Jump to content

Monochrom M246 DNG technical analysis


sandymc

Recommended Posts

 

The only way Leica can dig their way out of this is if the sensor and ADC combine to form a non linear response similar to a classic film S curve - with compression of the highlights and shadows. 

 

It seems to me that one should allow the images to talk (rather than the superficial science). The decision to go 12 bit was not just a convenience . . . it was because the results were better.

 

They don't have to excuse themselves technically as long as the image quality is up to scratch . . . . and at least one of the reviewers spent an inordinate amount of time trying to PROVE that the 12 bit files were not good enough (knowing that they wouldn't be ) and failed. 

  • Like 3
Link to post
Share on other sites

The decision to go 12 bit was not just a convenience . . . it was because the results were better.

 

Jono,

 

Well, you've used the camera, and I haven't, but for me to believe that 12-bits is better than 14-bits in more ways than file size/image write time I'd need to see some evidence. "As good as in practical use" I would have no problem with, but better?

 

Regards,

 

Sandy

Link to post
Share on other sites

Advertisement (gone after registration)

There's no question that the review images we've seen from the M246 seem quite good.  And it appears indisputable that the M246 significantly advances the high-iso bar.

 

What remains unanswered is why Leica chose to reduce the RAW files in the M246 to 12-bits.  It's particularly curious since the sensor they based the camera on (the M240) is 14-bits.

 

There can be no qualitative benefit to going to 12-bits.  We can surmise that there was certainly a good reason, presumably a compromise based upon some dilemma.  Performance, perhaps.  Or file size.  Or fabrication cost.  Something.  But what we do know is that it wasn't to improve image quality.  You don't get better image quality by reducing bit depth.

 

So... the impact of that decision lives somewhere.  Maybe it's only in very large prints that it exhibits itself.  Or in pixel peeping a gnats ass.  Or maybe, in real world use, it doesn't exhibit itself, living entirely within the ghost of gaussian noise somewhere.  Leaving itself as little more than a technical curiosity, a simple byproduct of the technical brief.  But it's somewhere.

 

Those of us on the list, hoping we'll get the call next week saying "your camera is here," would like to know where that is.

  • Like 3
Link to post
Share on other sites

I should have looked at the IFD Offset for Thorsten's files: Lightroom created a new IFD but left the original one created by the camera fully intact, just bypassed.

 

The Thumbnail 160x120 uses compression '0007'x, JPEG- which explains the variable size. 12-bit packed data starts right where it should. The M8/M9/M Monochrom did not compress the thumbnail. M Monochrom used a 320x216 thumbnail stored using a fixed size.

 

Things have changed since I started this code in 1995 or so... My wife needed to process TIFF files that stored multiple images in one file. Photoshop would only process the first one, so I needed to split them up into individual Tiff files.

Link to post
Share on other sites

Jono,

 

Well, you've used the camera, and I haven't, but for me to believe that 12-bits is better than 14-bits in more ways than file size/image write time I'd need to see some evidence. "As good as in practical use" I would have no problem with, but better?

 

Regards,

 

Sandy

Hi Sandy

Sorry I should qualify that: When I said "the results were better" I was meaning taking the camera performance as a whole rather than specifically the image quality.

As I understand it nobody has managed to see anything detrimental (and some have tried very hard). 

  • Like 2
Link to post
Share on other sites

FWIW, I decided to email Mr. Puts and ask if he had any thoughts on the 12 versus 14 bit issue.  Here's part of what he briefly replied...

 

"Basically the change from 14 to 12 bits is already implemented in the regular M with its uncompressed and compressed option. The compression is losless and effects only the high lights."

 

I'm sure he'll have more to say when he does his technical review on the camera.

 

Jeff

Link to post
Share on other sites

 

"Basically the change from 14 to 12 bits is already implemented in the regular M with its uncompressed and compressed option. The compression is losless and effects only the high lights."

 

 

For the record,  :(, and  :blink: respectively.

 

Sandy

Edited by sandymc
  • Like 5
Link to post
Share on other sites

Guest JonathanP

"Basically the change from 14 to 12 bits is already implemented in the regular M with its uncompressed and compressed option. The compression is losless and effects only the high lights."

Sorry but that's nonsense. The M240 uses lossless DNG encoding without loss (!) of any bits. How can the highlights be affected if it's lossless? :-)

  • Like 3
Link to post
Share on other sites

Don't shoot the messenger.   :)

 

I also didn't tell Mr. Puts that I would spread his email around, so please don't challenge him directly (don't email him) unless and until he issues a formal statement.  I don't intend to follow with him on this.

 

Jeff

Edited by Jeff S
  • Like 1
Link to post
Share on other sites

Lossles and yet affects highlights and loses 2 bits? How can that be? Lossless loses nothing by definition. Knowing Erwin an explanation will be forthcoming.

Experts: has the MM2 DNG an LUT as part of a compression again?

Link to post
Share on other sites

You guys know more about this than I ever will. If due to eyesight problems one most often needs EVF, then I am stuck using the M246 if I want a Monochrom camera.

 

My downside is 12 versus 14 bit files from what I can tell. In someones schooled opinion, will that be notciable in 30x40 B&W prints? 

Lou, 

 

From my POV I find it rather disappointing that Leica is still using an Epson/Olympus EVF that Olympus replaced two years ago with the far better VF-4. There was no need to change the plug or physical dimensions of the body/shoe, as my VF-4 fits on an M240 (but does not of course, work). It would just be the video suite and/or the software that would have needed updating from the M240 electronics. The VF-2 is marginal for accurate focusing of fast long lenses. The VF-4 is far superior with a 50% larger area screen, better diopter adjustment, brighter illumination, faster refresh and double the pixels. Leica has really missed a trick with this. The viewer on the T is also better than the now very elderly VF-2. 

 

Wilson

  • Like 2
Link to post
Share on other sites

For the record,  :(, and  :blink: respectively.

 

Sandy

Hi Sandy

Perhaps the proof of the pudding is in the eating. I'm not alone in finding the DNG files from the 246 to have more highlight detail and to be generally more processable than those of the MM. 12 bit notwithstanding (and we have known about this and started out sceptical).  Of course, ther must be an effect, but if I can't see it in a 40" print then I'm not worried.

  • Like 1
Link to post
Share on other sites

The Black level is set to zero, the white level is set to 3750. I am SPECULATING that the image is clipped to remove noise. The A/D is 14-bits, I will also speculate that the image is shifted to get rid of noise. The M Monochrom (original) used (carefully) at ISO 10,000 shows no signs of fixed-pattern noise, shows Gaussian (random) noise. The M246 used at ISO 12,500 shows fixed pattern noise, correlated lines that are brighter/darker than surrounding pixels. I suspect that if the 14-bit, unclipped images were stored- this fixed pattern noise might be present at lower ISO. Fixed pattern noise is hard to get rid of using threshold and averaging over the outliers, which is effective in the M Monochrom. I suspect the engineers believed it best to just clip fixed-pattern noise out of the image than to leave it in. 

 

As always, I reserve the right to be completely incorrect. A Leica engineer is welcome to step in and correct me, PLEASE!

 

Working with the M246 DNG files provided, I am hard-pressed to find a glaring difference between low-ISO shots with it and my much-loved, most perfect CCD camera ever created, the M Monochrom. At High ISO: I hate fixed pattern noise, dealt with it long ago (1982) with mixed Radix FFT's. Which is in the code I am currently using, but have not called it yet. I can tell my story again of the $130K attached array processor catching on FIRE because the power supplies overloaded while processing images. You would hate fixed-pattern noise as well. The boss bought me an Intel Hypercube to replace it.

 

I would like to see some ISO 12,500 shots made with: Liveview off because it is EVIL and might be warming the sensor. Do not fire the camera on continuous, or a lot of shots in quick succession with High ISO. Pretend it is a Leica M3 Double Stroke. Expect sensor noise to double about every 5degrees centigrade (from the M9). Use a slow SD card, I use 4x cards. The idea is to eliminate data bursts, keep everything as smooth as possible. I have never shot with the M240, have no idea what the nominal write rate is. AND- turn off RF emitting devices. Like a Cell phone. Or Iwatch or IWhatever messing with my A/D convertors! These are just "you never know until you try" suggestions.

Edited by Lenshacker
  • Like 6
Link to post
Share on other sites

On the M246 compression scheme- I need to look at it closely. I would have used a running difference frame, that would work well for these images and is truly lossless. That might be used here. 

 

Nikon's compressed NEF scheme which is advertised as "Lossless" is "visually lossless", I stopped using it after finding out it does not recover all intensity levels.

Edited by Lenshacker
  • Like 1
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.

  • Recently Browsing   0 members

    • No registered users viewing this page.
×
×
  • Create New...