Leica New Leica Mono coming soon???

Peter Klein

Regular
Location
Seattle
Real Name
Peter Klein
Brian: According to all I've read today, the M-246 takes only the old EVF-2. Just like the M-240.

I've been reading the reviews, and the files from the new camera looks great, and extend its available light abilities by about 1.5 stops over the M9-based MM. First thought is that I want one. And then, I got to play with an MM for an afternoon recently, and was astounded at how good it was. Maybe that's enough. I wonder whether used MMs will drop in price enough to tempt me to get one. But then again, there's the sensor corrosion issue and the prospect of being without one's camera for months it it hits.

I will continue to shoot my M8 and my Olympus E-M5 as I contemplate these existential dilemmas. M8Raw2DNG has been almost like getting a new camera.

--Peter
 

Amin

Hall of Famer
If I had an extra $7450 USD, I'd already be on the waiting list - but I don't. I can't complain, since I just traded for a Safari set...

I'm on the waiting list hoping $7450 falls out of the sky between now and when I get the call :rolleyes:.
 
Making the change to the KAF-18500 is not hard, and the status of the device changed at On Semi from "In Stock" to 13~16week lead time. Reading in between the lines: they may be revising the part.
 

Amin

Hall of Famer
This is a good review on the cameras, if you have not seen it already.

http://www.ultrasomething.com/photography/2015/04/sensors-and-sensibility/

I'm having a hard time understanding why his results look that way. My own black and white conversions with the M240 at high ISO are cleaner and more detailed than what I got from the M9 using the same software he's using.

Oops - I just realized MM9 was M Monochome (9). I thought it was M9 converted to B&W.

Oops #2 - Now I see that he got way better M240 results after applying color noise reduction, which I always do for high ISO files.
 
Last edited by a moderator:
What I noticed was the M Monochrom rendering of the sky was smoother than the M246. You could see less grey levels used by the M246 image.

Think of bits-per-pixel as a dimension of the image. the M246 is (roughly) Image( 6000, 4000, 4096) and the M Monochrom is Image( 5212, 3468, 16384).

I would like to see Leica fix this shortcoming of the M246, at least offer 14-bit pixels at lower ISO.
 
Last edited by a moderator:

Amin

Hall of Famer
What I noticed was the M Monochrom rendering of the sky was smoother than the M246. You could see less grey levels used by the M246 image.

This is what he said about that:

On first glance, the MM9 appears to give the smoothest gradations at base ISO. However, it’s important to remember that the file’s overall micro-contrast has been softened by the upsizing process, which has the effect of greatly minimizing the visible noise. In reality, files from the MM9 (that have not been upsized) have a nearly identical noise floor as the M240 and M246 cameras.
 
I've looked at M Monochrom files in Hex, truly raw pixel values. When I can do the same with M246 files, after unpacking the 12-bit images into 16-bit words- will be easy to see.

12-Bit cameras are now in the realm of entry level DSLRs. I expect more from a premium camera. Basically, this CMOS sensor cannot deliver the uniformity required for a 14-bit image.

The upres'd image would not make large sections of sky have more tones of grey in it, it might smooth out localized noise events.

The M240 came out shortly after the M Monochrom; Erwin Puts and others demonstrated that the 18MPixel 14-bit camera had higher resolving power than the 24MPixel 14-bit color M240. The question for anyone spending $7500 on this new 24MPixel 12-bit camera: will the next color M camera, most likely coming within a year, provide a better monochrome image than this dedicated camera. Look for someone comparing the M246 against a Nikon D810 to provide a reasonable comparison of what to expect within a year from Leica.

I also note that M246 DNG files have not been provided for download at this point, I am hoping this changes soon.
 
Last edited by a moderator:
There is another possible explanation as to why the Monochrome version of the M240 has 12-bit pixels: the original CMOSIS sensor might be using a 12-bit A/D convertor with a non-linear scale, same as used in the other CMOSIS full-frame (24x36) detector.

http://www.betopcom.com/public/imag...43.1.pdf?PHPSESSID=mse97nirl5c7lclml2rhfkvgs4

The above CMOS sensor has the full data sheet published. The A/D uses an indexed scale to convert voltage from the Pixels to digital. If this is how the M240 works, then the firmware could easily be converting the indexed/non-linear values to 14-bit linear. With a color camera, this works as the results are interpolated. There are still 4096 values, but after interpolation- it would work. Monochrome images do not get interpolated, the conversion from 12 bits to 14 bits would be very obvious and distracting. The A/D could be set for 12-bit linear mode, which is selectable as an operating mode. Means setting some register values in the chip. Just firmware.

If you want to get the new camera, you should. There will always be better cameras coming. However: 12-bits is a huge step backwards from the original M Monochrom. It means that 4 grey levels get binned into 1. Resolution is more than just rows and columns, the camera must also differentiate a change in intensity level to resolve an object. Many M Monochrom owners have found the ability to pull details out of the shadows a strong point: this is possible because of the 14-bit deep pixels. 12-bit depth with a monochrome camera is much more limiting than having 12-bits in a color camera. 14-bits is cutting it close, and 16-bits would have been a real improvement over the old one. Out of camera Jpegs will be great with this camera due to the heavy image processing applied by the firmware. Great for video, where you don't post-process every frame for exposure and contrast. Not so good for someone that likes to play with curves and tones in Lightroom.
 
Last edited by a moderator:
http://www.slack.co.uk/2015/Elliott.html

http://www.slack.co.uk/2015/monochrom/monochrom246/index.html#/view/ID255558

Look at the 6th image in Jono's series, the one with the two trees alongside the road, with open sky above. Look towards the upper portion of the image- you can see the effects of the quantized image.

Compare it with the M9 image converted to Monochrome,

17138851795_800b266e74_b.jpg
M1012665
by fiftyonepointsix, on Flickr

The above shot is straight conversion to linear-monochrome using my Fortran program.

and this one from the M Monochrom, straight export from LR.

17124027666_0cdd507f8f_b.jpg
L1005072
by fiftyonepointsix, on Flickr

If you do not see or care about differences in the image, don't worry about 12-bit vs 14-bit. I would like to see some DNG files posted from the new camera.
 
Last edited by a moderator:
First of DNG files made available for download by Thorsten Overgaard. These are High ISO 6400 and 12500 so bear that in mind.

At 100% view, you will see pattern noise following lines in the array. I will be happy to see some base ISO images.
 
It's interesting that the DNG file posted by Thorsten is now in Motorola byte order rather than Intel byte order. Mac and Windows based machines both use Intel format these days.

The M246 absolutely uses 12-bit pixels, linear. It does not use a look-up table. Black level is set to zero. White level is set to 3750 in all of the files, as reported before. This is with the 50/0.95 mounted, I don;t know of white level changes depending on which lens is mounted. White level on the M9, M Monochrom, and M8 stays at 16383 no matter what lens is mounted.

This means the camera can depict 3751 shades of grey. That's a lot more then 50, but less than 1/4th that of the M Monochrom.

The baseline noise reported for the two cameras at base ISO is "1". That is low. The baseline noise for the M8 is reported as "4".
 
Last edited by a moderator:
17172357350_d259bbb4ac_b.jpg
Skate and Fun Zone
by fiftyonepointsix, on Flickr

Above, My image from the M Monochrom at ISO 10,000

Below, Thorsten's sample image at ISO12,500 from the M246.

17173706649_844d7798d2_b.jpg
M246: Thorsten's Examples
by fiftyonepointsix, on Flickr

LR6 used to process, NR synchronized between the images. Sharpening set to 0, Luminance set to 30, detail to 30.

If you download Thorsten's original DNG you will see cross-hatch pattern noise in the ISO6400 and ISO12500 images. The ISO 12,500 has a lot of clipping in it. The M Monochrom has a lot of Gaussian Noise. Note that I use Sandisk 8GByte 4x memory cards, use of high-speed cards causes some pattern noise to creep in. The M Monochrom images are full 14-bit, the M246 images cover values 0:3750. The cross-hatch pattern noise is still an issue in the M246 images.

M Monochrom at ISO 5,000.

17358012392_dd759c0328_b.jpg
Skate and Fun Zone
by fiftyonepointsix, on Flickr

The claim that the M246 gives a 1.5stop advantage would mean comparing this image with the ISO 12,500 samples.

Thorsten used a 50/0.95 Noctilux and I used a 1950 Jupiter-3 with Zeiss glass in it. Both models are beautiful.

You cannot believe how much time I spent tracking noise sources in the lab, in the 1980s. Once took a 200pound VAX 11/730 onboard a P3 Orion to track down noise being induced on a digital Infrared sensor. I have that code ready to go anytime Leica needs it. The M Monochrom amazes me. I'll leave it at that. I also have a Tek real-time spectrum analyzer. That would also come in handy. It was expensive.

Am I missing something? The M246 shot almost looks like what you get from a flat-bed scanner with the length-wise non-uniformity.

This is Thorsten's image with sharpening Off, NR off, LR6 used.

17173199210_c3190f03bb_b.jpg
Thorsten Sample Image
by fiftyonepointsix, on Flickr

I see a lot of length-wise non-uniformity in this image, as opposed to Gaussian noise. Am I the only person seeing this?
 
Last edited by a moderator:

asiafish

All-Pro
Location
Bakersfield, CA
Real Name
Andrew
I'm not familiar with the terms you are using, the M Monochrom images look a lot smoother and more uniform to me.



Oh and yes, both models are beautiful, but yours looks like she is having a lot more fun.
 
One of my first jobs was writing software to characterize the detectors used for digital imagers, including determining source of noise once they were installed in the sensor. Early 1980s.

The images from the M Monochrom are very uniform, "Gaussian Noise" is random. Most likely why some people compare it with film grain.

The noise from the M246 has a pattern to it, like vertical and horizontal lines going through the image.
 

asiafish

All-Pro
Location
Bakersfield, CA
Real Name
Andrew
Now I see what you mean. The noise is perfectly symmetrical. I played with Thorsten's DNG a bit last night and it does look more digital. I will say it also appears to have slightly more dynamic range than what I get from the M Monochrom. I guess the reviewers are all correct in saying its got about an extra stop in the shadows.

Now that I look for it, the noise is not as pleasant as what I get from the old Monochrom, but I doubt most observers will notice, unless they pixel peep or print large.
 
More dynamic range with less resolution for intensity- means more quantization. It's ashame that it does not have 14-bits or better to resolve the image.
 
I'm very happy with my present M Monochrom.

I was surprised to read that some people were getting pattern noise in their original cameras at ISO 3200. I get clean images at ISO 10000- cannot make any fixed pattern noise out of it. I am careful as to which memory cards I use. And of course I never turn on my cell phone when shooting High ISO. Actually- I turn it on when dialing out.
 

Latest posts

Latest threads

Top Bottom