differences from the original files (including differences in monitor calibration), it still is all relative enough to decipher the differences in color rendering by using the eye-dropper to study the color mix percentages found in the Info dialog box from PS..

 

- Marc

 

You should be able to save each file and not have to do a screen grab - which is really problematic especially if you have a high gamut monitor.

 

I was tempted to work this all out to study the different charts under different light as depicted by a few converters using M profiles as well as others including camera simulation. But I really don't have the time to do a good job on it. And it became very obvious on casual observation that the M profile in C1 and DXO deviate quite a bit from most other cameras I've seen. So that likely explains a lot of the variation people are getting.

 

So I think most of you understand how to compare a chart using different profiles only.

 

Comparing charts shot under different lighting with each other add some challenges but you can easily get a rough idea of how the auto or manual white balance is working. And you could compare this with the same tests using other cameras. After adjusting them for matching gray tones, you should be able to see if there are any obvious color changes on specific colors. You really have to get the exposure close on both and try not to include any background. Maybe attach the color chart to a large gray card. The ideal light is diffuse regardless of source and hopefully there are no colors in the scene that affect the illumination too much. (e.g green grass or blue sky) A shot made illuminated with direct specular sunlight may not match the same chart shot in diffuse light of identical color characteristics.

[lct]

It may be wishful thinking, but I thing Phase One have improved the M240 profile. It seems to have calmer reds to me and less magenta in faces...

Still too much saturation for me. Red, magenta, cyan mainly (above). I think i'll stick to my profile for now (below) although blue is too dark probably.

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

[mwilliamsphotography]

I think this will all get sorted out in due time ... it is simply too important for Leica. Lots of very smart people working on this here on the LUF also. Hopefully, a next version EVF will also become available in due time ... and they'll make the control wheel black ... LOL!

 

So when my dealer called with my M240 last night ... I passed for now.

 

In the meantime, I'll stick with the S2P for color, and M Monochrome ... which eclipsed my M9P to the degree that I sold it for lack of use. It may take a while before my preference for the MM and B&W rangefinder work diminishes.

 

Plus, now I'm intrigued by the M Mini and what that might be. A smaller CL sized camera that takes M lenses may fit my color application needs just dandy.

 

Lots of exciting things from Leica these days!

 

-Marc

[Advertisement]

Hallo,

Look here: Leica M

[D&A]

Wow this is a real surprise . I measured the same charts but after going thru LR4 with various color profiles and I would never guess that the RED was higher in magenta in the original DNG of the M . Thanks for doing the analysis .

 

+1! Great analysis and it apparent to me in the M240 files I've worked with that magenta rears it's ugly head far too often, espcially with skin tone. As Marc has pointed out, I do believe Leica will eventually get this sorted out, especially with a lot of smart people workking intensely on identifying the problem. My question though is whether this can simply be fixed by a firmware update or if something in addition to this is required.

 

Dave (D&A)

[AlanG]

+1! Great analysis and it apparent to me in the M240 files I've worked with that magenta rears it's ugly head far too often, espcially with skin tone. As Marc has pointed out, I do believe Leica will eventually get this sorted out, especially with a lot of smart people workking intensely on identifying the problem. My question though is whether this can simply be fixed by a firmware update or if something in addition to this is required.

 

Dave (D&A)

 

It seems to me that it would not take too many smart people to realize they just have to use a different profile. They could just pick one they like from the DXO camera simulation and use it. As can you unless you need better in-camera jpegs. That is something they'll have to change in firmware.

 

Can't you use a different profile in Lightroom or adjust the one you are using?