Video: Leica M10-R Interview with Stefan Daniel

[LUF Admin]

Advertisement (gone after registration)

More: Leica M10-R | Technical Data | Jono Slack Review

Just before the introduction of the Leica M10-R, I did an interview with Stefan Daniel, among others with these topics:

• Introduction of the Leica M10-R
• Background information on the sensor design
• Behavior of old lenses on the Leica M10-R

The interview is in German, English subtitles available

 

 

 

[Advertisement]

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

[lburn]

What about camera shake at this resolution? Presumably it was impossible to fit stabilisation into the camera body or it would have been too expensive?

 

[Jeff S]

11 minutes ago, lburn said:

What about camera shake at this resolution? Presumably it was impossible to fit stabilisation into the camera body or it would have been too expensive?

 

I asked Jono (separate thread with his review), and he said it presented no problems for him.

Jeff

[lburn]

Thank you. Thats good to hear and is probably due mainly to the build and balance of the camera. I would be very interested though to hear in the future what experiences people shooting E.G. the summicron 90mm APO have with this. 

[scott kirkpatrick]

Thanks, Andreas, for hosting this interview.  Mr. Daniel made two points that were perhaps too technical for the Wetzlar session today, but were very interesting.  I was intrigued to hear that there are now very thin IR and UV filters in the cover layers of the S3 and M10R sensors (or at least in the M10R version) and that there was some innovation in the pixel design that contributes to a higher quantum efficiency by getting the support circuitry more out of the way.  That stuff is hard to find out.

[dkmoore]

4 hours ago, lburn said:

What about camera shake at this resolution? Presumably it was impossible to fit stabilisation into the camera body or it would have been too expensive?

 

No issues whatsoever with this after thousands of images taken with the M10M. M10R will similarly present no issues. 

[jcm0]

Advertisement (gone after registration)

S. Daniel's video explains that the new M10R 41M pixels sensor (same core sensor as the M10M - M10 monochrom) has been tuned to be 10% more effective at receiving light (thus better shadow recovery/better dynamic range), and a new UV filter layer has been added with almost no impact on the added glass on top of the sensor.  Moreother, Jonathan Slack on his hands on review (https://www.l-camera-forum.com/topic/311399-leica-m10-r-review-by-jonathan-slack) discovered that the M10R can avoid blown highly by a margin of about 2EV (whereas the M10M has almost no margin). That's great news.

However, can someone please explain

- how it is possible to get the same resolution on the M10R as the M10M (same sensor),  as the M10R use the Bayer array to convert some pixels to color. Should we understand that the effective resolution of the M10R is lower even if the pixel count is 41M pixels ?

-  how it is possible to have a vastly improved highlight margin (approx. 2EV) on the M10R vs the M10M ?

Anyway, what is highly interesting, as already mentionned above by Scott, is the newly added UV filter which could improve the color fidelity. I'm really curious to see more real pictures of the M10R and how the color rendering has been improved. The M9 was a real joy for color amateurs, but I am - by far - not a fan of the color rendition of the M240 (for skin color and moderate/high iso especially) or the M10 (too much orange / emerald tones).

 

 

Edited July 16, 2020 by jcm0

[elmars]

The effective resolution of a color sensor is ever lower than a b/w sensor due to the bayer pattern btw the lack of it, if both have the same pixel number. 

The dynamic range is not 2 f-stops higher. I did not test it, but my impression is, that it is nearly 1 f-stop higher. 

[scott kirkpatrick]

I think the question of how the highlights are handled depends on the amplification of the signal that occurs after the photons hit the sensor pixel and are converted to electrons.  Remember that if the scene doesn't change, the lens doesn't change,  and the shutter speed doesn't change but each pixel becomes only half as big in area, then half as many photons go into each pixel when we take a picture. To get the same pixel value out as before (say, 256 for middle grey) either every photon has to be turned into more electrons (that's noisy) or the electron signal that is read out, in volts, has to be amplified by more.  Since the M10 had a tendency to blow highlights, and the "base ISO" was somewhere between 160 and 400, the simplest answer is just that the S3/M10M/M10R chip doesn't amplify its output by as much, and produces a genuine ISO 100 base.  But it also goes to very high ISO, and there is mention in the press release of a "dual gain" architecture, which suggests that the gain gets turned up when you ask for 6400 or more ISOs.  Bill Claff's techniques of analysis look at dark current and noise, so they can show these effects.  If someone can send him the files he works with, this will get answered soon.  That's photonstophotos.net. 

[Alistairm]

Could the apparent improvement in signal to noise ratio be software rather than hardware driven? 

[onasj]

On 7/16/2020 at 3:00 PM, scott kirkpatrick said:

Thanks, Andreas, for hosting this interview.  Mr. Daniel made two points that were perhaps too technical for the Wetzlar session today, but were very interesting.  I was intrigued to hear that there are now very thin IR and UV filters in the cover layers of the S3 and M10R sensors (or at least in the M10R version) and that there was some innovation in the pixel design that contributes to a higher quantum efficiency by getting the support circuitry more out of the way.  That stuff is hard to find out.

The latter point (10% more photons reach each sensor pixel) made me think that the S3, M10-M, and M10-R sensors are finally BSI sensors, whereas the M10 and M10-P sensor is FSI.  If they are really BSI sensors, they should also offer less color cast at the edges and corners when using wide-angle lenses.

[Dr No]

Excellent and informative video. Thank you Andreas, very well done.

Stephan Daniel is always likeable, isn't he? A good face of the company.

[Dr No]

On 7/16/2020 at 4:14 PM, lburn said:

What about camera shake at this resolution? Presumably it was impossible to fit stabilisation into the camera body or it would have been too expensive?

 

Sometimes I am wondering if Sony deliberately released the 37MP A7r mk1 with the most terrible shutter imaginable that created shutter shock to put off competition and plant a seed and establish a perceived need in customers.

A well designed camera design has minimal shutter shock. You can see this with Hasselblad and Phase one cameras which have 100 and 150MP. You can see this between the Pentax 67 and Pentax 67 II. The shutter in the M10-P really is an exceptional design and the mass of design really helps too. This is no cheap toy.

 

[LUF Admin]

vor 32 Minuten schrieb Dr No:

Excellent and informative video. Thank you Andreas, very well done.

Stephan Daniel is always likeable, isn't he? A good face of the company.

Thanks. And yes, he definitely is, like many Leica employees.

[jonoslack]

On 7/17/2020 at 4:38 AM, elmars said:

The effective resolution of a color sensor is ever lower than a b/w sensor due to the bayer pattern btw the lack of it, if both have the same pixel number. 

The dynamic range is not 2 f-stops higher. I did not test it, but my impression is, that it is nearly 1 f-stop higher. 

Hi There

I didn't say that there was a 2 stop higher dynamic range, I said that you could recover highlights from an M10-R image which was 2 stops over-exposed I quote:

 I’ve managed to gain good detail from an image over-exposed by 2 stops

not at all the same thing! I didn't test the dynamic range either, but I'd b e inclined to agree with you that it's 1 stop better. My point really is that it's a pretty big advantage over the M10 where you don't really have any leeway.

Best

Jono

[LBJ2]

Oh ! I didn't realize that was you Andreas until now. Ha Ha. Thank you for asking so many important questions. Very helpful for an informed buying decision. I have one question left after watching the interview. What defines modern Leica M glass? Any glass designed/developed since 2010 ? 

[scott kirkpatrick]

You can find plots of dynamic range vs ISO setting in various places (e.g. photonstophotos), and for most of the range of ISOs that a camera supports, the DR drops by one stop for every factor of two increase in ISO.  We tend to identify the "base ISO" on such a curve as the value of ISO below which the curve isn't straight any more.  But what I would really like to see is two curves as a function of ISO, for a given sensor, one plotting where the highlights give out (the sensor pixel can't hold any more electrons) and one for the noise floor, below which there is just noise.  In the days gone by, we did that, but only for a few values of ASA or DIN rating, by shooting a textured surface at a wide range of exposures.  (You had to develop a roll of film for each ISO, and the people who really got into that seem to have all gone away somewhere...)  One possibility is that the specialists who design pixels, their analog amplifier outputs and A/D circuits to produce those 41 million sets of 16 bits in the buffer have found some tricks with which to bring back the toes and the shoulders on the response curves that made film so flexible.  Without this extra information, we don't know whether highlights are preserved for us by just "pulling" down the exposure at low ISOs or by creating toes and shoulders to absorb the extremes.

[Manicouagan1]

Is setting Exposure Compensation at -1/3 stop a good practice with M10-R to make highlights recoverable, I found it to generally be the case with my M10.   

[Advertisement]

Hallo,

Look here: Leica M

[Chaemono]

I’d do EV -1, at least, on the M10-R. The shot noise is very organic. 

[SrMi]

49 minutes ago, Chaemono said:

I’d do EV -1, at least, on the M10-R. The shot noise is very organic. 

I do not see any reason for persistent negative exposure correction. M10-R preserves highlights well.
Leica M metering differs whether using live-view (multi-field metering) or rangefinder (mostly center-weighted metering). That means that, especially with rangefinder shooting, the exposure depends on the framing. 
When I have time, or can prepare, I prefer to do manual exposure using histogram and blinkies in live-view. Otherwise, EV 0 works quite well.