16 consecutive frames damaged

[neelin]

Can anyone explain how not to have this happen?

I was taking marathon race shots this morning [Manitoba Marathon]. I took about 750 shots on 2 4-gig SD cards. This happened on the first card probably in about 150 shots, and ruined 16 consecutive frames. There was no anomolies in camera operation that I could determine the files were not being written properly.

Temperature was about 11 C, dry, and the camera was turned on-off many times during the morning.

All other shots were normal, aside for some Autofocus errors
If you want focus practice I highly recommend taking up a couple of positions at a marathon & focus away!

Robert

[PeterP]

First question is: what 4 gig memory cards were you using ?? According to Leica there are only 2 brands recommended Hama & extrememory.

Peter

[ho_co]

Robert--
I've never seen anything like this.

Were all the frames just like this? Left half okay, right half this same funny color, separation right down the center of the image? Or was there some variation?

The fact that the division seems to be exactly in the center and that the magenta seems so even makes it look to me like a sensor or circuit problem.

My suggestion is to call your Leica distribution center and tell them what's happening.


It may be the card as Peter said, but I haven't had problems with the Transcend 150x 4G cards (not on the list). But the fact that I haven't had problems yet doesn't mean I won't. As far as that goes, you could get a bad sample even of an approved card. So you could check: Does this happen on more than one card?

Originally Posted by PeterP

what 4 gig memory cards were you using ?? According to Leica there are only 2 brands recommended Hama & extrememory.

Actually, the count is now up to 3 brands: those two plus Platinum.

[neelin]

Originally Posted by ho_co

Robert--
I've never seen anything like this.

Were all the frames just like this? Left half okay, right half this same funny color, separation right down the center of the image? Or was there some variation?

>exactly the same, right down the center & exactly the same colour (on the right side of course).

The fact that the division seems to be exactly in the center and that the magenta seems so even makes it look to me like a sensor or circuit problem.

My suggestion is to call your Leica distribution center and tell them what's happening.


It may be the card as Peter said, but I haven't had problems with the Transcend 150x 4G cards (not on the list). But the fact that I haven't had problems yet doesn't mean I won't. As far as that goes, you could get a bad sample even of an approved card. So you could check: Does this happen on more than one card?

>I've shot thousands of images on my 2 transcend 150x 4-gig cards and this is the first time that one of them has hiccoughed. Unfortunately I've reformatted the cards so I don't know which one it was.

>perhaps I'll look for some software that can "exercise" the cards to see if one is developing errors?

[andybarton]

M8 sensors are split into two circuits. There were other reports of this happening, especially from older cameras.

Try searching here for reports

[luigi bertolotti]

Did you see the effect on the LCD screen just after the shots, or did it apeear only after having downloaded the files from the card ? In case you didn't look at the LCD while shooting, are the files already on the card so that you can make a reread on the LCD ?

I ask this because, as Andy says, the exact division in half makes it think of the "two-halves" sensor of M8... in this case I think you had to see the failure on the LCD too; wouldn't it be so, some problem of transferring/recording data in the memory card could instead be blamed..

[waterlenz]

I had something similar happen but it was not a split down the middle like yours. My recollection is that this did not appear on the M8 screen. Up-loading again did not help. This pic was from back in Dec 07. Only this shot was affected. The rest on the card were OK. I assumed the magenta part was from a previous IR shot but after looking yours maybe not. The snow-topped mountain appears to be an earlier shot from Alaska whereas the main shot from December shows Lake Erie, 1000's of miles away. I was using a 2 GB card (either a Lexar 133x or Sandisk Extreme III).
Tom

[pgk]

Problems like this have to be looked at logically. I'd say that you have 4 possible problems: camera, card, card reader (and just possibly, the computer).

First, I'd suggest trying to re-view the images on the card in the camera. If they view ok then it is probably a problem with extracting the files from your card onto the computer - I've had this problem myself - it showed in a similar way in that part of the image was corrupted - and it turned out to be some sort of problem reading the file structure of the card by the card reader (but not always - it showed only when I shot more than 1Gb on a 2Gb card and appears to have been related to the way the larger card was read by the card reader).

Next see if you can use another card reader to transfer the images (as I said, I've had probems with readers despite the fact that they should be ok to read any SD cards!)

If this doesn't work then it might be worth trying an image rescue piece of software to see if it can extract the images - if it can, then again it is likely to be the card or reader.

If the images show corruption when viewed via on the camera then it may still be the card or possibly now the camera. If it is an intermittent fault on either then the only way to eliminate one is to stop using the card and see if you get a repeat of the problem on another card at some point.

[Symeon]

It happens, especially when using the not-so dependable Continuous Shooting setting on the Leica dial. I had similar results once with totally blank frames besides the 2-color ones, but it is very rare. Somehow, it seems, the buffer overloads and the image processor cannot cope with writing data properly on the SD card, whatever size that is. But I wouldn't use 4Gb cards, I'd stick to 1-2Gb cards and carry more on me.
Regards,

[ho_co]

Your suggestions all seem good, guys, but to the point: Robert has already reformatted the cards, so he can't review these images in camera or check his card reader.

Robert, I'll stick with what I said above: Check with Leica.

Andy recommended searching the forum for similar occurrences, and that's the first best idea; but IIRC, similar problems ended with "I called Leica and they said to send in the camera."

pgk's approach is methodical and reasonable. If at the moment you have the time and inclination, keep shooting. Don't erase the card till you've checked that you've got good images downloaded. See whether it's just one card. (Have you changed card-readers recently?) (You do erase the card in the camera, right?)

Tom's example is fascinating because he has three images superposed. But since your situation has only one image, and divided directly in half, and occurred multiple times, despite the color similarity with Tom's shot, I think this is different.

In regard to 'testing' the card, that might be a good idea, but probably not much better than in-camera shooting.

Symeon's suggestion is a fresh one, but I believe the blank-frames issue was cured with the current 1.201 firmware, so the problem he mentions may be different.

Luigi, Andy and I are all bothered by the right-down-the middle effect. Look for more posts on the topic. As they said, there were a number of peculiarities with the earliest cameras. Do you know for sure that this body went back to Solms for the big recall, or that it was produced after the recall?

[luigi bertolotti]

Thinking again, and after having seen the strange Thomas' case... I am prone to think the problem is NOT in the sensor: borders are superstraight in both cases (and, obviously, have nothing to do with the two halves of the sensor in Thomas' case) and my idea is that if the problem would have been in some arrays of sensor 's pixels, the in-camera processing shouldn't produce so straight "jumps". I'd say that problem has to do with transfer of very well-defined strings of bytes... in some of the byte transfers / byte write that occur :
a) from camera buffer to card controller (this could be surely in Robert's marathon shot, probably not for Thomas, where seem to understand that the 3 pics were taken in very different moments, surely with camera off within).
b) from card controller to card (same above consideration applies)
c) inside the card itself (can happen... time ago I experienced such a problem with a 64 MB card of a mobile phone... wrong positons of agenda's data, and some of them "fool")
d) from card reader to PC (card reader "reads" wrong memory positions and/or consistently ordered wrong bytes in some memory positions - transfers to PC a fooled file)

I make the above speculations IMAGING how could be done the logical-blocks architecture of the system. (quickly...it was MANY years ago that I used to design such logical blocks schemes... ).. I do not know if, for M8, this design has been published... Andy, do you have something about it?

[jaapv]

It is not that, Luigi. The sensor is divided into two exact halves for purposes of readout. Should an error occur in one of the readout/amplifying cicuits it would produce just such a result. See the "centrefold" issue on some MF digital backs, where the amplification is dissimilar left and right.
One could postulate that in this case only the magenta channel was produced. This suggests an error after the interpolation of the Bayer pattern, but before further processing of the image.

Off to Solms it is, I guess...

[luigi bertolotti]

Originally Posted by jaapv

It is not that, Luigi. The sensor is divided into two exact halves for purposes of readout. Should an error occur in one of the readout/amplifying cicuits it would produce just such a result. See the "centrefold" issue on some MF digital backs, where the amplification is dissimilar left and right.
One could postulate that in this case only the magenta channel was produced. This suggests an error after the interpolation of the Bayer pattern, but before further processing of the image.

Off to Solms it is, I guess...

Jaap, that was my initial thought too... the "exact halves" make instantly think of the "halves" of the sensor, but it was the Thomas' example, so similar but with different areas, that made me speculate a different issue (and common to both... but nothing proves it's so)

[jaapv]

Luigi, I think that Thomas' example, despite looking very similar is a multiple-cause problem, the first being a repeated write-error to the card, and secondly one of the underlying images being produced in magenta only, that possibly being the same thing that happened to the righthand half of Robert's image. I hope I produced my thoughts understandably...

[marknorton]

I think this is clearly a sensor readout issue. As Jaap and others have said, the sensor has two readout channels; one of them is working correctly, the other is not.

Data is read out by "clocking" the sensor - an alternately high/low signal causes the sensor to present the output of the next pixel - all the DSP sees is a series of voltage levels which it interprets as red/green/blue light levels according to the order in which it sees them. It's possible the sensor has failed to respond to a particular clock edge so that the data is being shifted out in other than the correct sequence. Alternately, the critical power supply voltages to the sensor could be out of spec.

Either way, if it happens again, I'd send it back to Leica.

[jaapv]

Which, Mark, suggests a motherboard failure ?

[wparsonsgisnet]

Neelin, are your shots from raw images or jpegs?

I'm wondering if it's just a sensor problem, or if the jpeg logic is also involved.

Also, were your batteries fresh?

[marknorton]

Originally Posted by jaapv

Which, Mark, suggests a motherboard failure ?

It could be the power supply circuits on the motherboard, it could be an intermittent connection between the boards or it could be the sensor or circuitry on the sensor board. Really tough to know where the problem lies unless they have seen it before (which I expect they did during development). It might be, for example, one of the supply rails the sensor uses is on the margin so that one side of the sensor is happy, the other is not, some of the time.

Tough to diagnose, especially if the problem cannot be reproduced at will.

I don't think it's anything to do with the memory card because a memory card failure would lead to a file checksum error and, in any case, any corruption would not lead to such a coherent visible effect.

[biglouis]

Originally Posted by waterlenz

I had something similar happen but it was not a split down the middle like yours. My recollection is that this did not appear on the M8 screen. Up-loading again did not help. This pic was from back in Dec 07. Only this shot was affected. The rest on the card were OK. I assumed the magenta part was from a previous IR shot but after looking yours maybe not. The snow-topped mountain appears to be an earlier shot from Alaska whereas the main shot from December shows Lake Erie, 1000's of miles away. I was using a 2 GB card (either a Lexar 133x or Sandisk Extreme III).
Tom

I've had the same problem as Tom. From memory (a) the battery was low and (b) I was using a pretty old 2GB card, one which I'd be using continuously in about 3 different digital cameras for about 2 years. The clue was the inclusion in the frame of partial captures from days or weeks before. As it was not reproducible, I treated myself to a brand new Kingston Ultimate 2GB card and haven't had the problem since.

Despite it's retro carcass the M8 is at its core a computer and there are many potential problems in the path from sensor to card but mostly the weakest point.... is the card.

LouisB

[jimmy pro]

Originally Posted by neelin

Can anyone explain how not to have this happen?

Some people might say, get a Canon or Nikon. I can honestly say I've never seen, heard, or read of another camera doing what your M8 (and, it sounds like others to) is. Does anybody know if Leica hired a bunch of guys they fired from Jaguar back in the 80's? The M8 sounds just like the one I had. Great car when it was working, lots of potential, but you needed a second one for as often as it was in the shop. Man I feel like one lucky sonofabitch that my M8 is still working ok.