Jump to content

Poll for Article - M8 Vertical Band


sean_reid

Recommended Posts

Guest guy_mancuso

Advertisement (gone after registration)

That is correct and exactly what i have found also the dark frame reduction comes in around 1/30 of a second and it does vary , how much i can't figure . But in 1.09 we are seeing it more than in 1.06. That is the answer right there what did leica do in 1.09 from 1.06 , my guess is they are trying to optimize the processing and the pushed the gain up on the sensor. Basically it's over tweaked. I think what happens is the sensor data is split in two and sends signals to process at the same time and you are getting crossover than combine that with higher ISO's which the files are usually bigger and it slows down the signal and overlaps. It's another theory which i lost one of those already. LOL

Link to post
Share on other sites

  • Replies 85
  • Created
  • Last Reply

Top Posters In This Topic

Top Posters In This Topic

Posted Images

Hi Guy,

 

Actually, what I'm finding so far is that (based on torturing various test files) the problem is actually worse on my 1.06 camera (Gen 1) than on my 1.09 (Gen 2) example. It's possible that it's just a sample difference between the two cameras but it looks like Leica made progress with this aspect even though it wasn't making headlines on the forum. More after I do some more testing tomorrow.

 

Best,

 

Sean

Link to post
Share on other sites

Guest guy_mancuso

I'm traveling but will be checking in. also have some test planned for the 28th. I do think all of this is firmware and that is the good news. Because the hardware fix works as planned

Link to post
Share on other sites

Well Irakly has posted a thread that clearly shows he DOES have the banding problem, so it's getting pretty universal now as we dig into it.

 

I'll try some even higher shutter speeds of real objects in low light using something dark where the band appears. This will help investigate the higher shutter speed with no noise supression theory.

 

I have the Aptus 75, and have experienced the "centerfold" issue only once so far. Leaf just launched a new software update that substantually improves the performance @ ISO 800, but I do not know if it has addresssed the centerfold issue, the update doesn't mention it. It is exactly like the other issue shown here. But I can't understand why they would stitch two sensors together for the small Leica sensor. Maybe it has something to do with manufacturing with micro lenses?

Link to post
Share on other sites

It's a single sensor but the samples are read out using two channels, one to the left, one to the right, all to do with speeding things up compared to, say, the DMR.

 

The problem is that as soon as you have two different routes to measure pixel values, you run the risk of the two not matching and will then see a step change across the centreline.

 

I expect there's some internal calibration process to match the two which has gone awry in these examples. The camera will need to do this often as things will drift and temperatures change. It can establish a black reference any time it wants to as there are masked pixels on the sensor and my guess is that it checks pixel values either side of the centre line and allows only a small variation between them before re-calibrating. My further guess is that if you took a picture of a black/white card with the join exactly on the centre line, the camera would get really confused.

 

(sorry, Scott got there before me and I was distracted while posting)

Link to post
Share on other sites

Advertisement (gone after registration)

......if people with first generation M8s are seeing this particular kind of vertical band also then I would like to hear from them as well.........

First Gen. camera with firmware 1.06

First Gen. camera with firmware 1.09

Second Gen. camera with firmware 1.09.............Sean

 

Sean,

 

As my several posts have stated I have only been able to take pictures with a " First Gen. camera with firmware 1.06". Since then the camera has been in Solms.

 

I had severe vertical banding offset towards the right of centre, see pictures in other threads, both in normal exposure at ~1/250, 2500 ISO and with slightly underexposed shots at 1/30, 320 ISO. The band is only visible from DNG files, never from JPEG, and is independent of the conversion software used.

 

I also had a similar width dark vertical band towards the left side in a few pictures, see other thread.

 

I do hope this puts an end to those posts that claim and/or infer that this is a new problem. I fully accept that the second generation cameras and firmware 1.09 may have affected the severity and occurrence of the problem, but that is a different point.

 

The reason I did not put more information about this on the forum back in early November was simply that I had had it suggested to me that the vertical banding was closely related to the horizontal banding issue and would be cured when the camera was up-rated. It seems that that is still the opinion of some Leica people but the evidence is mounting that they are separate problems or that, if related, the current cure is not comprehensive.

Link to post
Share on other sites

I have one of the first cameras delivered in Germany. There was no vertical but horizontal banding. After the hardware and software upgrade in Solms no banding at all. I'm looking forward to recieve my IR filters.:)

 

Best wishes and Happy Hollidays to everybody

Link to post
Share on other sites

That looks like an issue that people are observing on high-end digital backs with wide angle lenses (e.g. Leaf Aptus-75, 33MP back). Its called there "centerfold" issue and seems to relate to the fact that these large sensors are assembled from more than 1 chip which can cause problems in calibrating each sub-sensor equally...

 

I have the same split (centerfold) effect at ISO 2500, 1/60 second (picture taken with the cap on the lens) + ev pushed up by 2.5 (or more), both in RAW and in JPEG files out of the camera. However, without exposure values being pushed up and in real life pictures, the effect is not noticeable. Neither does the issue show up at 1/15, not even in the torture test - or just very, very faintly - when exposure values are pushed up (why? - is this related to the in-camera noise reduction?). All this is nothing to worry about, I would assume.

 

No vertical band whatsoever in any of my pictures. My camera was bought in early November and just came back from the hardware fix. Firmware is 1.09.

 

Wolfgang

Link to post
Share on other sites

Guest guy_mancuso
Hi Guy,

 

Actually, what I'm finding so far is that (based on torturing various test files) the problem is actually worse on my 1.06 camera (Gen 1) than on my 1.09 (Gen 2) example. It's possible that it's just a sample difference between the two cameras but it looks like Leica made progress with this aspect even though it wasn't making headlines on the forum. More after I do some more testing tomorrow.

 

Best,

 

Sean

 

 

Sean i think it has been there all along and one or two things may have brought it more forward in 1.09 is maybe the hardware assembly board fix that made it work correctly brought it out or maybe if leica worked on the amplification of the sensor to maybe try to reduce some noise. The good news is it is just firmware and they just need to find that happy point. Personnally if there really pushing the sensor to get a better ISO 2500 than forget it . I would rather us get a exceptionally 1250 and call it a day. Pushing a CCD sensor to 2500 maybe pushing it too far. There is not a CCD sensor that i know of that goes to ISO 2500 and it maybe just a limitation of CCD technology. Something to think about. I think a question that may support this on the hardware fix , is there a difference between a unfixed hardware M8 running 1.09 than a fixed hardware running 1.09 and if there is than we can look at the hardware fix that maybe bringing it out more. Or a fixed hardware M8 running 1.09 compared to 1.06 than we can see if it is the firmware. I think lookig at it like this you find the root cause than go back in and adjust the firmware accordingly. The hardware fix works but fixing it and adding 1.09 at the same time maybe just brought it out more. So i think leica needs to test that than go in and work on the split processing for channel 1 and 2 and sync that up because it looks like crossover to me and our torture tests indicate that is exactly what is going on and the dark frame reduction is really just hiding it

Link to post
Share on other sites

Well Irakly has posted a thread that clearly shows he DOES have the banding problem, so it's getting pretty universal now as we dig into it.

 

I'll try some even higher shutter speeds of real objects in low light using something dark where the band appears. This will help investigate the higher shutter speed with no noise supression theory.

 

I have the Aptus 75, and have experienced the "centerfold" issue only once so far. Leaf just launched a new software update that substantually improves the performance @ ISO 800, but I do not know if it has addresssed the centerfold issue, the update doesn't mention it. It is exactly like the other issue shown here. But I can't understand why they would stitch two sensors together for the small Leica sensor. Maybe it has something to do with manufacturing with micro lenses?

 

Hi Marc,

 

Irakly's camera seems to have the light streaking problem that is fixed by the hardware change.

 

Cheers,

 

Sean

Link to post
Share on other sites

Sean i think it has been there all along and one or two things may have brought it more forward in 1.09 is maybe the hardware assembly board fix that made it work correctly brought it out or maybe if leica worked on the amplification of the sensor to maybe try to reduce some noise. The good news is it is just firmware and they just need to find that happy point. Personnally if there really pushing the sensor to get a better ISO 2500 than forget it . I would rather us get a exceptionally 1250 and call it a day. Pushing a CCD sensor to 2500 maybe pushing it too far. There is not a CCD sensor that i know of that goes to ISO 2500 and it maybe just a limitation of CCD technology. Something to think about. I think a question that may support this on the hardware fix , is there a difference between a unfixed hardware M8 running 1.09 than a fixed hardware running 1.09 and if there is than we can look at the hardware fix that maybe bringing it out more. Or a fixed hardware M8 running 1.09 compared to 1.06 than we can see if it is the firmware. I think lookig at it like this you find the root cause than go back in and adjust the firmware accordingly. The hardware fix works but fixing it and adding 1.09 at the same time maybe just brought it out more. So i think leica needs to test that than go in and work on the split processing for channel 1 and 2 and sync that up because it looks like crossover to me and our torture tests indicate that is exactly what is going on and the dark frame reduction is really just hiding it

 

Hi Guy,

 

I think it has been there all along and it just hasn't gotten much attention because, in many cases, it isn't posing a problem. (Or there's been a huge conspiracy among all of us who own the M8 to hide this from other people; a larger version of the reviewer conspiracy...<G>)

 

Based on what I'm seeing, the new camera and new firmware show the problem to a lesser degree in the side-by-side tests I'm now doing. Of course, I only have one of each type camera (old/new) to test so the differences might just be individual variations among those two specific camera (although I suspect that isn't the case).

 

I don't know if the problem can be solved with a firmware change or not and I suspect that no one outside of Leica/Kodak/Jenoptik knows that for sure. Based on what I'm seeing in some tests, I'm beginning to think that the banding is related to the read out of the two sides of the sensor (based on some interesting patterns I'm seeing in JPEG/RAW comparisons). But I'm not an engineer so what I'll mostly present in the next M8 article will be a summary of people's experiences, based on this survey and other communications as well as a discussion of what I'm seeing in RAW and JPEG files from the "old" and "new" M8s.

 

To All,

 

I'm still hoping to get more feedback from people who own the M8. Please chime in whether you've seen the vertical banding or not with your camera.

 

1. Are you using the first or second generation (hardware changed) M8? Which firmware version?

 

2. Have you seen the vertical band (discussed elsewhere on this forum) in your particular M8?

 

3. If yes, does it appear in properly-exposed pictures using your normal workflow? If so, at what ISO does it begin and how intense is it at each ISO? Picture samples welcome. About what percentage of your normal pictures made with the M8 show this band?

 

4. If the band only appears when you specifically test for it (alternate workflows, turning off default noise filters in RAW conversion, boosting EV levels or brightness, etc.), at what ISO does it begin and how intense is it at each ISO? Do you need to "torture" the file a bit to see it? What must you do to the file in order for this to be apparent?

 

Cheers,

 

Sean

Link to post
Share on other sites

Mine is from the second batch (early december) FW 1.09 and should be the "fixed new" from factory version...

 

Shows vertical band from 640-2500 ISO in almost every shot with "in front of the camera" light sources, images with the light source from the back of the camera never has this issue/problem/defect...

 

But if i push it very hard, let's say 4+ stops in raw conversion, i can see the vertical banding even in 160 and 320 ISO shot, under the same light conditions mentioned above..

 

http://www.alexandertufte.com/stripe3.jpg

 

(I was told by email from Leica, that they are working on a possible fix for this in firmware 1.10 - but could not garanty that this was "fixable" by firmware yet, the hopeless situation here is really the supposed release of this firmware upgrade in feb 2007..???? Come on Leica, why in gods name do we have to wait 2 freakin months for a firmware fix...

 

Here's the answer from Leica:

 

"A new firmware 1.10 will be ready in February and can be downloaded on our homepage then. As for the "new fault", there is only a few cameras which showed this problem so far. Not everyone who owns the M8 has made this experience. Once the cameras are here in Solms for the correction, we also check this matter. I don´t know if the stripes will be gone with the new firmware but with the information we have with the DMR for the Leica R8/R9 we had this problem too and it was gone with the new firmware. So, this is all I can say at the moment."

Link to post
Share on other sites

Hi Sean,

 

I put some examples of the issue I had last evening on the thread "Strange image taken today - another bug?" Note, I have been trying to reproduce the phenomenon today but have not been able to. So that means it only occured once (in about 100-150 images). It was interesting that the problem appeared in the DNG and JPG file, plus when the image is reviewed 'in camera' the split screen occurs about 3-4 seconds after the image is displayed on the camera LCD. i.e. the image is displayed then 'clicks' into a left hand/right hand adjusted image.

 

Phil

Link to post
Share on other sites

Second batch, Firmware 1.09

 

I can't see any vertical band, but some horizontal band when pushing the stop 2+ at ISO 1250 and 2500. I looks like the vertical ones but are horizontals and turns out darker than the rest of the pictures :confused: I do remember the verticals one to be lighter

Link to post
Share on other sites

Dear Sean

 

First batch, FW 1.06

not eveident in normal workflow

 

I can see it in over 640and over and short exposure (forced) 'cap on' tests' -- a set up that I usually try to avoid as post processing to recover an image thru a cap is tricky LOL

 

Would it show up? Maybe, but in attempts to get it to show with dark backgrounds, I have been unsuccessful. So I assume it is there, but subtle

 

regards

Link to post
Share on other sites

1. 1st Gen, 1.06

 

2. Yes.

 

3. I took about 300 pictures at a xmas party this week - only 12 @ 2500. Out of those 12, only 4 show the banding, only with high overexposure in C1. And no, out of the majority of pictures taken @ 640, I haven't found one with that band yet.

 

Dirk.

Link to post
Share on other sites

1. Have you seen the vertical band (discussed elsewhere on this forum) in your particular M8?

Yes, but only after have tested to induce it.

 

2. If yes, does it appear in properly-exposed pictures using your normal workflow? If so, at what ISO does it begin and how intense is it at each ISO? Picture samples welcome. About what percentage of your normal pictures made with the M8 show this band?

Very seldom.

640 ISO moderate, 1250 ISO and 2500 ISO more visible.

Maybe less than 1%

 

3. If the band only appears when you specifically test for it (alternate workflows, turning off default noise filters in RAW conversion, boosting EV levels or brightness, etc.), at what ISO does it begin and how intense is it at each ISO? Do you need to "torture" the file a bit to see it?
Correctly exposed shots and use of IR-cut filters make band less visibile.

 

What must you do to the file in order for this to be apparent?

Artificial light, incandescent more than halogen from my experience.

The subject should not receive any illumination from the front (= back of the photographer) and the light source needs to be relatively near the area of band. For instance if I position the light source on the left of the frame I can't see the band, which (when it shows) is always in the right half of the frame.

 

Cheers :)

Link to post
Share on other sites

First generation, 1.06 firmware.

 

I found one vertical band in a very underexposed (about two stops) 2500 ISO shot, 1/91 second shutter. Same light, same f-stop, 1/16 second (proper exposure) shot a few seconds later shows no band, even with pushed exposure.

 

Otherwise, I can't vertical banding in normal shooting at any ISO, DNG or jpg, even trying insanely pushed exposure in ACR for a couple of other poorly exposed 2500 shots.

 

If my camera stays like this, it is a complete non-issue for me.

 

Clyde Rogers

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...