Official Response from Leica on Laundry List

[wparsonsgisnet]

Advertisement (gone after registration)

Hi, Rubén.

 

I'll take whatever makes the most sense. I thot the DMR wrote a 16-bit file; is this not so?

 

Since it would be nice to do this with a firmware update, we may have to accept whatever that allows.

[rosuna]

The DMR writes a 16-bit file, uncompressed.

The non-linear compression made by the M8 is a good idea, but I cannot understand why they choose 256 variants.

[sean_reid]

Sean has talked about this on many occasions and we need to keep this on the wish list , no doubt.

 

I had one were you set the EV compensation and in the finder it flashes for a 1/2 second what that compensation is when you press the shutter. This way your fully aware of it

 

Leica hasn't forgotten about revising the camera to make for quicker access to certain functions.

 

Cheers,

 

Sean

[sean_reid]

No, I cannot see any clipping. The recovery slider cannot correct the problem. I see this same problem in several pictures, involving bright lights and skin. I would like to preserve more tonal range from the original file. I am not convinced by the arguments about the absence of negative effects of the 8-bit compressing. Detail in the highlights are lost, but Leica says it is "unnseen" in real cases.

 

Hi Ruben,

 

16-bit might help but the best solution for that lighting would be to work with a lower contrast lens.

 

Best,

 

Sean

[rosuna]

Of course, any reduction of the contrast in the motive is good.

I think linear 16-bits files are unnecessary. My point is that a non-linear encoding of 10-bits (instead of 8-bits) would provide a safer and wider container for the original information (which comes from a 16-bit space). That's all. I see 8-bit encoding as too dangerous. It is an excessive compression of the original data. I prefer to work with a security margin.

My question to Leica is, would it be possible to employ 10-bit encoding (=1024 variants instead of 256)?

[sandymc]

Thanks Guy! And all parties attending this list creation...

 

I quoted because there is an interesting message implied through these quoted items:

 

Nos. 7 & 16 are related... and there's more: the EXIF fields named

"Calibration Illuminant 1" and "Calibration Illuminant 2" are "some odd binary" in f/w 1.06 and 1.09, but in f/w 1.092...

Calibration Illuminant 1 Standard Light A

Calibration Illuminant 2 D65

 

Which would indicate--from EXIF field "Linearization Table", that magic used to interpret 14 bits to 8--that these EXIF field values have been literally normalized to a constant value in 1.092. In earlier f/w, these values were computed in camera much like that magic used to "compress" 14 bits into 8. What Stefan's reply implies is that C1 is just one of many DNG converters, "i e Adobe Camera Raw"... Some converters will use these "Illuminant" values, where C1 may not.

 

I agree with Leica that the "Lens Detection" should be a user specified Boolean( on/off ), and not left to the f/w to determine.

 

As for the presets... I would never trust them because they are a holdover from the type-casting of film stock, not the light drawn through the optics. "Auto" will always yield control to the camera, yet provides a good enough guess; "Manual" puts the photog in control, with "Pringles" lids, ExpoDisc or some external "mark twain"; and "Kelvin" yields yet more to those who sense ambient temps.

 

And No. 23: again, I am thankful for the "pure and clean" over the "assumed and asserted"... I'm newly curious about the use of these "Calibration Illuminant" values... are my ~1000 1.06 DNG files the first and last of their species? Is a D65 "white point" what I want? I suppose just as mythical as a "True 16bit" file

 

rgds,

Dave

 

The way the DNG format is specified, Matrix 1 tell the raw converter how to convert for Calibration Illuminant 1, and Matrix 2 tells the raw converter how to convert for Calibration Illuminant 2. For illuminats other than 1 or 2, the raw converter interpolates between the two matrixes. What the camera thinks the illuminant is, is encoded into the DNG as well. This is the way the Adobe products work - e.g., in Lightroom, you have the color temperature slider. However, from what Stephan was saying, clearly C1 ignores the DNG values, and uses a built in M8 specific color conversion system. Which may or may not give better results than the intepolation between matrixes system.

 

Sandy

[george +]

Advertisement (gone after registration)

Of course, any reduction of the contrast in the motive is good.

I think linear 16-bits files are unnecessary. My point is that a non-linear encoding of 10-bits (instead of 8-bits) would provide a safer and wider container for the original information (which comes from a 16-bit space). That's all. I see 8-bit encoding as too dangerous. It is an excessive compression of the original data. I prefer to work with a security margin.

My question to Leica is, would it be possible to employ 10-bit encoding (=1024 variants instead of 256)?

 

Ruben, 256 can be represented by one byte=8bits. If you want 1024, that takes two more bits=10. And how you handle that in a computer accross byte boundaries may require (much) more effort. This was not a major problem in the good old days when we sliced and diced bits any way we pleased. But it may be more than you want to tackle these days.

 

You could pack it of course into two bytes by which time you might as well go for sixteen bits and forget about the Leica trick. Just wait a bit longer for reads and writes - and pack less on a card. Anything is possible - at a price. Compromises make the world go round. And complainers of course.

[Sperandio]

Thanks Guy for taking this up with them.

 

27 has not been fixed to my experience.

 

Dirk

 

nor for my one!

[pklein]

Aren't most of the old M lenses able to be coded by Leica? If I remember correctly, it is only a handful of old M lenses that could not be coded, some of which were not compatible with the M8 regardless of whether they are coded or not.

 

RobSteve replied:

As for the screw mount lenses, I suppose Leica could offer screw mount adapters with the proper codes.

 

Yes, most recent Leica lenses can be coded. But not some of the older lenses-- including, for example, my beloved Dual-Range Summicron. And what about my VC Nokton 50.1,5, my 21/4 and my 90/3.5? Not to mention my old Canon Serenar 50/1.8 which might just be a great "sunny-day lens." (TM Sean Reid!).

 

The M8's price is high enough, and we will already have to buy filters if we have more than two lenses. The cost of coding all our lenses can add up. This is a negative point in the M8 purchase decision. For those of us who can't amortize our equipment in a photography business, a lens select menu is an attractive option, and a positive point in the decision to purchase or not.

 

As someone else mentioned, Leica doesn't have to put in other brand lenses. They already have a table, all we need is a menu to manually access it, so we can choose the "best equivalent" to our uncoded, old or non-Leica lenses.

 

--Peter

[Iron Flatline]

I'm really grateful to Leica for taking the time to answer us on a line-by-line basis. Few other companies would bother to do so.

[newyorkone]

I'm really grateful to Leica for taking the time to answer us on a line-by-line basis. Few other companies would bother to do so.

 

Ditto, and loving my new black M8!

[atufte]

Kudo's to Leica for being so responsive. Now with the compliments out of the way, on to my rant

 

 

 

I'm not buying this explanation. They trust the user to turn lens detection on and off and make decisions about exposure compensation and ISO selection via menus all of which can be left at a setting that can screw up your photos if you forget about it. But they want this one feature to be decided by the camera in 'point and shoot' mode. Leica M is all about direct manual control by the photographer. This is a feature users should demand.

 

I understand some marketing guy might think the present set-up gives Leica an edge. But I think this is wrong-headed and simplistic. Any thing that makes the M8 easier and more flexible in use, and gives users more options will accrue to Leica's long term benefit. The long line of used Leica lenses, CV and Zeiss M compatible products, all these options make the Leica M platform more attractive to users and work in Leica's favor and expand the market it serves.

 

If the reason for the lack of this option is an engineer wanting to protect us from ourselves then that would be even worse. When in doubt always let the photographer have control!

 

 

I could not agree more...this is only a sorry excuse to try to sell more new "coded" lenses, i really hate when company's like Leica speak to us like we are small children

and that we don't understand things like this, (WE DO!!!) the feature in the Nikon D2 lineup is by far the smartest and best solutions for using manual (non data coded) lenses digitally on the market today, and the feature is not even patented, SO COME ON LEICA get off your high horse and listen to the your users... This is guaranteed

the easiest solution for any lens compatibility issues with this entire system, and should

be taken very seriously by Leica...

[atufte]

Thanks to Guy for getting these questions to Leica, and to Leica for their responses.

 

 

 

Yes, please, to the black-and-white DNG-only workflow! Just the ability to display and evaluate the review image in black and white. The reasons why:

 

1. Batching the JPGs for mass delete doesn't work for me, because occasionally I do use straight jpg.

 

2. The jpgs do take space---for me, that space is wasted.

 

3. The camera is so much faster shooting straight DNG---about two seconds to write out an exposure vs. five seconds for jpg. I never fill the buffer when shooting dng, but sometimes do when shooting dng+jpg. Review is so much faster when the camera isn't busy writing to the card. After using DNG only, the camera feels slow in DNG+jpg.

 

Add to this Leica's B&W heritage, and it seems a natural. I don't need a dedicated B&W digital, the M8 is great for it---just let me review and zoom in monochrome or color, as I see fit.

 

What a great, simple, unique feature that would be available only on the M8!

 

Until later,

 

Clyde Rogers

 

 

Of course i forgot, the B&W DNG ONLY workflow, VERY important and should with Leica's long history with B&W shooters be an absolute minimum, why the h... not??

This is way to easy to implement and even cheap P&S cameras can do this today (and yes, even with raw...)

[wparsonsgisnet]

Ane other item --

 

If we are going to have to bracket our exposures, how about an automatic bracketing function on the M8.

 

This should be one stop either way?

[colorflow]

Alan I can say without a doubt that yours truly has really been all over this one more than anything, Stefan and Christain have heard my fair share on this one, it is the biggest issue we have and they know that very well. I do know there are a few camera's on the engineering bench being torn apart to figure this out. We talked about it a lot , it is pretty random and very hard to pin point but the symptoms are very similar so hopefully very soon the bright light will kick on saying we solved it.

 

 

Riley congrats

 

Guy, thanks for impressing upon Leica the importance of this problem. By the way, did you ever find out how they fixed your M8 problem? I remember you had one with similar problem to sudden death.

 

Alan

[Guest guy_mancuso]

Not yet, still waiting word on that.

[rosuna]

Ruben, 256 can be represented by one byte=8bits. If you want 1024, that takes two more bits=10. And how you handle that in a computer accross byte boundaries may require (much) more effort. (...) Anything is possible - at a price. Compromises make the world go round. And complainers of course.

 

The 8-bit encoding is used only for storing the information into the DNG file. It only affects file size. When you open the file the information is placed on a 16-bit space. Photoshop works on a 16-bits space. That is not the problem.

[george +]

The 8-bit encoding is used only for storing the information into the DNG file. It only affects file size. When you open the file the information is placed on a 16-bit space. Photoshop works on a 16-bits space. That is not the problem.

 

Yes Ruben ,that is exactly what we are talking about. Writing double sized files is what Leica wanted to avoid. Both the writing time and the storage it takes. Whether they were right in their compromise remains to be seen. An option to write 8 or 16 bits could be made into a user choice. Whatever . . . But ten bits - as you suggested above - does not make much programming sense. Sorry . . . .

[eronald]

Yes, Adobe does what you say. C1 uses an ICC-based workflow and so you need a full profile file for the camera, either one originated by Phase One, or one which has been supplied by a third party such as me.

 

Edmund

 

The way the DNG format is specified, Matrix 1 tell the raw converter how to convert for Calibration Illuminant 1, and Matrix 2 tells the raw converter how to convert for Calibration Illuminant 2. For illuminats other than 1 or 2, the raw converter interpolates between the two matrixes. What the camera thinks the illuminant is, is encoded into the DNG as well. This is the way the Adobe products work - e.g., in Lightroom, you have the color temperature slider. However, from what Stephan was saying, clearly C1 ignores the DNG values, and uses a built in M8 specific color conversion system. Which may or may not give better results than the intepolation between matrixes system.

 

Sandy

[Advertisement]

Hallo,

Look here: Leica M

[rosuna]

Yes Ruben ,that is exactly what we are talking about. Writing double sized files is what Leica wanted to avoid. Both the writing time and the storage it takes. Whether they were right in their compromise remains to be seen. An option to write 8 or 16 bits could be made into a user choice. Whatever . . . But ten bits - as you suggested above - does not make much programming sense. Sorry . . . .

 

If you write an unencoded 16-bit file you get a doble sized file, but a 10-bit non-linearly encoded file doesn't implies a double sized file. It would be a file marginally bigger than the actual 10MB file. That is the point.