Occasional unreadable dng

[Guest]

Advertisement (gone after registration)

10 hours ago, jaapv said:

Leicas design is teamwork. There are separate groups per model. Leica has an extensive mutual technical cooperation contract with Panasonic, announced a few years ago to replace the previous more restricted one. 
I still think that the problems arise from the ambition to shoehorn 2023 technology into a 1954 camera design. It should be noted there are very few such problems with cameras like the X and TL CL series nor with the SL and Q, no more than the large brands 

Based on Leica’s statements e.g., regarding ‘why no IBIS’, this is one case where they have said that space constraints were getting in the way.

From an ‘electronic’ system complexity perspective, the M11 is no more complex than other cameras. The rangefinder mechanism is purely mechanical and does not affect any digital system (other than taking up space in the chassis). The same is true for the M mount - it’s just a circular hole in the box ...

Ergo: The digital nervous system of the M11, which is letting us down with data corruption issues, does not have to perform any miracles that other cameras don’t have to do as well. The problem is one of software, not hardware. I have been a software engineer/hacker for more than 40 years; if you consider what we were able to do with the 6502 CPU in the C-64 in 1983 (<1 MHz and 2 arithmetic registers …), you would not ever take the Maestro architecture as an excuse for software failures - NASA has sent people to space with a fraction of that hardware …

I am quite sure that the root cause of issues is with Leica’s tradition of excellence in mechanical engineering. The software people are not first-class citizens, compared to Demi gods like Peter Karbe . They don’t get the budget and attention of the mechanical & electrical engineering design teams. Disclaimer: I don’t have surveillance equipment installed in Wetzlar, my deductions are based on the well-established pattern of ‘the innovator’s dilemma’ in missing new disruptive technologies, especially if the legacy cash cow continues to overperform in the short run. This is almost a repeat of Leica’s ‘no digital, film only’ mistake, driven by the same proud traditions that almost ruined the company the first time around? Not a surprise that the M team is the worst offender, as they are widely regarded as the most orthodox traditionalists even as part of their mission statement. Proud luddites, if one were to be unkind. Now imagine you are the poor guy on that team who has to write firmware for the M … Your colleagues will treat you as the black sheep in the family, almost a heretic; they don’t even speak your language = if you tell them that the M11 launch should be postponed (theoretically), how likely would they to understand of support your arguments?

Another point in case: The whole hype about the 'triple resolution' sensor, which amounted to nothing more than smaller file sizes (available in other brands for 10+ years ...) - clearly the thinking of a team that does not deeply understand or greatly care about software (e.g. not even considering that size reduction in post-processing allows for arbitrary reductions, at higher quality - rendering the in-camera feature trivial)?

Let's not forget that file corruption exists in multiple ways with firmware 2.0.2., if you include the over-exposure issues as well as the bit corruption. At this point, I'm sure Leica would fix it if they could. But that would require a world-class digital engineering capability. /end-of-rant

Edited December 22, 2023 by mzbe

[Advertisement]

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

[pgk]

5 hours ago, jaapv said:

Of course. But given the character of an M camera, all this electronic frippery is only crammed in to attract new customers. Much better to offer it in a suitable package.

The M11 is living proof: Built to the limit (or maybe just beyond?) of the possibilities of an RF camera on a vintage fundament, it has become temperamental. The other proof is the A7: built as a holistic design from the ground up, it is rock-steady.

To me the M11 seemed a move in an odd direction. Perhaps you are right, although I see no fundamental reason why an RF camera is inherently limiting of those electonic requirements to which it is most suited.

[UliWer]

vor 58 Minuten schrieb mzbe:

I am quite sure that the root cause of issues is with Leica’s tradition of excellence in mechanical engineering. The software people are not first class citizens, compared to Demi gods like Peter Karen. They don’t get the budget and attention of the mechanical design teams. Disclaimer: I don’t have surveillance equipment installed in Wetzlar, my deductions are based on the well established pattern of ‘the innovator’s dilemma’ in missing new disruptive technologies, especially if the legacy cash cow continues to over perform in the short run. This is almost a repeat of Leica’s ‘no digital, film only’ mistake, driven by the same proud traditions that almost ruined the company the first time around? Not a surprise that the M team is the worst offender, as they are widely regarded as the most orthodox traditionalists even as part of their mission statement. Proud luddites, if one were to be unkind. Now imagine you are the poor guy on that team who has to write firmware for the M … Your colleagues will treat you as the black sheep in the family, almost a heretic; they don’t even speak your language = if you tell them that the M11 launch should be postponed (theoretically), how likely would they to understand of support your arguments?

I don't think this reflects the approach to a new design for the M justly. A digital M has not many mechanical parts anymore. It's the rangefinder, which has seen some fine tweaks for the M10 but nothing new for the M11. Then there is the shutter which is not built by Leica but by Copal. For a new model of an M like the M11 you won't see many people from mechanical engineering occupied (perhaps someone who said that the bottom plate was not necessary anymore...).  

The M11 shares the Maestro III processor with the SL 2 (the Q 3 which came later than the M11 has a Maestro IV). If it is true, that the SL doesn't show the same glitches as the M11 - if this is true and not only the result of SL users not telling - we cannot assume that the processor is not sufficient, as it is much more occupied in an SL (autofocus, IBIS, video etc). 

So what's in a digital M which is not in another digital camera by Leica which may cause the glitches? When we look back at the first reports after the M11 was new there was a lot about lens detection showing inconsistent results. Of course an SL also uses lens detection but for different purposes and it's not applied the same way as for the M (6-bit code read by an LED sensor). Originally lens detection for the M was mainly important to correct some vignetting and color shifts  - so it was applied after the shot. Since then it has seen many additional functions like detection if there is a lens attached at all and recently closing the shutter when the lens is detached - so it is applied well before the shot in a way it was originally not meant to work. This does not look as if mechanical engineering gnomes stood in the way but much more like people of "new disruptive technologies" having an idea: we could use lens detection for this and for that and also for a third purpose. But it's no error-free technology. 

If I had an M11 and had the unreadable DNG files my first idea would be: switch lens detection off and see if these glitches happen again. 

 

[Guest]

8 minutes ago, UliWer said:

I don't think this reflects the approach to a new design for the M justly. A digital M has not many mechanical parts anymore. It's the rangefinder, which has seen some fine tweaks for the M10 but nothing new for the M11. Then there is the shutter which is not built by Leica but by Copal. For a new model of an M like the M11 you won't see many people from mechanical engineering occupied (perhaps someone who said that the bottom plate was not necessary anymore...).  

The M11 shares the Maestro III processor with the SL 2 (the Q 3 which came later than the M11 has a Maestro IV). If it is true, that the SL doesn't show the same glitches as the M11 - if this is true and not only the result of SL users not telling - we cannot assume that the processor is not sufficient, as it is much more occupied in an SL (autofocus, IBIS, video etc). 

So what's in a digital M which is not in another digital camera by Leica which may cause the glitches? When we look back at the first reports after the M11 was new there was a lot about lens detection showing inconsistent results. Of course an SL also uses lens detection but for different purposes and it's not applied the same way as for the M (6-bit code read by an LED sensor). Originally lens detection for the M was mainly important to correct some vignetting and color shifts  - so it was applied after the shot. Since then it has seen many additional functions like detection if there is a lens attached at all and recently closing the shutter when the lens is detached - so it is applied well before the shot in a way it was originally not meant to work. This does not look as if mechanical engineering gnomes stood in the way but much more like people of "new disruptive technologies" having an idea: we could use lens detection for this and for that and also for a third purpose. But it's no error-free technology. 

If I had an M11 and had the unreadable DNG files my first idea would be: switch lens detection off and see if these glitches happen again. 

 

The critical glitches we currently see are likely due to timing, concurrency, or overflow issues (or a combination thereof). Perhaps they 'lifted and shifted' code from the SL2 with different latency assumptions (different components) that should have been changed with the M11? Or added code in the wrong places e.g. for the 'triple res' nonsense ... Or the 60MP data volume, at identical data bandwidth, only works 'most of the time'?

What would help: Leica should come clean about the fiasco. Their cryptic release notes don't do the job. I can imagine their lawyers do not permit transparency, but the current attitude is hurting Leica's most precious asset - the perceived integrity of the brand (... together with non-software production defects, wait time for repairs, ...).

[hjddd]

3 hours ago, UliWer said:

我认为这并不能公正地反映 M 新设计的方法。数字M不再有太多的机械部件。测距仪对 M10 进行了一些细微的调整，但对 M11 来说并没有什么新内容。然后是快门，它不是由徕卡制造的，而是由 Copal 制造的。对于像 M11 这样的新 M 型号，您不会看到很多机械工程人员忙碌（也许有人说底板不再需要了......）。  

M11 与 SL 2 共享 Maestro III 处理器（晚于 M11 的 Q 3 配备 Maestro IV）。如果这是真的，那么 SL 不会出现与 M11 相同的故障 - 如果这是真的，并且不仅仅是 SL 用户没有告诉的结果 - 我们不能假设处理器不够用，因为它被占用了更多在 SL 中（自动对焦、IBIS、视频等）。 

那么，数码M中有什么是徕卡其他数码相机中没有的，可能会导致故障呢？当我们回顾 M11 上市后的第一份报告时，有很多关于镜头检测的结果不一致。当然，SL 也使用镜头检测，但用途不同，并且其应用方式与 M（由 LED 传感器读取的 6 位代码）不同。最初，M 的镜头检测主要是为了纠正一些暗角和色偏 - 因此它是在拍摄后应用的。从那时起，它出现了许多附加功能，例如检测是否安装了镜头，以及最近在镜头分离时关闭快门 - 因此它在拍摄之前就以一种原本不适合的方式应用。这看起来并不像机械工程侏儒挡道，而更像是“新颠覆性技术”的人们有一个想法：我们可以使用镜头检测来实现这个、那个以及第三个目的。但这并不是没有错误的技术。 

如果我有一台 M11 并且有不可读的 DNG 文件，我的第一个想法是：关闭镜头检测，看看这些故障是否再次发生。 

 

The len detection can not be turned off ， every time I put it to NO， once restart the camera， the LD automaticly turns on。 

[lct]

31 minutes ago, hjddd said:

The len detection can not be turned off ， every time I put it to NO， once restart the camera， the LD automaticly turns on。 

You may wish to try with an uncoded lens.

[jaapv]

Advertisement (gone after registration)

The camera starts up to the last user profile and reverts to the standard profile if there are any changes. So if you want lens detection off on startup you will have to make a user profile that contains this menu setting. 

[lct]

4 minutes ago, jaapv said:

The camera starts up to the last user profile and reverts to the standard profile if there are any changes. So if you want lens detection off on startup you will have to make a user profile that contains this menu setting. 

This is not necessary with uncoded lenses. Suffice it to select lens detection off and the camera will keep that setting after a startup. With coded lenses, whichever user profile is activated, the camera will start up with the coded lens profile to which it gives priority. It is the way my M11 works, or seems to work anyway. I say "seems" as a modesty but also because i begin to wonder if i understand anything when i read on this good forum that some M11 copies have firmware issues that others have not in spite of all of them having the same firmware 2.0.2...

[UliWer]

Well, I have said this before: If lens detection „works“ by recognizing a coded lens even if it was turned off I‘d look very critically at this function as a possible source of camera glitches. Off should mean off and not „on with coded lenses“. With  the M10 and the M10 Monochrom „off“ is really off and I see no problem leaving it this way. 

[pgk]

9 hours ago, jaapv said:

The camera starts up to the last user profile and reverts to the standard profile if there are any changes. So if you want lens detection off on startup you will have to make a user profile that contains this menu setting. 

This seems to be a counter intuitive and cackhanded way of doing things to me!

[lct]

14 minutes ago, UliWer said:

Well, I have said this before: If lens detection „works“ by recognizing a coded lens even if it was turned off I‘d look very critically at this function as a possible source of camera glitches. Off should mean off and not „on with coded lenses“. With  the M10 and the M10 Monochrom „off“ is really off and I see no problem leaving it this way. 

Matter of individual prefences i guess. My camera having no glitch watsoever, i don't tend to complain and I like the fact that the M11 prioritizes lens coding, as I use several lenses and i am a rather (my wife says hopeless) distracted guy.

[Derbyshire Man]

6 hours ago, lct said:

Matter of individual prefences i guess. My camera having no glitch watsoever, i don't tend to complain and I like the fact that the M11 prioritizes lens coding, as I use several lenses and i am a rather (my wife says hopeless) distracted guy.

Do you have any settings turned off like Fotos communication?

[lct]

5 minutes ago, Derbyshire Man said:

Do you have any settings turned off like Fotos communication?

I don't use Fotos at all if it may help but i did not set Connectivity to off.

Edited December 23, 2023 by lct

[Derbyshire Man]

56 minutes ago, lct said:

I don't use Fotos at all if it may help but i did not set Connectivity to off.

Have you connected to your phone Fotos? Or never done so? Is it running on your phone even though you don’t use it (eg for GPS)?

Edited December 23, 2023 by Derbyshire Man

[lct]

5 minutes ago, Derbyshire Man said:

Have you connected to your phone Fotos? Or never done so? Is it running on your phone even though you don’t use it (eg for GPS)?

Never used Fotos in any way sorry. 

[ianforber]

41 minutes ago, Derbyshire Man said:

Have you connected to your phone Fotos? Or never done so? Is it running on your phone even though you don’t use it (eg for GPS)?

I use FOTOs, have lens detection set to auto and also have never had any problems with my M11. My natural pace of work is slow though so I rarely press any button quickly after doing something else, which may help explain not having any problems. 

[UliWer]

vor 8 Stunden schrieb lct:

I like the fact that the M11 prioritizes lens coding, as I use several lenses and …

Same with me. The M10 prioritizes the detection of coded lenses when you have chosen a manual detection for an uncoded lens. As soon as you swap from your coded lens to an uncoded one your last manual choice will be used. Perfect.

Though when I chose to switch lens detection off for whatever reason as it is my own choice, I prefer the camera not counteracting this choice without asking me. Off means off and not „off as long as no coded lens is applied“. 

Edited December 23, 2023 by UliWer

[lct]

4 minutes ago, UliWer said:

Same with me. The M10 prioritizes the detection of coded lenses when you have chosen a manual detection for an uncoded lens. As soon as you swap from your coded lens to an uncoded one your last manual choice will be used. Perfect.

Though when I chose to switch lens detection off for whatever reason as it is my own choice, I prefer the camera not counteracting this choice without asking me. Off means off and not „off as long as no coded lens is applied“. 

It is one of the reasons why i prefer the M11 personally. Off still means off with coded lenses on the M11, but it does so as long as the camera is on only. As soon as the camera is switched off and on, it will take into account the code of the mounted lens as a priority. This way, one can use a different lens that will be recognized automatically, instead of having to juggle with a menu and miss the decisive moment this way. Setting lens detection off is still possible afterwards when the moment is over.

[Advertisement]

Hallo,

Look here: Interesting Leica gear

[UliWer]

vor 39 Minuten schrieb lct:

Off still means off with coded lenses on the M11, but it does so as long as the camera is on only. As soon as the camera is switched off and on, it will take into account the code of the mounted lens as a priority.

I know we discussed this before and I am not looking forward to convince you.

For me there is a fundamental contradiction if something is switched off and stays active at the same time so that it can detect a lens code to apply a certain code. The camera does not accept the main function "off" for lens detection.

It may be interesting lo look in which context the word "freeze" was first used in connection to the M11. I may have overlooked an early posting but I think it is here:

https://www.l-camera-forum.com/topic/329432-m11-will-freeze-when-lens-detection-off-bug/

And then it goes on:

https://www.l-camera-forum.com/topic/329232-m11-bug-with-uncoded-lens/

https://www.l-camera-forum.com/topic/338594-lens-selector-off/

https://www.l-camera-forum.com/topic/331172-m11-does-not-detect-lens-intermittent-fault/

https://www.l-camera-forum.com/topic/375926-m11-auto-lens-detection-problems/#comments

Edited December 23, 2023 by UliWer

[Crem]

On 12/22/2023 at 2:03 PM, mzbe said:

The critical glitches we currently see are likely due to timing, concurrency, or overflow issues (or a combination thereof). Perhaps they 'lifted and shifted' code from the SL2 with different latency assumptions (different components) that should have been changed with the M11? Or added code in the wrong places e.g. for the 'triple res' nonsense ... Or the 60MP data volume, at identical data bandwidth, only works 'most of the time'?

What would help: Leica should come clean about the fiasco. Their cryptic release notes don't do the job. I can imagine their lawyers do not permit transparency, but the current attitude is hurting Leica's most precious asset - the perceived integrity of the brand (... together with non-software production defects, wait time for repairs, ...).

As a software engineer, my M11 experience has felt like Leica releases M11 firmware on management's schedule. I find it impossible to believe the testing team used v2 for weeks or months and somehow didn't run into random red frame lines and corrupt DNGs. It's way more likely that v2 didn't get much real world testing.

I agree that Leica does a poor job of coming clean and communicating. I contacted support, reported bugs, and asked when they would get fixed. I got told that the engineers don't tell the support staff until the firmware is released. That's simply unacceptable. Engineers + support should be working close together on this.

Leica really screws up by not permitting firmware downgrades. I think we are all fine with beta quality firmware so long as we can downgrade to the previous good version.