Jump to content

The M8's Wacky Auto White Balance


Recommended Posts

Advertisement (gone after registration)

Below is a screen-shot of a CS3 Bridge thumbnail panel showing the randomness of the M8's automatic white balance facility. The underlying images were all shot with a 35mm Lux with a 486 filter within minutes of each other. (I shot the whole sequence in 11 minutes.) I used ISO 160 for all but the final five images..

 

My M8 only seems to know two white balance tunes: 3900K @ +12 tint, and 5900K @ +9 tint. Its choice between them seems entirely random; perhaps every even-numbered second it uses one, odd-numbered the other. (It's as good a hypothesis as any I've seen.)

 

I know that this auto white balance matter is an old subject, although it has become less frequently discussed here. But it's been a problem with the M8 since day one. In my opinion we need to hold Leica's feet to the fire to produce much better firmware for this $5,000 camera. Some of the programming in the M8 is downright primitive by any current standards. Like most others here, I've learned to deal with this problem (and others). But it still irks the hell out of me that I have to.

 

78636705.jpg

Link to post
Share on other sites

  • Replies 46
  • Created
  • Last Reply

...I think it's pretty....<g>

 

but seriously, we have all seen the same thing, I photographed a yellow sign outdoors the other day and somehow the camera thought I was indoors under tungsten!

 

I think it has to do with the angle of acceptance of the blue sensor on the front, a strong color or highlight can cause it to misread the color temp.

 

I am not sure firmware will do everything, there is a hardware component.

 

The camera really does not "know" what you are looking at since it doesn't see through the lens.

 

for example, the original mamiya 6 meter was awful, it was a photodiode in the rangefinder window. Wildly inaccurate. The subsequent mamiya 7 meter was very accurate, same idea, but updated hardware. The M8 may only ever get so good, until a revised photo cell is incorporated. M9?

Link to post
Share on other sites

Guest guy_mancuso

Well here is the rub you have daylight and you have tungsten , the camera is really just confused in this situation. Really does not know what you want it to do. This happens with the DMR also in a scene like this. Honestly there really not a lot you can do . The camera only knows 6 settings daylight, tungsten and such but really it only knows 6 kelvin temps and that really is it, it does not know what tempurature to pick here. You basically are fooling the AWB here and it is easy to do. The issue is it does not have a brain and can't think. some camera's maybe better at this becuase if it see's a certain temp it may just stick with it becuase it may not be as sensitive as the M8. it really may come down to the AWB is very sensitive and reason it jumps

Link to post
Share on other sites

I think it's rather an attractive color pattern, too, Robert. ;)

 

Of course I understand that the camera is choosing between extremely fixed settings for each shot. The point that we need to prosecute with Leica is that such programming is not acceptable in 2007. Given the same light my Canon cameras would nail this same scene with extreme consistency, and the cameras would modulate the color temperature in 10 deg K increments (and the tint in single ticks) to make it so. My little G7, which is a terrific camera at 10% of the M8's sticker, would also nail the white balance in this scene from shot to shot. In fact, just about any digital camera would get any scene's white balance at least consistently.

 

I know that I'm singing to the choir and that none of us can force Leica to spend the money to license (or develop) contemporary auto white balance detection technology. So we're left to shrug, take it, or leave it -- rather like mating up with a beautiful, intelligent, worldly lady who just happens to chew tobacco.

 

I just wanted to publicly point out this "old" issue basically to keep it in Leica's face. I know it's something we all want them to fix sooner rather than later. I also know it's something that will cost them margin to fix...which is why it's not already fixed.

Link to post
Share on other sites

I know, if you were depending on jpegs for a quick turnaround to a client you would be chuffed mad at this. And before anyone jumps on me for saying jpeg in church it is absolutely crucial sometimes to have that jpeg as good as it can be. I do kind of like that the raw is all over the map, it gives me options :) that i probably never would have considered...

 

If you read on over to dpreview, they have been on Canon's ass forever about their tungsten rendering, they find it too warm across the board. I like it, it preserves the warmth of the scene. But there is also a case for accuracy too. anyone else besides me who misses the look of EPY? there was a certain inaccuracy to its accuracy if you know what i mean.

 

I hope this is at the top of their list for firmware revisions. that and the shutter-depress-hold winding option.

Link to post
Share on other sites

Although I shoot exclusively in RAW, I still like my white balance to be close if for no other reason than to not have to have to explain to people over and over again that the bad color on my on my chimped LCD screen "doesn't matter" . Lecturing the average person on the advantages of RAW isn't my idea of M8 ownership. Anyway no one believes you and they just think your $5000 camera is an old fashioned piece of shit.

 

For the time being, I just dial in my best guess at the Kelvin temperature and leave it at that. Actually I am getting pretty good at gauging Kelvin temperature. I quess it is my natural tendency to look at the glass as half full rather than half empty.

 

If you ask my Leica should open up the code to the software and offer prizes to those amatuers who write improved code for the M8. I think the M8 community has a larger percentage of computer savy customers than your average DSLR group. Plus, I think we have proven that we are more tenacious and willing to work to improve the product.

 

Just an idea...a crazy one at that but hey, Leica is almost privately owned now and if Herr Whoever wanted to do something like that, it would be very interesting.

 

Rex

Hope springith eternal

Link to post
Share on other sites

Advertisement (gone after registration)

In another thread it was stated that Leica is not able to really improve the algorithmus they choose for AWB. Instead they seem to work on a completely new approach.

 

Whether it’s true I don’t know.

 

Best

Holger

Link to post
Share on other sites

Opening up the code and tools is a great idea and Leica could benefit greatly. They could have the code submitted to them and pick and choose the best code and features and offer it to the none programmers. I would love to write some code for it to do the things I think are important. I would gladly contribute it back in an open source way to leica for free distribution. I'm sure I am not alone in this,

Link to post
Share on other sites

I think it's rather an attractive color pattern, too, Robert. ;)

 

Of course I understand that the camera is choosing between extremely fixed settings for each shot. The point that we need to prosecute with Leica is that such programming is not acceptable in 2007. Given the same light my Canon cameras would nail this same scene with extreme consistency, and the cameras would modulate the color temperature in 10 deg K increments (and the tint in single ticks) to make it so. My little G7, which is a terrific camera at 10% of the M8's sticker, would also nail the white balance in this scene from shot to shot. In fact, just about any digital camera would get any scene's white balance at least consistently.

 

I know that I'm singing to the choir and that none of us can force Leica to spend the money to license (or develop) contemporary auto white balance detection technology. So we're left to shrug, take it, or leave it -- rather like mating up with a beautiful, intelligent, worldly lady who just happens to chew tobacco.

 

I just wanted to publicly point out this "old" issue basically to keep it in Leica's face. I know it's something we all want them to fix sooner rather than later. I also know it's something that will cost them margin to fix...which is why it's not already fixed.

 

I also have been frustrated by this problem and not just in mixed indoor lighting. I shoot primarily landscapes and I see this frequently in shot sequences taken seconds apart. I'm resigned to dealing with this in RAW processing, but it would be nice to have a more modern/accurate auto WB algorithim in-camera.

Link to post
Share on other sites

My M8 only seems to know two white balance tunes: 3900K @ +12 tint, and 5900K @ +9 tint. Its choice between them seems entirely random; perhaps every even-numbered second it uses one, odd-numbered the other. (It's as good a hypothesis as any I've seen.)

 

I did a quick review of a few day's shooting, indoor and outdoors, and on the frames that I know I hadn't touched, found only 3550 K and 5650 K (that's C1's interpretation on reading the EXIF), generally with tint -1. Also a few that said 7400 K when the shot was a little over exposed. It's probably not random, although it does seem that the decision boundary between 3550 and 5650 falls right in the middle of some fairly common conditions. I can see why Leica suggested that they were doing a thorough review of AWB.

 

It makes you wonder how they are supposed to sense color temperature in the first place. There are four sensors available, as shown in the "anatomy"thread, from which I pinched the relevant shot and attach it below: the main exposure sensor, the "blue dot" and two that are believed to be for flash control. To detect the relevant amount of blue and yellow you need to compare two different sensors with different characteristics. I've stuck the camera on a tripod and covered the blue dot without changing the color balance reading, so iI'm left with no clear idea how they do it, except to wonder about the two "flash sensors" on the sides of the main diode.

 

Covering the blue dot, BTW, does fool the camera into thinking that the lens aperture must be wide open (it's dark outside and bright inside), which calls up extra vignetting correction.

 

scott

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

Below is a screen-shot of a CS3 Bridge thumbnail panel showing the randomness of the M8's automatic white balance facility. The underlying images were all shot with a 35mm Lux with a 486 filter within minutes of each other. (I shot the whole sequence in 11 minutes.) I used ISO 160 for all but the final five images..

 

...

 

78636705.jpg

 

Ken, assuming you took the shots, which has the "right" color balance? Or none?

 

Maybe after a few alternate AWBs, it would be time to dial in a manual color temp that suits you.

Link to post
Share on other sites

Good question, Philip. Actually the answer is "none of the above". The warmer 5900K white point is closest to what I'd consider to be correct. My manual adjustment places the white balance at around 5200K.

 

I wish I had shot this same scene with a different camera for comparison. But at the time I was actually trying to take photos, not be an amateur camera tester.

Link to post
Share on other sites

My Crazy Sons, shots taken with M8 and 90 mm

 

2 pictures shot in a coffee shop big windows right and behind me, AWB way off in the first picture compare to the second one, shots are done maybe 5 seconds apart.

Is there a problem in general with the AWB?

My camera is it not doing all the time, but to often for my feeling.

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

My Crazy Sons, shots taken with M8 and 90 mm

 

2 pictures shot in a coffee shop big windows right and behind me, AWB way off in the first picture compare to the second one, shots are done maybe 5 seconds apart.

Is there a problem in general with the AWB?

My camera is it not doing all the time, but to often for my feeling.

 

We have to accept at least for now that AWB does not work to an acceptably reliable percentage, therefore it does not work.

 

It's one of the less-than-charming prices we pay for a camera that rewards amply in other ways.

 

Of course we should clamour for improvement. And we should also avoid shooting jpeg only. But in the end, we all know that the M8 does this and that we should never trust the AWB to get it right.

 

But think: how lazy are we with other manufacturers' gear that gets AWB close enough for us not to bother adjusting in PP, even though it is not right?

 

So we shoot RAW, hope for better firmware (and frankly if it were easy it would have been fixed already) and every day we use the camera, we make a decision to live with its quirks in order to benefit from its abilities.

 

Every shot we take, when we could have traded our M8 in for something else, reflects that decision. Most of us keep the camera.

 

t

Link to post
Share on other sites

So we shoot RAW, hope for better firmware (and frankly if it were easy it would have been fixed already) and every day we use the camera, we make a decision to live with its quirks in order to benefit from its abilities.

 

Every shot we take, when we could have traded our M8 in for something else, reflects that decision. Most of us keep the camera.

 

t

 

Oh brother. I hope you'll give the M8 blessing next Sunday, too.

Link to post
Share on other sites

Frankly, I would rather that it was consistently wrong in a particular direction than randomly wrong in unexpected ways. Of course, better yet would be for it to be constantly right!:)

Link to post
Share on other sites

AWB is broken; Leica's acknowledged this, haven't they? Yes, it needs to be fixed, but not before they fix other, more important stuff, like random banding from that edge light source and read-out problems.

 

Scott--the M8 doesn't have an outboard WB sensor like everyone else. They take WB, IIRC, from a white blade on the shutter panel. I'm thinking maybe there's a physical limit on how accurate it is on faster shutter speeds or something?

 

@ Theo--were those shots taken with the same shutter speed? Just curious.

 

Who knows. The best way around this--simple and very quick--is to make a custom WB. Voila--JPEG or RAW--problem solved.

Link to post
Share on other sites

Hello Jamie,

 

I have checked the files both are the same shutter speed 1/45 sec, Iso 160.

 

Quit frankly in my opinion, this fault should be on the toplist of Leica to FIX it.

 

I, and I think a lot of us can show more pictures like this and show us the same problem, this is not my first time that I discover it, I discover it after I get my camera back from New Jersey for adjusting the viewfinder.

 

Jamie Wrote:

The best way around this--simple and very quick--is to make a custom WB. Voila--JPEG or RAW--problem solved.

 

My answer:

If I have to do all the changes when I want to make a shot, then the moment is gone, I want a camera that works and we all expect that from Leica. Special for street photography I have no time to adjust everything, the camera have to do his job, come on Jamie be real here.

 

 

I hope that Leica read this tread and take notice of it to solve this.

 

Theo

Link to post
Share on other sites

Archived

This topic is now archived and is closed to further replies.

  • Recently Browsing   0 members

    • No registered users viewing this page.
×
×
  • Create New...