Weirdest thing about banding

[jrc]

Advertisement (gone after registration)

The oddest thing about the band phenomenon is not that it occurs, but that Leica seemed genuinely unaware of it -- yet it was discovered by THE VERY FIRST PERSON who took commercial delivery of an M8 (Pascal.) The first guy we know of who went down to the store and said, "Here's my money, give me the camera." If this has been tested for months (and we have reports of people testing it last summer) how could it have not been seen? I have one answer to that, just for myself -- and that is, the streaking was introduced at a late stage of firmware alterations. Like in the last two weeks, somebody said, "You know, we just need to put a couple of final touches on this baby and we can pat it on the ass and send it out." If I remember correctly, Sean Reid didn't use photos in his first M8 review because he didn't have the latest firmware, and that was only a couple of weeks ago. If this is the case, we're in good shape. If that's not the case, then, hmm, somebody needs to rework the testing regimes.

 

JC

[Advertisement]

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

[stunsworth]

Who knows John, if it's a hardware problem I can't believe that it went undiscovered during all the time the camera was in development.

 

Brian Read on the LUG has pointed out there that the 'streak' on one of the images posted there was 400 pixels wide on an 800 pixel image. Seems a bit of a coincidence that it's exactly hald of the image width. His thoughts were that this illustrated that it was probably a firmware rather than a hardware problem.

[pascal_meheut]

Because it doesn't appear on 99% of the pictures I took. I had almost no time to shoot so one day, I left work earlier and went for a walk by sunset. All the pictures I took this evening/night were ok except for one (or 6 as I shot several time with the same framing)

 

So it was easy not to notice the banding and I've just been unlucky.

[B&W]

I do find it even odder that Leica do not take the opportunity to post there action in this forum………….. but I presume their policy is to be very quite in all matters.

 

Up till now we have been informed by other member’s witch I am very thankful for:)

[rvaubel]

...........The oddest thing about the band phenomenon is not that it occurs, but that Leica seemed genuinely unaware of it -- yet it was discovered by THE VERY FIRST PERSON who took commercial delivery of an M8 (Pascal.)

 

JC

 

In retrospect, I first saw it in the JPEGs smuggled out of Photokina.

However, the effect didn't look so bad as to really register on my radar. That version of the software was 1.04 so maybe it got worse with the later version. Or maybe not. I also say the banding in the man in the cafe photo but I didn't come out of my tree then either.

 

Hindsight is always 20-20

 

Rex

[rob_x2004]

Dont be silly. Of course they knew about it. Probably find a lot of the testers outside Leica knew about it as well.

[M'Ate]

Advertisement (gone after registration)

Surely never an oversight. This must have been a judgement call. Undermining a product launch; capitalising on the enthusiasm of the buying audience boosted by the very positive reviews, versus, a quick fix programme and a firmware fix.

 

Leica controlled the use of images for weeks in advance of the launch, controlled the issue of cameras to the reviewers and started shipping a little earlier than most expected. Very well managed product launch, so why would they ever miss this banding issue.

 

The reviewers never had a chance to raise the flag, because their selected samples didn't exhibit the banding -

 

Criticism - not at all, it's risk management. They are still in control and know how and when the fix will be in place. Leica would not go into a launch with a problem they didn't have the answer to, IMO. They know the solution and the cost of the fix. The good news is they decided to go to market with it and therefore will recover.

[mwilliamsphotography]

Tried to further test for Banding with the M8 using controlled studio conditions so all conditions would remain constant except camera settings.

 

Set up a Profoto head in front of black velvet to simulate absolute black with a very strong specular light source directly into camera.

 

Here are 4 shots with the first 2 showing nothing, and the last two (#3 & 4) showing banding.

 

#1 and 2 were exposed properly for these purposes. In other words, in the case of #2 not trying to overexpose for the background velvet which is true black and should read that way ... while recording the off white overhead beam lit by tungsten overhead lights.

 

#3 was deliberately underexposed and lifted a small amount in processing ... #4 is lifting the same file even more to try and get it to the level of #2 exposure.

 

I'll welcome any fix Leica provides, and in the meantime make sure I keep an eye on exposures in conditions like these.

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!

[rob_x2004]

I would have called it a commercial decision. Leica dont consider the problem warranted holding back from going to market. Just the way it is. You dont have to condemn or defend it. Just reflects Leica's true to form.

[rvaubel]

Tried to further test for Banding with the M8 using controlled studio conditions so all conditions would remain constant except camera settings.

 

Set up a Profoto head in front of black velvet to simulate absolute black with a very strong specular light source directly into camera.

 

Here are 4 shots with the first 2 showing nothing, and the last two (#3 & 4) showing banding.

 

#1 and 2 were exposed properly for these purposes. In other words, in the case of #2 not trying to overexpose for the background velvet which is true black and should read that way ... while recording the off white overhead beam lit by tungsten overhead lights.

 

#3 was deliberately underexposed and lifted a small amount in processing ... #4 is lifting the same file even more to try and get it to the level of #2 exposure.

 

I'll welcome any fix Leica provides, and in the meantime make sure I keep an eye on exposures in conditions like these.

 

Did you notice the mirror dark image of the light symmetrical to the vertical centerline of the frame? Thats been a consistent pattern.

 

Try taking the same shot again but move the light over to the 1/3 point horizontally. I bet you well see a negative image of the light symmetrical about centerline.

 

God I wish I had my own M8 to do these tests that would be very:cool:

 

Rex...waiting for my very own

[ho_co]

Did you notice the mirror dark image of the light symmetrical to the vertical centerline of the frame? Thats been a consistent pattern.

 

Try taking the same shot again but move the light over to the 1/3 point horizontally. I bet you well see a negative image of the light symmetrical about centerline.

And it's clearly sensor-related, not internal reflections. The first image clearly shows a group of reflections, and they are bunched in a diagonal since the light is at a diagonal position from the center. Fourth image has internal reflections as well, but they're along the horizontal center since the light source is as well.

 

But the 'green smudge' is on the same level as the light source.

 

Might be good to do the shot at 1/3 point horizontally AND vertically. That should generate both the dark mirror image and internal reflections. That in turn would prove it's a sensor issue.

 

--HC

[cbretteville]

I would have called it a commercial decision. Leica dont consider the problem warranted holding back from going to market. Just the way it is. You dont have to condemn or defend it. Just reflects Leica's true to form.

 

Leica, just as all other publicly traded companies, have share holders to keep happy. Share holders like new products getting to market on schedule.

 

- Carl

[marknorton]

Rex/HC, I've done the tests you described. Whereever you put the light source in the frame, the green reflection is at the same height and is a lateral mirror image of it.

[rvaubel]

Rex/HC, I've done the tests you described. Whereever you put the light source in the frame, the green reflection is at the same height and is a lateral mirror image of it.

 

Mark

 

Although I dont have my M8 yet, I will make another prediction.

 

The banding on the blownout side is NOT mirrored on the other side. Not in the least.

 

Am I right?

 

Rex

[wattsy]

Mark and Howard do seem to be right about the green blob phenomenon - it does seem to be sensor related rather than an optical phenomenon. In a way, this is good news because, if it was optical, there is little that could be done other than change or increase the coatings on the sensor glass and/or rear lens element. If it's sensor related, Leica/Kodak/Jenoptik at least have an opportunity to fix it with a firmware or hardware alteration (although Mark suggests in another thread that it will be a harder problem to fix than the streaking).

[marknorton]

Mark

 

Although I dont have my M8 yet, I will make another prediction.

 

The banding on the blownout side is NOT mirrored on the other side. Not in the least.

 

Am I right?

 

Rex

 

Yes, Rex, I think you are correct. In this sample I posted, you can see the banding is confined to the side of the sensor where the bright light is and there's only the ghost mirror image on the other side.

 

http://www.leica-camera-user.com/digital-forum/8642-m8-streaking-hi-iso-3.html#post86069

[rvaubel]

Yes, Rex, I think you are correct. In this sample I posted, you can see the banding is confined to the side of the sensor where the bright light is and there's only the ghost mirror image on the other side.

 

http://www.leica-camera-user.com/digital-forum/8642-m8-streaking-hi-iso-3.html#post86069

 

Mark

 

Another prediction

 

For a blownout image to produce banding the image must be at least 4 F stops past clipping. In other words, an image that is 3 F stops past clipping will never band but an image that is 5 F stops past clipping will always band . Be carefull when you measure the blown highlight since some of it may be less than 4 F stops and some of it may be more than 4 F stops

 

I wish I had my own M8 to experiment with:mad:

 

Rex

[marknorton]

I wish you did as well, I think you'd be able to provide a useful extra pair of eyes on the analysis...

 

As regards exposure, I measured the light with a spot meter and correctly exposed, it's fine. Even +1 EV seems to show some banding.

[Advertisement]

Hallo,

Look here: Interesting Leica gear

[mwilliamsphotography]

FYI, here are 2 shots taken at the same time with a 5D.

 

BTW, both cameras were set to ISO 160 ... even on it's lowest setting the Profoto head would blow out the shots when set to any higher ISO.

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!

[scott kirkpatrick]

Yes, Rex, I think you are correct. In this sample I posted, you can see the banding is confined to the side of the sensor where the bright light is and there's only the ghost mirror image on the other side.

 

 

A couple of comments. The KAF10500 reads its data out to both sides along the horizontal axis (row-wise), and the streaks seen by several people stop at the midpoint. The Man in Cafe picture at 1250 streaks do this. The night scene in RPO's LeicaPassion article with streaks also shows them stopping at the midpoint of the frame. There may be a negative ghost image of the leftmost of the bright window areas mirrored across the frame in the Man in Cafe 1250 picture -- I'm not sure. Anyway, artifacts that exhibit the symmetry of the chip, and not the symmetry of the overall frame seem likely to have their origin in the data coming off the chip.

 

The phrase "banding" apparently means this sort of streaking to pixel-peepers, but it is used differently in the image processing community. Look at the help files for Capture One (if you have an M8 you have Capture One, if not, it's available as a free trial download). The banding suppression slider is intended to smooth transitions between colors that are represented with too few bits, and it does this by "adding noise bits," or in other words, by dithering the lower order bits of the image. Capture One also has a "pattern noise removal" action which you can check. It appears to be used for files coming from chips in the KAF family, such as PhaseOne backs, Some Olympus products, and the Leica DMR. I am familiar with "pattern noise" from my E-1 in its early days. Nobody is reporting this artifact for the M8.

 

scott