M10R (M10-R) sensor performance

[onasj]

Advertisement (gone after registration)

With the important caveat that this post is largely speculation, let's anticipate the performance characteristics of the M10-R sensor.

The Leica S3 sensor is 45x30 mm (1350 sq mm) and the DNG files are 9816 x 6512, for a total pixel count of 63,921,792 pixels, or 47,349 pixels per sq mm

The Leica M10 Monochrom sensor is 36 x 24 mm (864 sq mm) and the DNG files are 7864 x 5200 = 40,892,800 pixels, or 47,330 pixels per sq mm.

The pixel density of the two sensors is so similar that I assume the discrepancy comes from non-scaling edge usage factors, and/or small 0.01 mm-scale rounding errors in sensor size.

So if we assume that the S3 sensor and the M10 Monochrom sensor come from the same process, differing only in size and the presence of the Bayer color filter array in the former, then it's reasonable to assume the M10-R sensor also comes from the same process, and is basically a 36 x 24 mm crop of the S3 sensor (as others have hypothesized in this forum previously).  

If this reasoning is correct, then you can expect the M10-R sensor to be the same as the M10M sensor, plus a Bayer CFA, and to perform more or less the same as the S3 sensor on a per-square-mm basis.  Therefore, I would expect:

- DNG files that are 7864 x 5200 = 40,892,800 pixels

- Dual gain design

- 4.6 µm pixel pitch

- Base ISO = 100

- ISO range = 100-50,000 (basically one stop lower than the 160-100,000 ISO range of the M10 Monochrom, as one might predict from the lack of the Bayer CFA in the M10M)

- 14 bits per pixel

Good ISO 12,500 photos and decent ISO 25,000 photos would be terrific, and a significant improvement over the M10/M10-P, as noted here:

https://photorumors.com/2019/01/20/mega-shootout-sony-leica-and-phase-one-comparison/

[darylgo]

The M10-P high iso has been excellent.  I use it at base whenever possible with shadow recovery improved over  the M240.  So, this is good news (speculation), I wouldn't want to give up the gains of the M10-P.  I have heard the Q2 and SL2 were slightly less, both measured and by some users. 

Edited June 29, 2020 by darylgo

[Advertisement]

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

[Dennis]

3 hours ago, onasj said:

Therefore, I would expect ... 

OnasjRumors dot Com 🙌 

[Jeff S]

3 hours ago, onasj said:

With the important caveat that this post is largely speculation, let's anticipate the performance characteristics of the M10-R sensor.

The Leica S3 sensor is 45x30 mm (1350 sq mm) and the DNG files are 9816 x 6512, for a total pixel count of 63,921,792 pixels, or 47,349 pixels per sq mm

The Leica M10 Monochrom sensor is 36 x 24 mm (864 sq mm) and the DNG files are 7864 x 5200 = 40,892,800 pixels, or 47,330 pixels per sq mm.

The pixel density of the two sensors is so similar that I assume the discrepancy comes from non-scaling edge usage factors, and/or small 0.01 mm-scale rounding errors in sensor size.

So if we assume that the S3 sensor and the M10 Monochrom sensor come from the same process, differing only in size and the presence of the Bayer color filter array in the former, then it's reasonable to assume the M10-R sensor also comes from the same process, and is basically a 36 x 24 mm crop of the S3 sensor (as others have hypothesized in this forum previously).  

If this reasoning is correct, then you can expect the M10-R sensor to be the same as the M10M sensor, plus a Bayer CFA, and to perform more or less the same as the S3 sensor on a per-square-mm basis.  Therefore, I would expect:

- DNG files that are 7864 x 5200 = 40,892,800 pixels

- Dual gain design

- 4.6 µm pixel pitch

- Base ISO = 100

- ISO range = 100-50,000 (basically one stop lower than the 160-100,000 ISO range of the M10 Monochrom, as one might predict from the lack of the Bayer CFA in the M10M)

- 14 bits per pixel

Good ISO 12,500 photos and decent ISO 25,000 photos would be terrific, and a significant improvement over the M10/M10-P, as noted here:

https://photorumors.com/2019/01/20/mega-shootout-sony-leica-and-phase-one-comparison/

The least you could do is quote Nicci, whose thread discussion you participated in over 6 months ago, wherein he laid out most of this information.  He also noted how Leica started the process with the S007, which led to the M240.  
 

See especially Nicci’s posts # 22, 26, 38 and 73...

Jeff

[onasj]

1 hour ago, Jeff S said:

The least you could do is quote Nicci, whose thread discussion you participated in over 6 months ago, wherein he laid out most of this information.  He also noted how Leica started the process with the S007, which led to the M240.  
 

See especially Nicci’s posts # 22, 26, 38 and 73...

Jeff

No disrespect to Nicci was intended; thanks for finding his post (my imperfect memory of which was what I meant by "as others have hypothesized in this forum previously"), and for adding additional details.

I don't consider any of what I posted to be surprising or new—just a point of optimism that if the M10-R really uses a cropped S3 sensor then I hope it offers the same benefits as the S3.  I haven't found high-ISO sweep test images of the S3 but anecdotal reports suggest its high-ISO performance is good.

[Jeff S]

11 minutes ago, onasj said:

I don't consider any of what I posted to be surprising or new—just a point of optimism that if the M10-R really uses a cropped S3 sensor then I hope it offers the same benefits as the S3.  I haven't found high-ISO sweep test images of the S3 but anecdotal reports suggest its high-ISO performance is good.

Well, the relationship between the S007 and the M240 might provide some perspective, since this was where Leica began sharing wafers, as Nicci notes.  But obviously there’s much more to how cameras perform, including what’s bonded to that wafer such as micro lenses, filter arrays, etc, let alone differences and qualities that extra sensor size potentially affords. And then there’s obviously all the other hardware and software.  But we should have faith that Leica demands high and consistent standards across product lines, even sharing lenses in many cases.  
 

In this case, we also have the M10 Monochrom as a better basis for comparison, as it is likely the same sensor, albeit with Bayer array, in the same body.

Jeff

[fatihayoglu]

Advertisement (gone after registration)

Also, as far as I know, same sensor can have different base ISO values. So S3 might have a lower base ISO and M10R might have a higher base ISO.

[Alberti]

Of course because the M10-R has a different sensor filter layer. I expect some light loss.

As well, some sensitivity loss may come from using some signal used to fight vignetting - the new sensor covering with two layers has no micro prisms, so electronic handling might sacrifice some sensitivity because of some subtraction in the centre?

But the advantage to me is the sensor source: I love the M240 colours (based on S007 type for long time the best in the world ahum) and the new origin of the sensor from the S3 promisses a lot to me - so  I pulled the trigger on the M10-R. Even after seeing the M11.

[Camaro5]

I can only compare the real-world performance of the M10 to the M10-R.  I went out to do some nighttime street photography was impressed by the M10-R's ability to shoot in very low-light situations at ISO levels much lower than what I could do with the M10.  Because of that, obviously, there was no noticeable noise, and the images were as clean as a whistle.  Hand-holding the M10-R at night I could shoot at 1/60th of a second at f/2 and get no camera shake with a 35mm lens, a little less than 1/2F.  

By comparison, with the M10 I would've been using a higher ISO and maybe a little slower shutter speed.  Those images were sharp, and also looked good, but sometimes did have a little more noise.  

I think the stats and numbers are nice to look at, but you really have to go out and shoot to get a good indication of how the sensor/camera works in a particular situation.  

[jdlaing]

3 hours ago, Alberti said:

Of course because the M10-R has a different sensor filter layer. I expect some light loss.

As well, some sensitivity loss may come from using some signal used to fight vignetting - the new sensor covering with two layers has no micro prisms, so electronic handling might sacrifice some sensitivity because of some subtraction in the centre?

But the advantage to me is the sensor source: I love the M240 colours (based on S007 type for long time the best in the world ahum) and the new origin of the sensor from the S3 promisses a lot to me - so  I pulled the trigger on the M10-R. Even after seeing the M11.

The offset micro prisms are still there.