Side-by-side comparison of M11 vs. M10-R vs. M10-M (M10 Monochrome) at ISO 12,500 and 25,000

[SrMi]

Advertisement (gone after registration)

9 minutes ago, zwieback said:

100% pixel peeping - that is what I think is necessary if you also really want to use the 60MP. I would never do this for the pictures that I take but here it is about technical performance relevant or not.

I also think that the nose in the pictures posted above is very significant but this is just a screen grab from the forums pic, last post:

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

To me this level of noise on such a low resolution is very significant, for somebody else it might be ok because you otherwise do not get the shot at all.

That is not the way cameras should be compared, because the final output should be the measure.

[UliWer]

2 hours ago, digitalfx said:

"The M11 images were scaled down to the the same size of the M10-R and M10-M images."  

Is that the best way to compare the M11 image?

Yes, I think it's the right way. Otherwise you compare different enlargements of an image's section and of course the bigger enlargement shows more problems.

Though I fear the "model" was not ideal for the comparison. Even if she or he didn't move during change of cameras there was some breathing at least. And the close focus/ strong crop makes much difference then. 

[Advertisement]

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

[zwieback]

vor 23 Minuten schrieb SrMi:

That is not the way cameras should be compared, because the final output should be the measure.

so what's wrong, is it a mistake in your eyes to look at 60MP output of a 60MP camera... To me 60MP IS indeed the final output if I would use a camera like this. If the "final output" is instagram an Fuji Xt1 is alredy far bejond what is needed.

Edited January 16, 2022 by zwieback

[onasj]

I conducted the test in an extremely well-lit room (the room has 3,000 real-life watts driving 6,500 LEDs that are spectrally matched to recreate sunlight, minus the UV) so that I could shoot the 50 APO at f/5.6 and f/8—where the lens is sharp enough to not be the bottleneck to detail capture.  Likewise, under these conditions shutter speeds were very, very fast (the slowest shutter speed was 1/4000 s), so my sleeping cat was definitely not motion-blurring the shots, despite his occasional breathing :)

You can download all six original RAW files at the link I provided in the first post and analyze them using your preferred methods.

The reason the M10-M has a fundamental advantage of about 2 stops to start with is because the color filter array of almost all modern digital cameras eats about 3/4 of the photons. Which is to say that <1/4 of the photos actually make it to a photodiode.  That's the job of the color filter array: to block out non-red, non-green, or non-blue photons.  The lack of a CFA also helps preserve sharpness because there is no need to de-mosaic the image, as is done with the color cameras.

[UliWer]

2 minutes ago, onasj said:

cat was definitely not motion-blurring the shots, despite his occasional breathing :)

I am sure the cat was not responsible for blurring the individual shots.

But what happened when you changed the camera?

[onasj]

14 minutes ago, zwieback said:

so what's wrong, is it a mistake in your eyes to look at 60MP output of a 60MP camera... To me 60MP IS indeed the final output if I would use a camera like this. If the "final output" is instagram an Fuji Xt1 is alredy far bejond what is needed.

What matters most is image quality (noise, detail, etc.) per picture height.  Otherwise you could conclude a 1-pixel image was perfect in detail capture.  So I could either scale up the M10-M and M10-R images, or scale down the M11 images.  In practice, the difference between either workflow does not change any of the findings.  I added an un-scaled 1800-pixel crop of the M11 ISO 12,500 image, below (and in the Dropbox folder).  You can see it is substantially similar to the scaled-down version, but slightly more noise and slightly less sharp, as math would predict.

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

Edited January 16, 2022 by onasj

[onasj]

Advertisement (gone after registration)

5 minutes ago, UliWer said:

I am sure the cat was not responsible for blurring the individual shots.

But what happened when you changed the camera?

Oh the images are definitely not identical—I took them handheld after all.  But that wasn't the point of the test.  I just wanted to compare high-ISO noise and detail capture.  It's true you have to compare non-identical regions of fur, etc. but I think the findings are pretty clear regardless.

[darylgo]

@onasj Thank you for this comparison, great information, and kudos to your cat for being a great subject.  I am finding the whiskers and fur to be an excellent test, perhaps better than fruit and charts.  I am on the fence about purchasing the M11, the differences from model to model and generation to generation are getting more nuanced.  

[zwieback]

thanks for the explaination onasj but but like I have written I do not compare to anything but look at the absolute output only. The point at which noise comes in does not depend on the image height, the relevance of noise to image quality on the other hand does. I saw quite some noise coming in at 1600 - 60MP DNG, that's it, not saying that it is high or low compared to any other camera. Look at the noise level yourself and when it starts to increase and you will probably find that it is in the 1600 range.

The personal rating certainly also matters and again, even only looking at the low res pics you posted here on the forum I rate the noise to be significant, you might find it OK, high, low whatever is your taste and I am fine with that.

[Jipster]

1 hour ago, onasj said:

I conducted the test in an extremely well-lit room (the room has 3,000 real-life watts driving 6,500 LEDs that are spectrally matched to recreate sunlight, minus the UV) so that I could shoot the 50 APO at f/5.6 and f/8—where the lens is sharp enough to not be the bottleneck to detail capture.  Likewise, under these conditions shutter speeds were very, very fast (the slowest shutter speed was 1/4000 s), so my sleeping cat was definitely not motion-blurring the shots, despite his occasional breathing :)

You can download all six original RAW files at the link I provided in the first post and analyze them using your preferred methods.

The reason the M10-M has a fundamental advantage of about 2 stops to start with is because the color filter array of almost all modern digital cameras eats about 3/4 of the photons. Which is to say that <1/4 of the photos actually make it to a photodiode.  That's the job of the color filter array: to block out non-red, non-green, or non-blue photons.  The lack of a CFA also helps preserve sharpness because there is no need to de-mosaic the image, as is done with the color cameras.

How do you come up with 3/4? I am genuinely curious.

[onasj]

1 hour ago, Jipster said:

How do you come up with 3/4? I am genuinely curious.

I think the typical light loss from standard RGGB Bauer CFAs is 5 db (factor of 3) from technical papers I read long ago, but I suspect the CFA used by the M11 passes narrower wavelength bandwidths (perhaps like the Phase One IQ3 Trichromatic), which would explain the better color and lower native ISO. In addition to raw photon loss from wavelength filtering, there’s also spatial information lost from the fact that only 1/4 (RB) or 1/2 (G) of the sensor detects light of that color. So a 60-million-photodiode sensor only has 15 million red-detecting photodiodes. Together these factors explain why the M10-M overall has about a 2-stop edge over the M10-R in terms of noise and acuity.

Edited January 16, 2022 by onasj

[SrMi]

2 hours ago, onasj said:

I conducted the test in an extremely well-lit room (the room has 3,000 real-life watts driving 6,500 LEDs that are spectrally matched to recreate sunlight, minus the UV) so that I could shoot the 50 APO at f/5.6 and f/8—where the lens is sharp enough to not be the bottleneck to detail capture.  Likewise, under these conditions shutter speeds were very, very fast (the slowest shutter speed was 1/4000 s), so my sleeping cat was definitely not motion-blurring the shots, despite his occasional breathing :)

You can download all six original RAW files at the link I provided in the first post and analyze them using your preferred methods.

The reason the M10-M has a fundamental advantage of about 2 stops to start with is because the color filter array of almost all modern digital cameras eats about 3/4 of the photons. Which is to say that <1/4 of the photos actually make it to a photodiode.  That's the job of the color filter array: to block out non-red, non-green, or non-blue photons.  The lack of a CFA also helps preserve sharpness because there is no need to de-mosaic the image, as is done with the color cameras.

All PhotonsToPhotos measurements show only one stop difference (Q2 vs Q2M, M10-R vs. M10M). 3/4 number looks too high.

Edited January 16, 2022 by SrMi

[Torontoamateur]

WOW ! a $9,000 camera with a $8,000 lens to take pictures of a cat. Give me a break !!!!

[onasj]

2 hours ago, Torontoamateur said:

WOW ! a $9,000 camera with a $8,000 lens to take pictures of a cat. Give me a break !!!!

You’re right—I should have used an IQ4-150.

[onasj]

2 hours ago, Torontoamateur said:

WOW ! a $9,000 camera with a $8,000 lens to take pictures of a cat. Give me a break !!!!

More worthy subject: M10-R, Leica Stemar 33 mm @ f/3.5 (takes stereo pairs). If you cross your eyes, you can see the subject in 3D:

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

[tightsqueez]

4 hours ago, onasj said:

More worthy subject: M10-R, Leica Stemar 33 mm @ f/3.5 (takes stereo pairs). If you cross your eyes, you can see the subject in 3D:

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

Just don’t cross your eyes while using it… might get your feet wet. 
 

Appreciate the comparison. 

[M9reno]

5 hours ago, onasj said:

More worthy subject: M10-R, Leica Stemar 33 mm @ f/3.5 (takes stereo pairs). If you cross your eyes, you can see the subject in 3D:

I’m glad to know that the M10-R fits a Stemar (as I remember, an M240 did not).

[onasj]

2 hours ago, M9reno said:

I’m glad to know that the M10-R fits a Stemar (as I remember, an M240 did not).

I had to modify the lens slightly to get the Stemar to fit the M10 family, documented here: 

 

[M9reno]

2 hours ago, onasj said:

I had to modify the lens slightly to get the Stemar to fit the M10 family, documented here: 

 

Thank you very much. It is good to know that the modification is not a radical one. However, for the moment, as long as I continue to use an M9 the Stemar will be used with that, since it works with it unmodified. (I also use it on film M’s).

[Advertisement]

Hallo,

Look here: Interesting Leica gear

[adli]

13 hours ago, Torontoamateur said:

WOW ! a $9,000 camera with a $8,000 lens to take pictures of a cat. Give me a break !!!!

 

This forum demonstrates a rather rough environment for the few people who actually contributes positively...

 

Cudos to onasj for sharing his thoughts with us.