M9 Banding and noise issues are gone!

[Erik Gunst Lund]

Advertisement (gone after registration)

I'm working as Patent Engineer, educated as mechanical engineer B.SC. evaluating and working with all sorts of inventions within; electronics, software, hardware, mechanical, hydraulics, design etc. been in the same engineering company since 1989 we have in-house about 500 engineers and researchers.

Part of my job is as Pro photographer, NPS and LPS, Shooting digital since Nikon D1 came on the streets, then: D1Xs, D2Xs, D3 & M8u and a lot of Modification of lenses especially conversions and chipping, adding CPU's to Nikon MF lenses.

 

I am just interested in finding the correct cause of your problems or lack of problems with you M9... Sorry if I have offended you.

[Paul J]

No offense taken at all, I asked out of interest. If the problem doesn't make sense from an engineering stand point then I'm baffled.

Edited September 27, 2012 by Paul J

[Advertisement]

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

[brianv]

I just looked at the cropped sections of the images showing the banding.

 

In images, does the banding run across the entire frame OR does it start into the image, starts at a pixel, then continues to the end of the image?

 

The noise is coming into the image in the analog domain, before being converted to digital. The question of where it starts is related to it being burts of noise, and the band being created due to the CCD shifting. If it is throughout the entire image, does not start at one pixel in the line, then the noise would be likely constant.

 

The Engineering term for "freakish" events like this is "Gremlins". Regarding noise in systems, and changing components that would not seem to make any difference that make the problem go away- happens more than one might think.

 

After reading this thread, I reset the image review of my M8 to "OFF". The extra power draw on the system to power up the LCD as data was being written to the card occurring at the instant of making the next exposure in a sequence: Gremlin Food.

[Paul J]

Interesting insights, thanks.

 

Here are two samples. One with the new card and one with the old card. Both shot with the lens cap on. FWIW Both shot at 1/4000th @ f4.

 

I have increased the exposure on these to show the difference. The third frame is one of the unadjusted frames as a reference point.

 

So just to clarify - the banding showed up even when correctly exposed. It was always there, showing up mainly in dark areas or areas that had enough contrast to show the difference between noise.

 

I don't see any noise or pattern banding now. I have greatly reduced noise and colour blotching also. The only artefacts here are are normal for any other digital camera and also show how the sensor is made in two halves.

 

Edited to add - these are not crops. They are displaying the whole frame

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 September 27, 2012 by Paul J

[k-hawinkler]

Paul,

 

So, I would like to do this test with my camera exactly as you did. What matters?

 

Lens cap on, M9, lens?, ISO=?, 1/4000, f/4, a handful of different memory cards, shoot DNG? + JPG?, process in CS6? or LR4? or C1?, which settings in M9, which in image processing software?, full battery, memory card formatted in M9, use card reader?, transfer images via USB?, save data to card first, use tethering to transfer directly from M9 to Mac laptop?

 

Please, could you spell out all that matters?

 

Thanks.

Edited September 27, 2012 by k-hawinkler

[Paul J]

Paul,

 

So, I would like to do this test with my camera exactly as you did. What matters?

M9, lens?, ISO=?, 1/4000, f/4, a handful of different memory cards, shoot DNG? + JPG?, process in CS6? or LR4? or C1?, which settings in M9, which in image processing software?

 

Please, could you spell out all that matters?

 

Thanks.

 

Sorry, I thought someone would ask and that I should have had all that as I was typing it. I'm trying to get a job retouched and out the door at the moment!

 

M9 with 1.176fw

Noctilux f0.95

ISO 160 - 1/4000th @ f4

**Lens cap on**

Lightroom 4 adjusted with +5 exposure. Pulled the highlights in the curves to far left with auto sync on both frames.

 

I don't think it matters which lens or software you use if you are comparing cards for yourself, just make sure you have consistency obviously.

[AlanG]

Advertisement (gone after registration)

This person also traced his banding to the card.

 

Yet Another Leica M9 Bug | Mark's Photography Blog

[thighslapper]

Observations:

 

Banding has a fixed periodicity

 

L and R side sensor read-out not exactly the same.

 

Worse at bottom of sensor (ie. top of frame)

 

For my money this is interference or noise in the sensor affecting the channel read-out .......

 

I'm just a simple soul, but I cannot conceive of how this can be generated IN the card or the data transfer process to the card.

 

I think you have a ropey sensor and that changing the card has just lessened the problem..... again probably all down to voltage issues or some sort of interference in the sensor cells/channels.

 

I have NEVER had banding on any M9 photos, even 60 sec night exposures, irrespective of whatever card I have used (and thats 2 M9's and an M9P)

[Paul J]

I almost didn't post the samples because I thought someone would say this.

 

All digital looks like that in some form under these sort of adjustments. The M9, like many sensors are made in two halves. Which is what you are seeing. Do the same with your M9's and you will have the same affects no doubt.

 

I made this test to show how the banding looked in answer to brianv. Also to demonstrate how it looks with the two different cards.

 

Here is a Phase One P65+ under the same test. It has incredible IQ.

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 September 27, 2012 by Paul J

[brianv]

If the problem was the sensor, it would likely be across the entire frame. A CCD reads out the signal sequentially. It's like something is switching on, creating the noise as the analog signal is being digitized. The noise looks very even, not intermittent. There are multiple readout muxes, and multiple ADC's on the sensor. Maybe the difference in levels across the scan lines is caused by the noise creeping in on one ADC more than the other.

 

Gremlin. Changing the SD card could have changed the voltage draw, simply been less noisy, changed the impedance, etc. If I worked at Leica I'd offer you a new camera to get this into the Lab and find out first hand. That's what Pseudo-Random Bit Sequences, O-scopes, and logic analyzers are for. And you fixed the problem without any of that, Congratulations!

[k-hawinkler]

If the problem was the sensor, it would likely be across the entire frame. A CCD reads out the signal sequentially. It's like something is switching on, creating the noise as the analog signal is being digitized. The noise looks very even, not intermittent. There are multiple readout muxes, and multiple ADC's on the sensor. Maybe the difference in levels across the scan lines is caused by the noise creeping in on one ADC more than the other.

 

Gremlin. Changing the SD card could have changed the voltage draw, simply been less noisy, changed the impedance, etc. If I worked at Leica I'd offer you a new camera to get this into the Lab and find out first hand. That's what Pseudo-Random Bit Sequences, O-scopes, and logic analyzers are for. And you fixed the problem without any of that, Congratulations!

 

 

Hi Brian,

 

Congratulations to Paul indeed for fixing his M9. I understand your explanation as follows:

 

Certain memory cards seem to influence the electrical system of some M9s in such a way that noise, generated upstream of the final storing of the data on the memory card, intermittently finds its way primarily into the dark parts (least significant bits) of some images. This happens more likely when using high ISO values.

 

That also implies that the memory cards themselves did not directly corrupt the images, but only indirectly by influencing the electrical system of the camera.

Please, correct me where I am wrong with this characterization.

 

Thanks.

[brianv]

That's what I believe is happening- sometimes it helps to remember that at some level, everything is analog. "Digital" is just an analog signal with a threshold applied.

 

I worked on the Data Acquisition System (DAS) for "Digital Imagers" a LONG time ago. The second system used a 14-bit ADC, ~1984. The DSP chip inroduced 9 bits of noise into the signal as it was being digitized. Quickly learned the importance of isolating digital and analog components, and that every wire is an antenna....

[k-hawinkler]

That's what I believe is happening- sometimes it helps to remember that at some level, everything is analog. "Digital" is just an analog signal with a threshold applied.

 

I worked on the Data Acquisition System (DAS) for "Digital Imagers" a LONG time ago. The second system used a 14-bit ADC, ~1984. The DSP chip inroduced 9 bits of noise into the signal as it was being digitized. Quickly learned the importance of isolating digital and analog components, and that every wire is an antenna....

 

 

Hi Brian,

 

Many thanks. So true.

[Paul J]

really well put Brian. Thanks.

[Paul J]

Certain memory cards seem to influence the electrical system of some M9s in such a way that noise, generated upstream of the final storing of the data on the memory card, intermittently finds its way primarily into the dark parts (least significant bits) of some images. This happens more likely when using high ISO values.

 

That also implies that the memory cards themselves did not directly corrupt the images, but only indirectly by influencing the electrical system of the camera.

 

Brilliant!!!

[ringohung]

I have had this banding problem for a long time. I knew it was also a card problem because I have tested a few cards and found out that different cards have different degrees of banding noise. However, when I updated to the latest M9 FW, majority of the banding problem are gone now, although I still got some problems occasionally when the power drops below 25%. I hope this help.

[k-hawinkler]

I have had this banding problem for a long time. I knew it was also a card problem because I have tested a few cards and found out that different cards have different degrees of banding noise. However, when I updated to the latest M9 FW, majority of the banding problem are gone now, although I still got some problems occasionally when the power drops below 25%. I hope this help.

 

 

Many thanks.

Your observations seem to be consistent with two of the three improvements of firmware version 1.196 as pointed out by Leica, namely quote:

"

• Further improvements of SD-Card compatibility.

 

• Improved power management.

The camera is now measuring battery stability internally. In case of weak battery stability, the camera switches off automatically.

Using old batteries could have lead to camera crashes and banding stripes under certain conditions with the previous firmware versions. Especially the usage of nearly 6 years old M8 batteries proved to be critical.

"

 

Well, in the past on my M9 I have seen system crashes with a new and fully charged battery.

I don't believe everything I read and spinning of a message seems to be common place.

But I certainly welcome the progress Leica has made with the 1.196 firmware update.

More updates seem to be forth coming http://www.l-camera-forum.com/leica-forum/2188700-post55.html, quote:

 

"Firmware-Updates zur M9 wird es geben"

 

Great.

[Reds]

What kind of snake oil, pixie dust is this whole thread about! Made my morning.

 

So digital 0s and 1s written to an output medium, via a processor can mutate and generate specific, 'recognisable to humans' banding and noise patterns, completely in the context of photography. (Rather than just unreadable, or damaged files). Depending on the output medium used.

 

Must be the cosmic ray effect, splange valve inversion code

 

What is this forum coming to....... That and arguing for 4 pages over 5mm on a camera no-ones even held yet. Must be silly season.

[k-hawinkler]

What kind of snake oil, pixie dust is this whole thread about! Made my morning.

 

So digital 0s and 1s written to an output medium, via a processor can mutate and generate specific, 'recognisable to humans' banding and noise patterns, completely in the context of photography. (Rather than just unreadable, or damaged files). Depending on the output medium used.

 

Must be the cosmic ray effect, splange valve inversion code

 

What is this forum coming to....... That and arguing for 4 pages over 5mm on a camera no-ones even held yet. Must be silly season.

 

:eek:

[Advertisement]

Hallo,

Look here: Leica M

[pop]

So digital 0s and 1s written to an output medium, via a processor can mutate and generate specific, 'recognisable to humans' banding and noise patterns, completely in the context of photography. (Rather than just unreadable, or damaged files). Depending on the output medium used. .

 

There is indeed evidence that images made by an M9 exhibit banding and there is evidence that the banding appears to vary with the type of storage card used.

 

What you call an "output medium" is in reality a piece of electronic gear which does indeed depend on a power supply and on signal quality. Not all "storage cards" are created equal.

 

It is not all that unusual that the quality of stored images or sounds vary with the quality of the output medium used and that not all output media gave the same results on every type of hardware. Tape (on reels and in cassettes) and film come to mind.