DXO Mark Leica M10 Score

[dunos]

Advertisement (gone after registration)

I've said it before:  It is ridiculous to judge a sensor on the way the camera noise reduction works - it is a camera software matter full of compromises and trade-offs. It is legitimate to have a preference - but not to have an absolute judgement.

 

I find it difficult to judge a camera system by focusing on a single component. An assessment of a single component adds weight to the whole, but if people are looking for the best (for their personal definition of best) in every single component without compromise then they will be disappointed, not just with Leica but with everything they purchase.

 

That is unless they like driving The Homer.

Edited January 9, 2018 by dunos

[Advertisement]

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

[jaapv]

Precisely, as Leica has always been biased to leave as much as possible to user input.  Especially something like reducing luminance and colour noise; a subject-related postprocessing workflow is surely superior to an in-camera one-size-fits-all precooking.

[Mark II]

Precisely, as Leica has always been biased to leave as much as possible to user input.  Especially something like reducing luminance and colour noise; a subject-related postprocessing workflow is surely superior to an in-camera one-size-fits-all precooking.

 

 

It seems likely that every model since the M240 has substantial noise reduction at high ISOs applied to the RAW files - it was explicit in some of the block-diagrams for the Maestro processor, and it is obvious if you compare equivalent high-ISO shots vs pushed low-ISO. Similarly, the "green shadows" problem was probably the result of trying to mask pattern noise by manipulating the sensor data prior to writing the DNGs. And all M-series CMOS cameras have what appear to be fairly arbitrary limitations on long exposures and no ability to disable dark-frame subtraction.

 

Much of this is, unfortunately, common practise in the camera industry. It is an attempt to make the files out-of-camera appear as good as possible "before" processing. Aside from dark-frame subtraction (which is useful in most cases, though it should still be optional), generally the less in-camera processing that is made to a RAW file the better.

[Chaemono]

Could be, but the comments of this reviewer on the M10's sensor capabilities convinced me to sell the M240 and get the M10 http://www.artphotoacademy.com/leica-m10-preview/

[M11 for me]

I am missing the GPS tagging in mine. Just as much in the direction of accuracy of information in that article.

[Rollin]

It seems likely that every model since the M240 has substantial noise reduction at high ISOs applied to the RAW files - it was explicit in some of the block-diagrams for the Maestro processor, and it is obvious if you compare equivalent high-ISO shots vs pushed low-ISO. Similarly, the "green shadows" problem was probably the result of trying to mask pattern noise by manipulating the sensor data prior to writing the DNGs. And all M-series CMOS cameras have what appear to be fairly arbitrary limitations on long exposures and no ability to disable dark-frame subtraction.

 

Much of this is, unfortunately, common practise in the camera industry. It is an attempt to make the files out-of-camera appear as good as possible "before" processing. Aside from dark-frame subtraction (which is useful in most cases, though it should still be optional), generally the less in-camera processing that is made to a RAW file the better.

 

M10 is my first Leica - I'm really appreciating the lower NR on the jpgs vs. other manufactures that leave the images mushy, but do wish Leica would have included a High, Normal and Low options for jpg NR like others.

 

Loving how the DNG's are full of detail, if it even means noise is more apparent. However, I was surprised to find luminance NR applied in LR automatically on the DNG's, so had to change the profile for import to set that to 0.

Fuji is known to cook in NR on their RAW's and partly why they do well in comparisons sometimes, but you lose detail. I agree, Leica's approach is much appreciated.

[jaapv]

Advertisement (gone after registration)

It seems likely that every model since the M240 has substantial noise reduction at high ISOs applied to the RAW files - it was explicit in some of the block-diagrams for the Maestro processor, and it is obvious if you compare equivalent high-ISO shots vs pushed low-ISO. Similarly, the "green shadows" problem was probably the result of trying to mask pattern noise by manipulating the sensor data prior to writing the DNGs. And all M-series CMOS cameras have what appear to be fairly arbitrary limitations on long exposures and no ability to disable dark-frame subtraction.

 

Much of this is, unfortunately, common practise in the camera industry. It is an attempt to make the files out-of-camera appear as good as possible "before" processing. Aside from dark-frame subtraction (which is useful in most cases, though it should still be optional), generally the less in-camera processing that is made to a RAW file the better.

The main "trick" used by Leica to reduce out-of-camera noise is shifting the black point. The overall noise reduction is quite modest on all their cameras, compared to some other brands.

[Chaemono]

Loving how the DNG's are full of detail, ...

 

Before I decided to get the M10, I did a quick comparison with the X1D here: https://www.smugmug.com/gallery/n-9FSKSS/. I just loved the detail the Summicrons stopped down a bit showed in the DNGs. I'll do α7R III + 35/2.8 Sonnar FE vs. M10 + 35 Summicron-M next.

[Mark II]

The main "trick" used by Leica to reduce out-of-camera noise is shifting the black point. The overall noise reduction is quite modest on all their cameras, compared to some other brands.

Over the holidays I shot a series of images with an M262 to try to capture the Christmas lights using an ISO-less shooting style to preserve colour in the highlights (basically expose for the highlights at a low ISO and then push the shadows in post). For some shots I had both ISO-less (pushed) and normal exposures. The pushed shots show considerably more detail, as well as more noise. The ISO 3200 shots look visually as if they have been low pass filtered (I have not tried an FFT to check the spectrum). There are a lot of hot pixels in the pushed versions, but all of this cleans up nicely using the NR tools in Capture One.

 

Incidentally, making such comparisons with any raw developer is fraught with difficulty. The same NR setting in CO10 produces a different degree of NR and extra sharpening (!) depending on both the camera and the image EXIF ISO rating, so you have to use a raw converter that does not do this when making comparisons.

 

Ultimately, the experiment failed because of the pain involved to correct the green shadows. But if Leica would avoid messing with the raw pixel data an iso-less shooting style ought to work very well in high contrast situations.

 

FWIW, the DXO results and some experiments with DNG files have pretty much killed any desire for me to upgrade to an M10. With equivalent processing (in camera or out) the only advantage I can see is about 1/2 stop more DR above ISO 800. For me, this is not enough to counter the pain caused by the lower capacity batteries when travelling (changing batteries is not a problem - but recharging 3 or 4 batteries overnight is!).

[Chaemono]

All this negativity in this thread. I read that negative thoughts increase the stress hormones in our body.

[xiaubauu2009]

Well, it's fair to say that the tech in Leica's camera sensor is a bit lagging behind, but their parameters are measured differently and they are design for a purpose of using the M mount lens that has a certain requirement to them that only M can provide....

So I think people don't buy M for all kind of purposes, it's either they get/love/understand it or they don't.

 

DXO is just using this as a sensationalized topic....

[tomasis7]

Well, it's fair to say that the tech in Leica's camera sensor is a bit lagging behind, but their parameters are measured differently and they are design for a purpose of using the M mount lens that has a certain requirement to them that only M can provide....

So I think people don't buy M for all kind of purposes, it's either they get/love/understand it or they don't.

 

DXO is just using this as a sensationalized topic....

+1

[Mr Fjeld]

If the sensor is the main reason for those looking to buy an expensive camera, then I wish them luck. Obviously, Leica M is not a suitable choice for the sensor fetishists so they should sell their Leica M and buy a Nikon. I believe that for my needs the sensor in the M is probably good enough, and that should be sufficient for photographers wanting a range finder camera. 

Edited January 10, 2018 by Mr Fjeld

[pgh]

There seems to be some cognitive dissonance happening here.

There are those that say the sensor is fine, and it is, it is fine. In most circumstances it will have nothing to do with whether or not your picture is successful. It will boil down to those hard/impossible to buy things like curiosity, timing, skill, etc etc.

 

BUT this is a $7k camera. I don't think there's any justifying a just fine sensor in a camera that is more than twice any of the models that it competes with. I can rightly and fairly say that Leica has a subpar sensor given what you pay for, and this opinion was the case before any Dx0 score came out. That just merely confirmed impressions that the Leica sensor is lagging behind current offerings - and for some types of imagery, this definitely DOES matter for some photographers. I love the M10 - it feels like home after years of using my M6's and a never quite right M8. But the sensor is the weak spot. I wish it was higher resolution, I wish it held tonal ranges better. The low light thing for me - it's fine in that area, but of course I wouldn't be mad at an improvement here. I'm fine with more detail preferred over lower noise. Working with the highlights and the shadows in the M10 files is more of a chore than in any camera I've had in recent memory...heck, I even recently opened a few X pro 1 raw files and was surprised to see the highlights and shadows in this camera were even more malleable than in the M10. The sensor is disappointing. The camera itself, when it works, is a great piece of equipment. People like to talk about cars. Well, it's like a nice car with the engine of a Honda Civic. The Honda will do the job most of the time anyways, so it's fine. 

[adan]

Pop quiz - when was the last point in history where paying for a Leica ever bought one cutting-edge camera technology (as opposed to lens technology)? Probably 1954 (multi-focal-length M3 viewfinder).

 

FWIW, the DXO results and some experiments with DNG files have pretty much killed any desire for me to upgrade to an M10. With equivalent processing (in camera or out) the only advantage I can see is about 1/2 stop more DR above ISO 800. For me, this is not enough to counter the pain caused by the lower capacity batteries when travelling (changing batteries is not a problem - but recharging 3 or 4 batteries overnight is!).

 

Using those parameters, a logical decision. I skipped the M240 upgrade from an M9 because the absolute ISO increase without banding was only 2/3rds stop.

 

But I'd say the M10 has at least a 1.5-stop advantage over the M240 in where shadow banding appears (ISO 12800 vs ISO 4000). Which, as I mentioned, DxOMark fails to measure.

Edited January 10, 2018 by adan

[pgk]

Well, it's fair to say that the tech in Leica's camera sensor is a bit lagging behind, but their parameters are measured differently and they are design for a purpose of using the M mount lens that has a certain requirement to them that only M can provide....

 

The question is whether it is lagging behind other cameras which can utilise Leica M mount wide-angles as well as the Leica can?

 

The answer isn't as simple as ascribing a score or indeed utilising various graphs, unless of course they take Leica's wide-angle lens requirements into account.

[Steve McGarrett]

Pop quiz - when was the last point in history where paying for a Leica ever bought one cutting-edge camera technology (as opposed to lens technology)? Probably 1954 (multi-focal-length M3 viewfinder).

 

Well, I'd say 2015 (4.4Mp built-in EVF...)

[FlashGordonPhotography]

There seems to be some cognitive dissonance happening here.

There are those that say the sensor is fine, and it is, it is fine. In most circumstances it will have nothing to do with whether or not your picture is successful. It will boil down to those hard/impossible to buy things like curiosity, timing, skill, etc etc.

 

BUT this is a $7k camera. I don't think there's any justifying a just fine sensor in a camera that is more than twice any of the models that it competes with. I can rightly and fairly say that Leica has a subpar sensor given what you pay for, and this opinion was the case before any Dx0 score came out. That just merely confirmed impressions that the Leica sensor is lagging behind current offerings - and for some types of imagery, this definitely DOES matter for some photographers. I love the M10 - it feels like home after years of using my M6's and a never quite right M8. But the sensor is the weak spot. I wish it was higher resolution, I wish it held tonal ranges better. The low light thing for me - it's fine in that area, but of course I wouldn't be mad at an improvement here. I'm fine with more detail preferred over lower noise. Working with the highlights and the shadows in the M10 files is more of a chore than in any camera I've had in recent memory...heck, I even recently opened a few X pro 1 raw files and was surprised to see the highlights and shadows in this camera were even more malleable than in the M10. The sensor is disappointing. The camera itself, when it works, is a great piece of equipment. People like to talk about cars. Well, it's like a nice car with the engine of a Honda Civic. The Honda will do the job most of the time anyways, so it's fine. 

 

Price is not an indicator of sensor performance. The most expensive Nikon and Canon bodies have sensors remarkably close to the M10 and with lower scores than mid priced models. Comparing on price alone the M10 does quite well and is within line of the 1DX2 and D5. If Leica's flagship has a sub par sensor then we need to say the same thing about the D5.

 

It's already been stated that Leica couldn't get the microlenses they needed from Sony. Does anyone besides Sony make a commercially available sensor above 25MP? Is it currently possible to make a 36MP sensor that works in the corners with M lenses? What are the impacts on camera shake and usability with a 42MP sensor (the M10 isn't exactly known as a tripod camera)? How many of us actually print M10 files bigger than 24"? Does a 42MP Sony sensor generate more or less heat than the one in the M10?

 

All cameras have compromises. The M10 has bigger issues than sensor performance. If the sensor is that important you can get the huge D850 or the complex A7R3.

 

This incessant whinging is getting depressing. The M10 is what it is. It doesn't make worse pictures than  it did before the scores went up. Most of us will get no where near it's potential.

 

Gordon

[Advertisement]

Hallo,

Look here: Leica M

[Mark II]

But I'd say the M10 has at least a 1.5-stop advantage over the M240 in where shadow banding appears (ISO 12800 vs ISO 4000). Which, as I mentioned, DxOMark fails to measure.

Yes, DXO ignores banding, even though it is often a seriously problem.

 

I suspect that the banding was at least partially addressed in the M 262 series - while there is some pattern noise it is much much less than I expected based on reports here (and it is negligible compared to my old Canon 5D III). Or maybe I just got lucky.

 

There is nothing inherently wrong with any of these sensors. Understanding how they behave is both interesting from a technical perspective, and also useful for understanding how to get the best results from a given setup. Understanding the hardware, firmware and software is not fundamentally different than exploring the merits of different films, papers and chemicals...

[rrdiaz30]

...Understanding the hardware, firmware and software is not fundamentally different than exploring the merits of different films, papers and chemicals...

 

 

Well said!