Jump to content

green cast in M9 with new firmware 1204


pibitller

Recommended Posts

Advertisement (gone after registration)

Sorry for the delay with posting the pictures. I am having difficulties with the new forum software: I can log in on my iPhone & iPad but have problems uploading some images even though they comply with forum regulations. I can currently not login from stationary computers (awaiting solution from @LUF Admin).

I will post the rest of the images from my comparison (Elmarit 28, C-Biogon 35, Summicron 50, Apo Summicron 90) once the technicalities have been sorted out.

Kind regards

Mathias

 

EDIT: Login from stationary computer still does not work, but serial posting from tablet now works. Somehow, the new forum software seems to cache uploaded files so that submitting several posts in a single thread is a bit fiddly, at least to me. Anyway, I will add the missing pictures now.

 

 

Edited by schattenundlicht
Link to post
Share on other sites

Leica Elmarit 28 mm ASPH 

without lens detection

(All images shot in broad daylight, auto white balance, ooc jpeg, M9, fw 1.216, 1/4000 or 1/3000, f/5.6 or f/8)

Green hue most prominently visible in upper right corner, even in low quality forum jpeg.

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!

Edited by schattenundlicht
Link to post
Share on other sites

Zeiss C-Biogon 35 mm f/2.8

with manual lens profile (declared as „MATE@35mm“)

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!

Link to post
Share on other sites

 

Zeiss C-Biogon 35 mm f/2.8

without lens profile 

Green hue is less starkly visible than with the Elmarit 28, in forum quality, but it is still very obvious in the full res image.

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!

Link to post
Share on other sites

Leica Summicron 50 mm f/2 (Cron V)

with lens detection

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!

Link to post
Share on other sites

Leica Summicron 50 mm f/2 (Cron V)

without lens detection

In forum resolution on an uncalibrated monitor/ tablet, the hue will hardly be noticeable, but on full res image on a calibrated workstation, it is still visible.

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!

 

 

Link to post
Share on other sites

Advertisement (gone after registration)

Leica Apo-Summicron 90 mm f/2 ASPH

with lens detection

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!

Link to post
Share on other sites

 

Leica Apo-Summicron 90 mm f/2 ASPH

without lens detection

With naked eye, the hue is virtually undetecteable in forum quality on an uncalibrated screen. However, one can discern the difference on fullres images when reading RGB values from corresponding ROIs. Thus the effect is reproduceable across all tested focal lengths, but the magnitude depends on FL.

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!

Edited by schattenundlicht
Link to post
Share on other sites

Hi,

Good to see there are other users reporting similar issues. When I've detected this issue, I was in NY, I passed by Soho Leica store just to ask if they have had similar reports, or if this was a know problem with the new sensor or my Elmarit. They didn't had there any Elmarit but we tried with a Elmar 24mm, and the issue was also there as I've reported before here, this not a problem affecting the Elmarit 28mm Asph. I've did tests with DNGs and Jpeg OOC and both showed the same hue more pronounce at the bottom corners of the frame. I was not able to see it with both my Summicrons, 35mm and 50mm. 

I though this was a problem with my camera but now starts to seem obvious this is an issue with the 'old' profiles requiring an overhaul to accommodate the new sensor design. I hope someone from Leica can help us with a new firmware update for the new sensor M9's. 

As reported before, I have DNGs from my Elmarit, before the sensor replacement and they were perfect! Actually, I love this lens, reason why I would be very happy to have this solved. Right now I can solve it on LR applying a local brush but it's far from perfect.

Let's check if other users report, the more the better.

Link to post
Share on other sites

CONCLUSION:

  • The Elmarit 28 ASPH is not at fault at all. No need to sell this wonderful lens. :)
  • The problem stems from the M9 sensor but is corrected in-camera for jpegs if the appropriate lens profile is chosen manually or detected automatically.
  • DNGs shot w/o lens profile show the full cyan/green hue, whereas DNGs shot with the appropriate lens profile show somewhat corrected color on default in Lightroom, even when lens profiles are deactivated in LR. 
  • This is a case in point, that „raw“ files are far from being truly unprocessed sensor data ;)
  • The correction of the DNGs is only partal, however. There remains a subtle greenish hue in the corners, although much much less garish than w/o lens profile. This is a nuisance, since non-uniform hues are harder to correct in postprocessing.
  • I find it disappointing to use different M bodies (e.g. M9 & M10) at the same time on the same location and get such highly inconsistent color rendering.
  • Any further firmware improvement by Leica would be very welcome.

 

Edited by schattenundlicht
Link to post
Share on other sites

3 hours ago, schattenundlicht said:
  • The Elmarit 28 ASPH is not at fault at all. No need to sell this wonderful lens. :) 
  • The problem stems from the M9 sensor but is corrected in-camera for jpegs if the appropriate lens profile is chosen manually or detected automatically.

I agree that one should not blame the 28 Elmarit ASPH alone or in particular. But in reality it is a combination of a lens, the short mount distance of the M cameras, and sensor microlens design and IR filtering. Leica could have designed the Elmarit optics to be more like a massive SLR lens, or the Sony Alpha FE 28mm f/2.0 - twice as long, and 10% fatter - and probably eliminated almost all the "native" cyan vignetting. But that is not consistent with the scale of a Leica M camera, and in the case of the Elmarit in particular, a lens that was designed to 1) have the "size and feel," and approximate FoV, of a 35mm on 2) the cropped sensor of the M8 (the lens and camera were introduced simultaneously).

15 years ago, Leica was insisting that it was not possible to make a digital M at all, and technical problems like the cyan vignetting are exactly what they knew they would face.

Market forces eventually forced them to address the problem, by a Rube-Goldberg combination of 6-bit coding, IR filtering so "thin" that it was only 50% effective, a cropped sensor in the first iteration (M8), extensive in-camera processing to correct the remaining vignetting, and so on.

https://en.wikipedia.org/wiki/Rube_Goldberg_machine#/media/File:Rube_Goldberg's_"Self-Operating_Napkin"_(cropped).gif

It helps to remember that the M9 sensor silicon is "last-millenium" technology - based on the Kodak CCD architecture for, first, medium-format SLR digital-back sensors (c. 2000), and then cut down for use in the Leica DMR back (2005), and then carried over into the M8 (late 2006), and then "de-cropped" for 24x36 in the M9 (2009). The "new" M9 sensor changed virtually  nothing - except for using a less corrosion-prone cover glass type.

3 hours ago, schattenundlicht said:

• This is a case in point, that „raw“ files are far from being truly unprocessed sensor data ;)

As has been well-known (at least by anyone paying attention) since - oh - about 2003. ;)

3 hours ago, schattenundlicht said:

• The correction of the DNGs is only part(i)al, however.

Yes, and for a very simple reason. To remove the cyan tint, the red channel is pushed or amplified. On a curved gradient from zero amount in the center to a larger amount (up to 2 stops with an older 21) in the corners. Which means more noise in the red channel. One of the very first things I noticed trying my 21 pre-ASPH on an M9 on the day of introduction (9/9/09) was how many red noise speckles appeared near the edges as the red luminance was cranked up to counteract the cyan tint. Leica does a "balancing act," varying the amount or "curve" of correction with the ISO in use.

3 hours ago, schattenundlicht said:

I find it disappointing to use different M bodies (e.g. M9 & M10) at the same time on the same location and get such highly inconsistent color rendering.

Just a question of perspective. Kodachrome II did not render color the same as original Kodachrome, nor did Kodachrome 25 render color the same as KII. Ektachrome 100 in 2005 (and probably in 2018) rendered color differently than E100 (1980). Velvia 50 renders very differently than Fujichrome 50 RF (1980s) - and "Velvia" film simulation in a Fuji digital is not really the same as Velvia itself. Time and technology move on.

3 hours ago, schattenundlicht said:

Any further firmware improvement by Leica would be very welcome. 

For those using the M9, I hope so. However, 9 years and 2 generations on, at some point the M9 becomes "a finished product." Looks like the last M8 firmware update was five years ago - and the M9 is only 3 years newer.

Edited by adan
  • Like 1
  • Thanks 1
Link to post
Share on other sites

1 hour ago, adan said:

[...]

Thanks for your knowledgeable comments, which are appreciated!

1 hour ago, adan said:

But that is not consistent with the scale of a Leica M camera, and in the case of the Elmarit in particular, a lens that was designed to 1) have the "size and feel," and approximate FoV, of a 35mm on 2) the cropped sensor of the M8 (the lens and camera were introduced simultaneously).

Yet, on the M10, the Elmarit works rather flawlessly. Since I own both bodies, I can freely use lenses as needed, however I like to reserve the M10 for the more extreme focal lengths when I am out shooting, because of the EVF option. Thus I would have liked to use the "simpler" lenses which do not outperform the M10's sensor on my M9. Well, you can't have everything ;)

1 hour ago, adan said:

Market forces eventually forced them to address the problem, by a Rube-Goldberg combination of 6-bit coding, IR filtering so "thin" that it was only 50% effective, a cropped sensor in the first iteration (M8), extensive in-camera processing to correct the remaining vignetting, and so on.

https://en.wikipedia.org/wiki/Rube_Goldberg_machine#/media/File:Rube_Goldberg's_"Self-Operating_Napkin"_(cropped).gif

That one gave me a good guffaw!

1 hour ago, adan said:

The "new" M9 sensor changed virtually  nothing - except for using a less corrosion-prone cover glass type.

Going back to my legacy files from the M9 before the sensor replacement, I find the effect less pronounced. Thus I would like to venture that the change of cover glass did change something.

1 hour ago, adan said:

As has been well-known (at least by anyone paying attention) since - oh - about 2003. ;)

Well, due to other obligations, my dilettante photography was in a latency period at that time, so I might have missed that one ;)

1 hour ago, adan said:

Yes, and for a very simple reason. To remove the cyan tint, the red channel is pushed or amplified. On a curved gradient from zero amount in the center to a larger amount (up to 2 stops with an older 21) in the corners. Which means more noise in the red channel. One of the very first things I noticed trying my 21 pre-ASPH on an M9 on the day of introduction (9/9/09) was how many red noise speckles appeared near the edges as the red luminance was cranked up to counteract the cyan tint. Leica does a "balancing act," varying the amount or "curve" of correction with the ISO in use.

That's a very helpful technical explanation. Thanks.

1 hour ago, adan said:

Just a question of perspective. Kodachrome II did not render color the same as original Kodachrome, nor did Kodachrome 25 render color the same as KII. Ektachrome 100 in 2005 (and probably in 2018) rendered color differently than E100 (1980). Velvia 50 renders very differently than Fujichrome 50 RF (1980s) - and "Velvia" film simulation in a Fuji digital is not really the same as Velvia itself.

 But then, in the 1980ies I would not carry multiple bodies for the same shooting with a mix of Kodachrome and Ektachrome...

1 hour ago, adan said:

However, 9 years and 2 generations on, at some point the M9 becomes "a finished product." Looks like the last M8 firmware update was five years ago - and the M9 is only 3 years newer.

That's fair enough, I guess.

(This is true for all electronic devices these days: My Apple notebook hardware generally outlasts the company's willingness to support older hardware with the latest OS, or with security updates for the older OS, thus forcing me to eventually convert them to Linux machines which pile up over the decades ;)  Numerous other examples could be found).

Kind regards

Mathias

Edited by schattenundlicht
  • Thanks 1
Link to post
Share on other sites

  • 2 weeks later...
I've just completed an interesting test on my Leica lenses after reading about this  problem on LUF. The problem appears to be only with the M9 with the latest Firmware. Mine is 1.216. 
 
My lenses are an ancient 50mm f2.0 Summicron 1593333, a new 28mm f2.8 Elmarit ASPH 4662257 and an old 135mm f2.8 Elmarit 2656227. 
 
In each case I used Manual lens detection and with each iteration ie for the 28mm the ASPH setting 11606, and the non-ASPH 11809, 11804, for the 50mm I used 11817, and the 11819 etc and the 135mm  the 11829 
 
I set the camera on a tripod and took a series of photos from the lowest to the highest f number of a slightly off white door. Moving the tripod so I had approximately the same amount of the door in the frame for each lens and in similar lighting conditions. 
 
Each photo was then viewed in Microsoft Picture Manager. Brightness was adjusted to +40 and the whole viewed at 50% zoom. 
 
As predicted, the 28mm was the worst no matter which lens detection setting I used with a noticeable green cast in the lower left hand corner. The cast remained throughout the f stop range, though when the detection  was set to 11606 it was almost gone by f16. 
 
The ancient 50mm Summicron  had no cast with both of the detection settings throughout the f range. 
 
The 135mm Elmarit showed a very slight green tinge all across the bottom 1/3 of the picture which almost disappeared by f22. 
 
So I have a problem too.
 
 
 
Link to post
Share on other sites

Sad to see more people with the same issue, but at least it clearly shows it's not an isolated problem.

In the meantime, I sent a mail to Leica Customer care and I got no feedback for the past 12 days. Not sure if I will to be honest.

Any suggestions how to make us heard?

I really hope they are silently analyzing this strange behavior of the new sensor ;)

Link to post
Share on other sites

I've contacted Leica UK and got these replies, both within an hour of sending my email. Well done Leica UK.

 

''The problem you’re seeing is sometimes referred to as ‘The Italian Flag’. It’s the reason Leica initially told everyone a digital M was impossible.

It occurs because the lenses being used are not ‘telecentric’. Many M lenses were designed in the days of film, when telecentric design was not important. 

But even when lenses are designed today, it’s difficult to make M lenses telecentric due to the small diameter of the M bayonet. 

I sometimes wondered if there was any difference between the ‘new’ and ‘old’ M9 sensors, but (apart from that one guy and yourself) no one has ever commented on this. 

I would say the probable cause is either due to a change in the sensor, or (more likely) down to the new IR cover glass being used. 

It’s not really a firmware fault as such. The revised firmware will try and correct any colour shifts as much as possible. But there may be limits.''

Then later:

'' Leica say the 28mm f2.8 Asph is the most ‘demanding’ of all M lenses in terms of delivering light rays to the sensor at an oblique angle. 

It’s partly due to its’ small size; quite a few optical tricks had to be pulled to make it small.
 
This lens was developed during the film era, years before Leica offered a full-frame M. 
The later sensors on the M (240) and M10 handle non-telecentric lenses better than the M9. So the Italian Flag problem is not really an issue…''
 
I said I'd get an older Elmarit 28mm fo compare when money allowed and here's his reply:
 
''The older type IV 28mm f2.8 is comparable to the new lens in terms of sharpness and contrast, and better in terms of telecentricity, but it’s also much larger… 
Longer lenses (like the 135mm) are telecentric simply by virtue of the long distance between the rear element and the sensor. ''
 
I hope this helps people who're having trouble. When I'm up to it I'll get the RGB numbers from the different areas  in the different lenses from Photoshop. But my findings just show how good at this the ancient, 1958, Summicron , is compared to the 1973 135mm Elmarit and the brand new 28mm Elmarit. 
 

'

 

  • Thanks 1
Link to post
Share on other sites

  • 1 month later...

Hello all, been a lurking member,

I would also like to say the 28mm Elmarit Asph (6 bit coded) also exhibits green cast on my M9P with new sensor with latest firmware,  both camera and lens has been to Weztlar for tests, and came back with no solution.

Another lens I noticed was a 35mm Pre_Asph lux v2, when using this lens and manually selecting the 35mm 1.4 profile, the dng file when imported into lightroom shows the Exif as 35mm Summicron IV instead. It happens on the M9P only. I have, on another camera M240 the exif shows the correct lens as 35mm 1.4. Just wondering if this is a related problem of the new firmware regarding lens profiling. 

Anyone with the above care to test their lens and check the exif?

Currently my other lenses shows the correct Exif lens profile but I do not have other old lenses to try out.

Link to post
Share on other sites

17 hours ago, ngjeremy said:

Another lens I noticed was a 35mm Pre_Asph lux v2, when using this lens and manually selecting the 35mm 1.4 profile, the dng file when imported into lightroom shows the Exif as 35mm Summicron IV instead. It happens on the M9P only.

M9P as well here and can not confirm that:

exiftool -s -FirmWareVersion -LensId L1008334.DNG
FirmwareVersion                 : 1.204
LensID                          : Summilux-M 35mm f/1.4

1.204 is not the last firmware but is a post sensor change firmware anyway.

I don't use lightroom though, maybe the bug lies therein.

Link to post
Share on other sites

Thank you Jul, I have thought about that but its not Lightroom.

I have actually tested it again on new files and same result. What I did was took some photos by manual selecting 35mm 1.4, copied the file from SD card onto my hard drive and then and opening the files with Capture One. Under the information tab it still showed as 35mm summicron IV. This only happens to the M9P.

I have firmware 1.216 at the moment by the way. 

I'm not very sure If there was any issue with the 1.204 firmware as I cant recall which firmware I had in my prior photos (2017). but my 2017 lux files also shows as summicron 35 IV. I did not use this lens before 2017.

 

Link to post
Share on other sites

21 hours ago, ngjeremy said:

.....

I have actually tested it again on new files and same result. What I did was took some photos by manual selecting 35mm 1.4, copied the file from SD card onto my hard drive and then and opening the files with Capture One. Under the information tab it still showed as 35mm summicron IV. This only happens to the M9P.

 

You need to use the Lenstagger plugin with Lightroom to change the lens info from “35mm Summicron IV” to the pre-ASPH 35 Lux. Lightroom does not recognize the lens code Leica records for the pre-ASPH 35 Lux. 

Link to post
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

×
×
  • Create New...