Bug in M10 6-bit coding

[Matlock]

Advertisement (gone after registration)

30 minutes ago, verbivore said:

The entire 6-bit coding system is a charming but anachronistic kluge. (Even providing a 7th bit would have doubled the number of codable lenses from an inadequate 64 to a reasonable 128.) If that is Leica's concession to the 21st century, they should at least implement it in a way that's useful to their loyal and patient customers. 

Oh so that's where I have been going wrong, I haven't quite reached 128 lens yet, must try harder. 😁

[Advertisement]

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

[lct]

2 hours ago, verbivore said:

To LCT: this would be straightforward. The photographer would manually enter an M lens in the Lens Detection  > Manual M menu item, and whenever the camera detects that an uncoded lens is mounted, it would use that metadata. The photographer would manually enter an R lens in the Lens Detection > Manual R menu item, and whenever the camera detects that the R-Adapter is mounted (through its unique 6-bit code), it would use that metadata. Since the photographer would never have a reason to manually enter an M or R lens when a coded M lens is mounted, this would handle all the cases but one: the tiny number of photographers who use a Novoflex or Kipon R-to-M converter, which is uncoded. Leica could be forgiven for not supporting those items. 

With the Macro-Adapter-M, the firmware change would have to be a smidgen more complicated, but it's still pretty simple. When the Macro-Adapter is mounted (detectable through its unique 6-bit-code), the photographer could enter a second M lens, which would be applied to the image when, and only when, the Adapter is mounted. Granted, this would require that the camera remember TWO M lenses (one for uncoded lenses, one for the Macro-Adapter). This would require that a high-school-level programmer spend 15  minutes, rather 5 minutes, patching the firmware code. [...]

But why would i have to enter anything if i use the same lens on my adapter? And this each time i switch on the camera? I simply don't want that. Seems like you're reverting the onus of chimping here. Just kidding .

[mujk]

1 hour ago, verbivore said:

I agree, Mujk. This would be even more useful if there was a manufacturer who sold L39-to-M adapters with 6-bit coding pits (the currently available ones are either shoddily made, and don't provide proper infinity focus, or lack precisely milled pits), and/or replacement rear flanges with the pits, which could be swapped onto the third-party lenses. Sharpies could then be used to enter 6-bit codes, ideally one that was not already allocated to a Leica lens but was user-settable. But for now I'd be happy enough if Leica simply fixed the bug. 

The entire 6-bit coding system is a charming but anachronistic kluge. (Even providing a 7th bit would have doubled the number of codable lenses from an inadequate 64 to a reasonable 128.) If that is Leica's concession to the 21st century, they should at least implement it in a way that's useful to their loyal and patient customers. 

Yes, Leica could have used more bits and reserved some values for Leica use and left a number of values for customer and/or third party use. Probably they were in a hurry and just wanted to make sure that coding would work reliably for their own future, current and some older M lenses. So they made the pits fairly large and left space between them to ensure correct reading in all situations.

They could also have put in a small sensor on the frameline selector and made it possible to use that information to indicate a user-configurable uncoded lens or even to check if the 6 bit code had been read correctly (a sort of primitive parity bit).

[helohe]

1 hour ago, verbivore said:

I agree, Mujk. This would be even more useful if there was a manufacturer who sold L39-to-M adapters with 6-bit coding pits (the currently available ones are either shoddily made, and don't provide proper infinity focus, or lack precisely milled pits), and/or replacement rear flanges with the pits, which could be swapped onto the third-party lenses. Sharpies could then be used to enter 6-bit codes, ideally one that was not already allocated to a Leica lens but was user-settable. But for now I'd be happy enough if Leica simply fixed the bug. 

The entire 6-bit coding system is a charming but anachronistic kluge. (Even providing a 7th bit would have doubled the number of codable lenses from an inadequate 64 to a reasonable 128.) If that is Leica's concession to the 21st century, they should at least implement it in a way that's useful to their loyal and patient customers. 

As far as I know the camera also uses the information from the lens mount flange (the part that brings up the correct framelines). As there are 3 options for this, this allows a total of 192 lenses to be coded. 64 in each set of framelines.

[verbivore]

Matlock, the point isn't that any single user would have more than 64 lenses, but that different users  have different subsets, and the entire set numbers more than 64. Leica has made more than 64 lenses in its history, and will presumably continue to add new ones, to say nothing of the compatible Zeiss's, Voigtlanders, and so on. 

LCT, if you only had one uncoded M lens, or one uncoded R lens, then you would not need to re-enter it each time. The camera already remembers the last lens you manually entered, and this would not change if they fixed the bug.  

Helohe: Clever suggestion! Yes, Leica could take advantage of that feature, in theory. 

Pico: Yes, the Rayqual adapter is of excellent quality. Its only limitation is that the 6-bit coding area is a continuous flat surface, rather than 6 precisely positioned pits. Even using a Match Technical template (which unfortunately has some play), and black and white extra-fine Sharpies, I've been unable to inscribe marks that the M10 can read. 

 

 

[lct]

Just now, verbivore said:

LCT, if you only had one uncoded M lens, or one uncoded R lens, then you would not need to re-enter it each time. The camera already remembers the last lens you manually entered, and this would not change if they fixed the bug. 

Then i don't get it sorry. How would you manage to keep the last lens in memory if the camera registers “uncoded lens” as you suggested? 

[verbivore]

Advertisement (gone after registration)

LCT:

When the camera detects an uncoded lens while the camera was set to Lens Detection > Auto, it would default to the last M lens that was manually entered. That is, it would *detect* that the lens was uncoded, but would not enter "uncoded" as the lens metadata. 

If the photographer didn't want any code entered, he or she could switch to Lens Detection > Off, just like now. 

Another possibility is to add a "None" option in the list of manually coded M lenses, in case a photographer both wanted to leave Lens Detection  set to Auto AND had an uncoded lens for which he did not want to have any lens metadata stored. 

[pico]

Such trivial concerns. Who really needs auto lens detection? We are not stupid.

[lct]

7 minutes ago, verbivore said:

When the camera detects an uncoded lens while the camera was set to Lens Detection > Auto, it would default to the last M lens that was manually entered. [...]

The camera could not keep any lens profile in memory then given that the lens is not coded. Next time you switch on you would get the adapter's profile in your exif data, not that of the previous or any other lens IINM.

[pico]

Just another of my stupid questions - how very much does 6-bit coding contribute to image outcome? Really!

 

 

[verbivore]

No, that doesn't follow. The user manually sets an M lens, and the camera remembers it indefinitely, or until the user sets a different lens, just like now. It applies the lens info to the image when and only when it detects that an uncoded lens is mounted. If an adapter with a 6-bit code is mounted, it applies that code. This works in every case except an uncoded 3rd-party R adapter. 

[verbivore]

Pico: The 6-bit code matters for two reasons. One is that (particularly with a wide-angle lens) the image is altered in the camera according to the lens code (vignetting and distortion), unless the user remembers to undo that correction in post-processing. The other is that photographers perfect their skills, and learn how to optimize their gear, by looking at their results, including the choice of lenses. That's why every shot from a Canon, Sony, Nikon, Leica S and SL, etc. has lens metadata. Given that Leica engineers implemented a crude version of this for the M, why should they have an implementation that introduces errors, rather than doing it properly? 

[lct]

1 hour ago, verbivore said:

No, that doesn't follow. The user manually sets an M lens, and the camera remembers it indefinitely, or until the user sets a different lens, just like now. It applies the lens info to the image when and only when it detects that an uncoded lens is mounted. If an adapter with a 6-bit code is mounted, it applies that code. This works in every case except an uncoded 3rd-party R adapter. 

Seems like we'll have to agree to differ here. If you set lens detection to auto, as you suggested above, the camera won't remember anything but the profile it knows when you switch on, i.e. that of the adapter. 

Edited August 26, 2019 by lct

[jdlaing]

1 hour ago, pico said:

Just another of my stupid questions - how very much does 6-bit coding contribute to image outcome? Really!

 

 

In some cases quite a bit. In others...not.

[UliWer]

Am 24.8.2019 um 17:59 schrieb verbivore:

1. The R-adapter-M has a unique 6-bit code: 110111. When the camera’s “Lens Detection” is set to “Auto,” the camera should always register “R-Adapter-M,” or the last R-lens manually entered under the Lens Detection menu.

Instead, when an R-adapter-M with an R lens is mounted, it switches out of “Lens Detection Auto,” and registers the last lens that was manually entered. If that was an R lens, it switches to “Lens Detection Manual R,” which is OK. But if the userI had previously entered an M lens under “Lens Detection Manual M,” the camera switches to “Lens Detection Manual M,” and falsely registers that M lens.

This makes no sense. Since the R-Adapter-M has its own 6-bit code, when the camera when set in “Lens Detection Auto,” it should always register an R lens.

2. The Macro-Adapter-M also has a unique 6-bit code, 101000. So when the camera’s “Lens Detection” is set to “Auto,” it should register “Macro-Adapter-M.”

Instead, it registers “4/90.” ...

3. When the camera is set to “Lens Detection Auto,” and an uncoded lens is mounted, the camera registers the last lens that was manually coded, whether it was under “Lens Detection Manual M” or “Lens Detection Manual R.” It’s hard to understand the logic behind this.

I fully agree to you point 1 (with an addition I describe below under 2).

ad 2: I'd wish the Macro-Adapter would generally switch to the detection for the 1:4/90mm Makro-Elmar M. If your's really does, something must be wrong with the coding for my adapter, since it only shows the last lens which I manually chose from the list (same as with the R-adapter). The Macro-Elmar being the lens for which the adapter was orginally made and will be bought in most cases together with this lens, it would be the detection you'd expect when the adapter is coded. Of course it should be possible to change the detection manually, if you use any other lens together with the adapter, which is possible. So the logic should be: As long as you use "Auto", the Macro-Adapter detects the Macro-Elmar; but it should be possible to choose another prefered lens from the list if you wish, so that this lens is automaticalle detected when you attach the adapter. 

The same logic should apply to the R-adapter: you should be able to choose any R-lense which is generally been detected if you attach the adapter, and of course you should be able to change your preference.  

ad 3: There is a logic behind the switch to the last manually chosen lens when you use "Auto": If someone generally uses coded lenses with one exception (in my case 135mm which is not coded), he will just chose the uncoded lens manually and stay on "Auto".  All his lenses will be detected, without any need for changing of lens detection manually. Of course this does not help if you use more than one undetected lenses. There is no way for electronics to make a decision between two different options if these options are not detected (no code).

With your proposal you would always have to choose a certain lens from the menu, and have no chance that the detection works according your preferences. 

[UliWer]

vor 2 Stunden schrieb pico:

Just another of my stupid questions - how very much does 6-bit coding contribute to image outcome? Really!

 

 

I always had the impression that coding lenses and camera's lens detection was a lesson to become unhappy.

In general it might reduce vignetting. I have an old screw-mout 28mm Summaron which cannot be detected - I never missed the reduction of vignetting. For the M9 there were some more reasons to use lens detection for wide-angle lenses (colour shifts etc.). For the M10 I have never "detected"  any  substantial difference in this respect.

Rests Auo-ISO, which you can use focal-length dependant, so the camera always choses a shutter time of e.g. 2*focal-length. Maybe helpful sometimes.

The most important reason for the fetish of lens coding and detection is the EXIF...another instruction to become unhappy. 

 

[verbivore]

"With your proposal you would always have to choose a certain lens from the menu, and have no chance that the detection works according your preferences. "

No, not true. With a coded lens, the camera would recognize that lens. With an R-adapter, the camera would recognize that it was an R-adapter (unlike what it does now). With a macro-adapter, the camera would recognize that was a macro-adapter (unlike what it does now). With an uncoded lens, the camera would recognize that it was an uncoded lens. That is, the camera would register the lens correctly, instead of incorrectly. I'm surprised that so many participants in the forum are not seeing that the problem is a blatant bug in the M10 firmware -- and one that distorts the image (e.g., when it applies a wide-angle correction to an image that was not taken with a wide-angle lens). This has nothing to do with whether a "real photographer" should car about metadata, use a notepad, or anything else -- it's a bug. 

In addition to just getting this right instead of wrong, the camera should intelligently apply the information that it already collects and remembers, namely one manually coded M lens and one manually entered R lens. The M lens code should be applied when the camera has an uncoded M lens attached. The R lens code should be applied when the camera has an R adapter attached. This would not (and does not) require the user-re-entering the lens information every time -- the camera can (and does) remember. It's just a question of applying the information sensibly rather than irrationally. 

Some readers have suggested in addition that the camera remember two manually coded M-lenses, one for uncoded lenses, one for the macro adapter. I agree that this makes sense, but it is a request for Leica to add a new feature, which goes beyond what I'm suggesting. I'm suggesting only that Leica implement a feature it already has correctly, rather than incorrectly. 

And with all of these suggestions, the user always has the option of turning Lens Detection off, and would still have that option if Leica fixed the bug. 

I

[lct]

1 hour ago, UliWer said:

[...] I'd wish the Macro-Adapter would generally switch to the detection for the 1:4/90mm Makro-Elmar M. If your's really does, something must be wrong with the coding for my adapter, since it only shows the last lens which I manually chose from the list (same as with the R-adapter). The Macro-Elmar being the lens for which the adapter was orginally made and will be bought in most cases together with this lens, it would be the detection you'd expect when the adapter is coded. Of course it should be possible to change the detection manually, if you use any other lens together with the adapter, which is possible. So the logic should be: As long as you use "Auto", the Macro-Adapter detects the Macro-Elmar; but it should be possible to choose another prefered lens from the list if you wish, so that this lens is automaticalle detected when you attach the adapter.[...]

I use the macro adapter with various lenses so i prefer keeping the last lens profile personally. That's the way my CL is working (don't recall on my M240) when the lens profile is chosen manually and your M10 should to the same i guess but i have no experience with it. Now in auto mode the default lens profile is "Macro-Adapter-M" and the focal length registered in the exif data is 90mm. I see nothing to change in that but YMMV. 

[Advertisement]

Hallo,

Look here: Leica M

[lct]

36 minutes ago, verbivore said:

With a macro-adapter, the camera would recognize that was a macro-adapter (unlike what it does now).

Exactly what it (at least my CL) does in auto mode with the "Macro-Adapter-M" default profile. Hard to believe that the M10 does differently but i have no experience with it. 

39 minutes ago, verbivore said:

With an uncoded lens, the camera would recognize that it was an uncoded lens.

That's what it does with the "Macro-Adapter-M" default profile referred to above. The camera having no info about the current lens, 90mm is chosen as default focal length which is the better choice to me. But again it is in auto mode, not when a lens profile is chosen manually, in which case the last profile is kept by default. I see no reason to change anything to this.

[Matlock]

8 hours ago, pico said:

Such trivial concerns. Who really needs auto lens detection? We are not stupid.

Really? As they say "there are more out than in". So perhaps we could get back to taking pictures.

On another angle. Why does my M4 not record what lens, film, developer, etc. etc. has been used? I do believe that some people should "get a life" 😆