Re: Possible serious Issue with AWB & M9?

[D&A]

Advertisement (gone after registration)

Hi All,

 

This post may be premature and it does require even the most cursory investigation to validate any issue, so at the moment, it was a brief observation. I am just helping an acquaintance setting up his M9 . I had my M8.2 with me. We took very similar shots of the same indoor scenes, lit by general household tungsten light bulbs (although a few were those new florescent type screw in bulbs in certain locations.

 

We either used the same lens and moved forward or back to frame the shot similarly or used different lens but the framing of each respective shot with each camera was very close. RAW was shot with both although we did try jpeg.

 

What I quickly noticed was that when we brought up the images on the back LCD screen of each camera...the M8.2, AWB appeared to be good (close) as it always was with the latest M8 firmware and a good representation of the color balance of the scene. With the M9, every shot more was excessively warm..an extreme yellowish-orange color cast over the entire image...which only could be altered by changing the Kevin adjustment dramatically. The tungsten setting had little effect. Obviously the next step was to look at the files themselves in a RAW converter, and in each, the color balance came close to that seen on the LCD? I'm a bit stumped why the difference but bet the answer is so obvious, that Its right in front of me, and I'm not seeing it. Anyone else notice this or have any suggestions. It's not that the LCD is supposed to be relied on to make assessment of accuracy of color of the recorded image (I've been involved with digital files way to long to understand that)...was just shocked to see each respective image looking totally opposite from one another. I'm not concerned once getting into the RAW converter, since profiles have to be developed for the M9 obviously. Its what is observed on the LCD's of both respective cameras. No daylight shots taken in comparison yet...obviously the simple thing to do next. Thanks!

Edited December 28, 2009 by D&A

[larrylgreenhill]

I noticed the same problem of the yellow-green cast to the Automatic White Balance after trying out my new M9 for the first time today, and after comparing it to my trusty M8.

 

All pictures taken under incandescent light are very yellow green when the white balance is set to AWB, flourescent, tungsten, or AWB. Furthermore, any areas that go off into shadow at 1250 ISO have extremely high grainy digital noise. The M8 has NONE of these problems. If I then put a flash unit on the M9, the yellow-green look disappears. Any natural light from outside, then the histogram works and the sensor's white balance is normal.

 

There is another problem associated with the yellow-green cast to all images taken under incancescent light. No matter how bright or dark it is when I shoot indoors under incandescent light, the histogram shows total underexposure -- off the graph to the left. I set the M9 to an Auto or tungsten white balance. No matter how it's set, the histogram shows total underexposure even when there is overexposure in the image as proven by the red clipping indicator in the info screen.

 

Do you also have the same problem with your M9?

 

Larry Greenhill

(Concerned M9 owner)

[Advertisement]

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

[D&A]

First, Thank you for your response.

 

Unfortunately both you and I (along with my aquantence) are memebrs of the same club. I'm wondering how many others are experiencing the same thing. I've observed each and every thing you mentioned with his M9. In fact, my next question a minute from now was going to be about ISO 1250 and how I noticed under the same tungsten lighting, I've never expected to see such noise in the shadows just as you described..(probably due to underexposure, exactly how you described...regardless of exposure compensation). My next question is whether everyone with M9's are experience this or not...maybe just those recently released? At least from what you described, outdoor lighting (daylight) is handled normally. As I mentioned, the tungsten setting didn't help in the indoor situation nor did raising the exposre compensation, in the ways you observed.

 

(Could all this only be on just released M9's???)

 

 

Lastly, which I didn't talk about cause its secondary at the moment...the SDHC cards I am temp using with his camera are Transcend cards which I know some can't get to work. Normally I always download my images to a folder on the desktop before ever opening any image.Mine cards so far work fine, but if I bring the images via a card reader reading the images off the card directly into AdobeBridge, then open any image ain photoshop after adjusting the RAW in bridge...then insert that SDHC card back into the camera....any image I opened from Bridge to Photoshop off the card, will show up with just a number on the cameras LCD sceen...but images just viewed in Bridge...can still be viewed on the M9. This probably isn't the camera per se, but the camera interfacing with this kind of SDHC.

 

I hope to hear from others on the main issues. Thanks again!

 

Dave (D&A)

Edited December 28, 2009 by D&A

[jaapv]

I would not call it an issue. The LCD of the M9 has indeed a tendency to produce a yellowish cast for artificial light in my experience. It does not bother me, as you say it is largely irrelevant for postprocessing and a camera LCD is no way to judge images anyway. I cannot reproduce the exposure thing, however.

[scott kirkpatrick]

Actually I am a little concerned about frequent strange results using AWB. Not judged on the LCD, but when the raw files are opened in Capture One.

 

With the M8, there was always a tendency for low light shots to be yellowish, since the AWB stops at a color temperature of 3000 when the actual light temperature might be in the low 2000's. Either that was to avoid other troubles or Leica just felt that the candlelit look was what people most want to see. So I am used to cooling indoor shots a bit.

 

But with the M9 (still pretty new, about 1400 exposures, no conclusions yet) I am seeing some very bad choices made under flourescent and mixed lighting as well. Does anyone know if the AWB decision is based on the actual image captured or on the response of the sensor to light reflecting off the closed shutter's silver stripe? If the latter, then the more spot-meter like exposure sensitivity of the M9 might be a factor. Of course shooting in raw, one can usually fix things, but it is nice to start with a close approximation.

 

scott

[lars_bergquist]

You mentioned the presence of fluorescent energy saving bulbs. These play havoc with AWB, creating a very strong yellow colour. Neither AWB nor the tungsten setting or any fluorescent setting can handle this. What happened was probably that a slight shift in framing brought a strong light source of that kind into the picture, and poof went the balance.

 

Handling this light in itself is not too difficult. I did a manual white balance on the light from Osram fluorescent bulbs, and incorporated this white balance into a dedicated user profile, in order to have it handy. But a mixture of light sources is indeed intractable. Sometimes the only remedy is to shoot or convert to black and white. We are going to see more of this problem as common matte tungsten bulbs are phased out, by E.U. ukase.

 

The old man from the Age of Kerosene Lamps

[scott kirkpatrick]

Advertisement (gone after registration)

I agree that the new spiral <18 watt EU flourescents are trouble, but I've shot several hundred exposures in a gymnasium with aged overhead old-fashioned long tube flourescents providing the only illumination, and find these are requiring strong corrections. My office lighting is similar but I don't have the camera with me. Will make some checks with a WhiBal certified gray card and think about it some more.

 

scott

[jaapv]

Actually I am a little concerned about frequent strange results using AWB. Not judged on the LCD, but when the raw files are opened in Capture One.

 

With the M8, there was always a tendency for low light shots to be yellowish, since the AWB stops at a color temperature of 3000 when the actual light temperature might be in the low 2000's. Either that was to avoid other troubles or Leica just felt that the candlelit look was what people most want to see. So I am used to cooling indoor shots a bit.

 

But with the M9 (still pretty new, about 1400 exposures, no conclusions yet) I am seeing some very bad choices made under flourescent and mixed lighting as well. Does anyone know if the AWB decision is based on the actual image captured or on the response of the sensor to light reflecting off the closed shutter's silver stripe? If the latter, then the more spot-meter like exposure sensitivity of the M9 might be a factor. Of course shooting in raw, one can usually fix things, but it is nice to start with a close approximation.

 

scott

Afaik the AWB is read off the sensor. I find that the real problems start where EU rules have been implemented with incandescent lightbulbs being replaced by energy-savers. The WB varies from one shot to the next in those cases. I'm not sure whether to blame Solms or Brussels.

 

I just walked downstairs in my practice and tried some shots with fluorescent lighting and a mixture of fluorescent, halogen and daylight. In both cases the M9 exhibited an excellent white balance, as good if not better than the M8. However, these are full-spectrum fluorescents. I think the problem is with the discontinuous spectrum light sources.

Edited December 28, 2009 by jaapv

[vikasmg]

I find that the real problems start where EU rules have been implemented with incandescent lightbulbs being replaced by energy-savers.

 

Interesting. I've wondered about this as well. Actually I first noticed the problem on a Panasonic GH-1 when diferent images shot at the same table in a restaurant exhibited different colour balance characteristics. I wonder if it related to the "flickering" nature of fluorescent and compact fluorescent lights certain exposure times. I wonder: does the frequency of the light source result in different colours at slightly diferent shutter speeds?

 

- Vikas

[luigi bertolotti]

The issue of new Energy Saver lamps surely exists and probably it is worth a deep tech review... I have the feel that the various kinds of ES lamp on the market don't yet provide a consistent spectrum... at home I am into a sort of "transition phase" (lamps replaced with ES type when needed) and my rough experience at home is like this:

 

1) Kitchen : new ES lamp with "long" lighting power ramp curve

2) Dining room : old incandescent

3) Sitting room : new ES lamp(s) with instant lighting power

 

the 3 rooms are adiacent... apparently, you don't have the feel of a different color dominant passing from one to another... a white sheet of paper with M8 in AWB mode gives 3 different hues (not difficult to adjust in LR).

[C_R]

A slight tendency towards yellow and green is very frequent in my AWB guided shots. I have no comparison with the M8, but the magnitude is about the same as the aberrance towards cool temp with my former Canon gear. In Capture One I pull the temp slider about 300 to 400 K towards blue, and often also one or two points towards magenta. Not a big deal, but incandescent light is another story.

 

A recent portrait series with incandescent (old bulbs !) light was way too yellow, and high ISO quality was really disappointing (uncompressed DNG, Capture One 5). I can highly recommend manual WB for this situation: a reshoot with manual WB from grey card was perfect, very good up to ISO 1250.

[microview]

According to an electronics engineer friend, the more of the fluorescent energy saving bulbs you have the worse the sound from your hi-fi will become!

[jaapv]

That is exactly why I have a huge mains filter in the power supply of my Hifi.

[gravastar]

I find the M9's auto color balance quite unusable where the low energy compact spiral fluorescents are concerned. Overall yellow green cast and a rendering of colors that can't be completely fixed in post processing.

 

The solution for me was to use custom white balance to remove the yellow green cast and the X-Rite Color Checker Passport to generate a fluorescent lighting profile to improve the color rendering. This seems to compensate very well for the somewhat uneven spectrum of the lamps. High quality domestic CFLs have a Color Rendering Index of about 80, poor ones can be as low as 60 compared to daylight's 100. So profiling is necessary to effectively improve the CRI.

 

Bob.

Edited December 28, 2009 by gravastar

[D&A]

Hi All,

 

Yes, for some of these initial test shots, there was the presence of those fluorescent energy saving bulbs as I indicated and they do play havoc with most notably the M9 AWB. Interestingly though and one of the reasons I initiated this "post", was that the same shot with the M8.2 (framed exactly the same and changing the framing of the shot didn't alter things to any degree concerning WB)...the M8.2 under the same lighting came quite close (relatively speaking since post processing adjustment was still somewhat required)..while the M9 was as others descibed, was maginitudes "off". The differences between the two cameras handling of this scene, framed exactly the same way, was night and day..with the M8.2 coming extremely close to accurate rendering and exposure. The M9 also seemed to underexpose the same shot to a degree, but the jury is out on that one. I'm just trying to get a handle on why such great WB (AWB) differences between the two cameras under tugsten or specifically this sort of lighting.

 

Its as though the AWB of the M9 is where the early M8's used to be, with their early firmware...before siginificant improvements implemented with later firmware (for the M8). Curious, there must be another explanation for the striking differences at least with the aforementioned lighting between the two cameras (M8 and M9). It's just a educated guess, but like the it was initially with the M8, I'm certain the M9's AWB will be addressed in future M9 firmware updates.

 

In regards to excessive shadow noise at ISO 1250 with M9 images under such lighting, not only does grossly incorrect WB lead to noise at higher ISO's, especially in underexposed areas, but so does of course underexposed areas in general. Combine both undesirable attributes in a particular shot, and I suspect this played a prominant role. Just getting the WB correct in postprocessing, would have probably eliminated some of the excessive noise I and some others descibed in this particular lighting situation. I haven't gotten to that point with his test files.

 

Appreciate all the comments so far. Thanks!

 

Dave (D&A)

 

 

You mentioned the presence of fluorescent energy saving bulbs. These play havoc with AWB

Edited December 28, 2009 by D&A

[D&A]

I was too late to edit my previous post (above), but also wanetd to mention that I tried a variety of differently framed shots under the lighting conditons I described with both cameras where the predominant light was centered or off to the side...even blank white walls that were fairly evenly lit under such lighting. Whereas there M8.2 came close to nailing the WB correctly (using AWB), the M9's WB (using AWB) was completely off...consistantly. I Just wanted to add these additional facts. Thanks!

 

Dave (D&A)

[sm23221]

I have found the AWB to be off in most cameras. The M9 does have a bias toward the yellow region. A color temp meter looks more and more enticing as time goes on.

[D&A]

Agree...but one of the basica points I was making is why the M9 is so different and so far off (under the conditions I described), compared to the M8/M8.2 with latest firmware updat. Thats what I'm surious about.

 

Dave (D&A)

[jaapv]

AWB seems to be very time-intensive firmware to write, see the time it took to get the M8 to the point where it is now. Contrary to popular belief, the M9 sensor is not just an extended M8 sensor. the microprism architecture is different, which probably does not affect WB, there is an IR filter on it, that certainly makes a difference and the colour filtering in the Red-Green channel is different. I am sure Leica had to start virtually from scratch in their WB algorithms. I have a strong feeling we are witnessing a work in progress here. Having said that, AWB is one of the least important features of digital. There is, after all, no correct WB, just as much as there is no correct exposure. These are creative decisions that we take sitting in front of our calibrated monitors.

[Advertisement]

Hallo,

Look here: Leica M

[D&A]

Jaap Wrote >>> "I am sure Leica had to start virtually from scratch in their WB algorithms. I have a strong feeling we are witnessing a work in progress here. Having said that, AWB is one of the least important features of digital. There is, after all, no correct WB, just as much as there is no correct exposure. These are creative decisions that we take sitting in front of our calibrated monitors."<<<

 

Jaap,

 

We're in complete agreement. Not only with the M8, but look how long and many generations of DSLR's it took Nikon since the introduction of the original D1 (1999) before their AWB began to resemble something close to many recorded scenes. I also beleive the M9 AWB is a work in progress and expect to see changes to it as time goes by, much like we saw with the M8. In shooting RAW, it's a convienience when the AWB gets it into the ballpark, both when qickly reviewing on the LCD screen as well as when having to adjust incredably large numbers of diverse files for a shoot. As you say, it's not manditory as RAW "post processing" and color balance (among other things) stil needs to be applied to the files and of course one's own subjectivity comes into play during that adjustment phase.

 

I just wanted to see if others were observing the same things with their M9's under similar lighting conditions as I described...rather than any one particular M9 responding differently than most others so that it could preclude that particular M9 might have an issue . Thanks!

 

Dave (D&A)

Edited December 28, 2009 by D&A