High ISO - M10 vs Q

[Chaemono]

Advertisement (gone after registration)

Oh, come on now. Where are the greys in the Sony picture?

[indergaard]

Are you using the Adobe Color profile for both the A7rIII and the M10? It doesn't look like it. The M10 profile lifts the shadows by default, which seems to be happening. So it might appear that the shadows are more open. But it's just a LUT manipulating the raw data. My A7rIII shows considerably more open shadows than my M10 does, regardless of the scene or lens used, if used with the exact same profile (Adobe Standard or Adobe Color in LR, or the default profiles in Capture One 11).

Edited September 13, 2018 by indergaard

[Chaemono]

Of course, I am. And unlike your Sony and M10 pictures, I show mine. I'll do tons of them now, just because you insist. And I'll keep posting the link.

[indergaard]

Of course, I am. And unlike your Sony and M10 pictures, I show mine. I'll do tons of them now, just because you insist. And I'll keep posting the link.

 

I didn't insist on anything. And I have previously shown several images highlighting the ISO variance between the metering of the M10 and A7rIII. There is also a big variance between the metering of the M10 and Q, M240, MM (v1) and M9-P, all of which I have owned. There are tons of posts on this forum, and many others, where a lot of people have discovered that the M10 is using a different method of calculating ISO values than previous Leica cameras.

 

But I think everyone would like to either see totally unaltered versions (no WB fix, no exposure adjustment, no sharpening, NR, no manufacturer profile applied, etc), or get access to the raw files.

 

Not a single image from my A7rIII shows harsher blacks than my M10 does, regardless of the scene. In fact, the M10 is known to produce far harsher blacks than the M240 (which is also my own experience). So your example images makes no sense, unless there is a profile or defaults that are making adjustments.

 

Looking forward to download your raw files. I haven't bothered with an M adapter for my Sony this time, as it makes no sense. 

[jdlaing]

A lot of things make no sense.

Edited September 14, 2018 by jdlaing

[Chaemono]

Raw files of the last two pairs posted earlier.

 

α7R III + APO 50 Summicron-M ISO 200 f/2.0 @1/250 sec.

https://cc2032.zenfolio.com/img/g47598086-o750076470.dat?dl=2&tk=BiQf4Ya9TxmekbsOxU5Fg68P1aey3vFO1UREEeD33Lg=

 

M10 + APO 50 Summicron-M ISO 200 f/2.0 @1/250 sec.

https://cc2032.zenfolio.com/img/g495210704-o750076470.dat?dl=2&tk=fOmRRZaepf1rfLPqFnnPMGKB_25xKPsLLMs3yU7emPA=

 

 

At ISO 640 one needs to lift the shadows less with the M10 file to see details. M10 here superior IMO.

 

α7R III + APO 50 Summicron-M ISO 640 f/2.0 @1/500 sec.

https://cc2032.zenfolio.com/img/g15294270-o750076470.dat?dl=2&tk=5n041xbWn56x2MzKO3_2l2NP5EVZlyqf3Cy0_WENy4w=

 

M10 + APO 50 Summicron-M ISO 640 f/2.0 @1/500 sec.

https://cc2032.zenfolio.com/img/g9556073-o750076470.dat?dl=2&tk=N2O4szIvj5xyyCLE6HnKQdJK_E0iXT-OBN9xrbYUhfY=

Edited September 14, 2018 by Chaemono

[Chaemono]

Advertisement (gone after registration)

I kind of like this 'please, let me see the raw files.' Makes life a lot easier for everyone. I will use the same lens on both cameras over the next few weeks and post the pictures here. Until then links to raw files of one pair I posted earlier in the year. Now, here we are talking the incredible 35 Summicron-M (so small and light) on the M10 and the Sonnar FE 35/2.8 on the α7R III. The Sony lens is, of course, no match for the Leica lens, but just for the heck of it.

 

Raw files

α7R III + Sonnar FE 35/2.8 ISO 640 f/4.0 @1/60 sec.

https://cc2032.zenfolio.com/img/g134996006-o750076470.dat?dl=2&tk=CVA4PSeQaCsyRiVpEWm8yOQueCIZYNZ2NqYADKXDK7k=

Do you like the size of the file? Well, there is 81.6 MB of information in there.

 

M10 + 35 Summicron-M ISO 640 f/4.0 @1/60 sec.

https://cc2032.zenfolio.com/img/g14018288-o750076470.dat?dl=2&tk=sUMAv2KiFyiLzAJUwh0AcAg7ub3WftfMsSTzPjkdq5E=

23.4 MB of information only.

[adan]

BTW - earlier (post #73) I mentioned:

 

 

Some firmware maven early on (sandymc?) reported that he found many things in the M10 firmware and EXIF, at least in the beginning, that were "sloppy and rushed," and this may be another such.

 

I just ran across that original source, in researching a different M10 technical question...

 

http://chromasoft.blogspot.com/2017/01/leica-m10-raw-file-dng-analysis.html

[Chaemono]

Okay, Raw lovers, two more from today. There is mixed light outside so I have to find places indoors with no daylight. Here, the 81.7 MB of information in the Sony ARW file are paying off. Not because of shadow recovery, the M10 is doing great there, but because of detail retention in the highlights. In the upper left hand corner, if one puts Highlights at -32 in LR for both, one can actually read "BLASTER GUN WITH CUP" in the Sony picture. Amazing. This is why the α7R III sensor needs lenses like the 55 Otus, or Leica M lenses, or even the FE 55/1.8 which is very sharp corner to corner but has some CA/PF issues.

 

Raw files

α7R III + 35 Summicron-M ISO 640 f/2.0 @1/125 sec.

https://cc2032.zenfolio.com/img/g622476266-o750076470.dat?dl=2&tk=mv-aHSfV5pNzidna19-XIJQQWBQaRZYzY7uXi2m286k=

M10 + 35 Summicron-M ISO 640 f/2.0 @1/125 sec.

https://cc2032.zenfolio.com/img/g998459165-o750076470.dat?dl=2&tk=KCYOUqXtO9xoJLXhNJ46hbuH2bU-VdpYdV17v9FItYI=

Edited September 15, 2018 by Chaemono

[Guest Nowhereman]

What does this have to do with the discussion in this thread? We're not talking about resolution.

 

By the way, yesterday after transferring DNG files from the M10 to the Lightroom, I took a few pictures in my study at ISO 200 — just to remind myself how the M10 meters and renders, because I haven't shot with it since May.  As the wall was Zone 5 grey, I metered for that, but intentionally underexposed by ½ stop: there was a window in the left side of the frame, and the late afternoon light, on an overcast day, was neither bright nor contrasty. The file the M10 produced managed to blow some highlights in the the window part of the frame. This is what I mean about the M10 having an annoying flaw: this image would not have blown highlights on an M9 or an MM. The annoying part is that it seems unpredictable when, and to what degree, the highlights are going to blow out — although someone who has shot with the M10 more than I, or with more testing, may be able to provide some explanation of the parameters that allow more predictability.

_______________

Alone in Bangkok essay on BURN Magazine

Nowhereman Instagram

 

[Chaemono]

‘Please, stop posting links to Raws. The M10 has weird sensitometry because, because I said so.’ Pathetic. Tons more of Raw pairs to come dude.

[Chaemono]

How about the claim that the M10 tends to blow highlights and has weird 'sensitometry?' Well, it ain't so. 

 

Raw files

α7R III + 35 Summicron-M ISO 200 f/2.0 @1/1000 sec.

https://cc2032.zenfolio.com/img/g934961501-o750076470.dat?dl=2&tk=di8FhYVciifIZ124ks6mjy0PtH3v3TlWXqN07B330zk=

 

M10 + 35 Summicron-M ISO 200 f/2.0 @1/1000 sec.

https://cc2032.zenfolio.com/img/g896029428-o750076470.dat?dl=2&tk=fdA5wMCZiPrL2WLiAoxvoiYyk-G7S5GfU97yDjmBUEA=

 

 

But what is clear is that the Sony definitely retains more details in the highlights that can be recovered by moving the slider in LR to the left. 

 

Raw files

α7R III + 35 Summicron-M ISO 640 f/2.0 @1/125 sec.

https://cc2032.zenfolio.com/img/g876439471-o750076470.dat?dl=2&tk=i2JFJaLARS-NEDTyPNqNBMFE55UzQWCnKfaEeT-DEfk=

 

M10 + 35 Summicron-M ISO 640 f/2.0 @1/125 sec.

https://cc2032.zenfolio.com/img/g855750261-o750076470.dat?dl=2&tk=r8nSAIKEYZHOqAEF20xxkvgyPTcUwMcjeF24ZkQ_tm4=

[Chaemono]

Here, again, I much prefer the way the M10 exposes and am not sure that the Sony would have required faster shutter speed or lower ISO.

 

Raw files

α7R III + 35 Summicron-M ISO 200 f/2.0 @1/2000 sec.

https://cc2032.zenfolio.com/img/g909906693-o750076470.dat?dl=2&tk=ucyiWuGwsLEIctlhEUUJxnqYAi6sEhNMRyNbBjs8iRg=

 

M10 + 35 Summicron-M ISO 200 f/2.0 @1/2000 sec.

https://cc2032.zenfolio.com/img/g766653911-o750076470.dat?dl=2&tk=4w0ZWX5PFC_LMt3p0fL_UXSrVk4dg-ujc3exodMp4Uo=

 

 

AWB by the cameras. Love the M10 output.

 

Raw files

α7R III + 35 Summicron-M ISO 640 f/2.0 @1/125 sec.

https://cc2032.zenfolio.com/img/g759490348-o750076470.dat?dl=2&tk=dkxVLeKerANCV8uttTfrlXJ1HIMgWbGyLD00O5RDOFs=

 

M10 + 35 Summicron-M ISO 640 f/2.0 @1/125 sec.

https://cc2032.zenfolio.com/img/g878065960-o750076470.dat?dl=2&tk=s-i6fqiCnlfOGa-TV1aNfh46BuYfWBuso-p2kOa7KN4=

Edited September 15, 2018 by Chaemono

[indergaard]

Wow, thanks for providing a ton of data to examine Chaemono. I really appreciate that. Looking forward to look at your samples!

 

Who knows, maybe there is a variance in the M10's meter causing the perceived "issue". I love my M10, but it's still just a camera, and will be compared with other cameras equally. Which I have done. And even with it's shortcomings it's still my favorite camera to shoot with. But my A7rIII provides more usable dynamic range and resolution (when paired with good lenses). I also prefer the A7rIII for typical tripod and landscape work, as it's just so much easier to work with.

[Chaemono]

You’re welcome, indergaard, but I’m not done yet. And, whenever BS claims about the M10 appear, this is the third thread now, I will repost the links to the Raw files. My aim is to get to about 20 to 30 pairs to counter the BS.

[adan]

As the wall was Zone 5 grey, I metered for that, but intentionally underexposed by ½ stop: there was a window in the left side of the frame, and the late afternoon light, on an overcast day, was neither bright nor contrasty. The file the M10 produced managed to blow some highlights in the the window part of the frame. This is what I mean about the M10 having an annoying flaw: this image would not have blown highlights on an M9 or an MM.

 

Well, as you know, we agree on this aspect in general. But how do you actually know the M9 would not have blown that particular overcast in that particular window - and with that particular lens?

 

I go back through my pix over the years, and can find up to 3-stop differences between "overcast day" exposures - all depends on the thickness of the clouds: rather thin stratus vs. a 15-km-deep supercell. But all count as "overcast."

 

There is almost no worse light-meter than the human eye. It accomodates like crazy (both with a variable aperture, and in the nervous system). And the one worse light-meter is - human memory.

 

How do you know the wall was really Zone V gray ± a half stop? Pop quiz - which (if any) of the gray patches below really is/are a Zone V gray?

 

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!

 

So let's check the M10's actual highlight latitude or "headroom". First we recheck the M10 "classic" semi-spot metering. First image shows the metering area, on a gray-card/density scale. Second image shows the metered area average-filtered in Photoshop, (all pixels within the meter averaged together). And hey-presto, it comes out just fractionally (+0.1 stop) brighter than the correct value for "middle gray/Zone V." That is, BTW, a hint for the pop-quiz above.

 

The M10 meter will average whatever falls within its limited area as a medium-gray brightness  - as all reflected meters should. BTW - in all these shots the camera was set to zero EC and ISO 200, and only default "0" settings were used for processing the images (except for the final check for recoverable highlights). Lens was 35mm Summarit for the first meter check, and 135 Tele-Elmar last version (1995) for the car pix.

 

 

Now we will meter something, and then overexpose it progressively (bracket up) until we get a blown white. We are looking at the white car, and particularly a point (marked, faintly) between the door handle and driver's mirror, which in the metered exposure comes out exactly gray-card gray. At +4 stops of exposure, that same point hits 254 on the digital brightness scale, and the highlights cannot be recovered. They can be made "grayer and darker," but the brighter sun-reflection from the mirror is no longer distinguishable.

 

 

Now, as a back check on that, we look at the exposure at 3.5 stops over - and in that image, the car "white" of 252 is just recoverable, to reveal the bright reflection patch (255) as a different tone (recovered, those values are 213 and 235).

 

 

Therefore, we can say that M10 highlight headroom is at least 3.5 stops brighter than medium gray (recoverable) but not 4 stops brighter (not recoverable). Put another way, meter the brightest near-white where some detail is desired, and open up 3.5 stops, and there will be recoverable details in that highlight.

 

I'll toss another monkey into the wrench - lens contrast. Here are two pictures with two lenses, my old beloved 1983 21mm Elmarit non-ASPH, and a bright shiny 35 Summarit v.1. Whole frames (just for reference) plus a troublesome highlight detail. Exact same exposure (ISO 200, f/5.6, 1/180th, EC -0.67), exact same Camera Raw settings. 35mm image resized in detail to match the 21 size.

 

It's pretty obvious a modern, contrasty 35 blows the highlights more than the 35-year-old 21, all other things being equal. More bald white area. I don't know where the 28 Summaron falls as to gross (as opposed to edge) contrast. But Erwin Puts notes the original has "high overall contrast," and that's with the 1950s coatings. Nor, for that matter, do I know what that lens would have done on an M9, for example - might have blown highlights equally.

 

 

 

 

[jaapv]

Let's go back to our reliable DR source: photonstophotons, which gives us the usable dynamic range (as opposed to DXO, which is the measurable DR, less relevant) The jumps indicate digital manipulation of the response curve, the shadows get noise reduced.

 

The M10 curve has no such trickery. Disregarding electronic enhancement, the dynamic ranges are virtually identical.  

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!

[Mark II]

Let's go back to our reliable DR source: photonstophotons, which gives us the usable dynamic range (as opposed to DXO, which is the measurable DR, less relevant) The jumps indicate digital manipulation of the response curve, the shadows get noise reduced.

 

The M10 curve has no such trickery. Disregarding electronic enhancement, the dynamic ranges are virtually identical.  

 

 

I believe that the problem with the Photons to Photons chart is that the ISO axis uses the value stated by the manufacturer. Because of the scope for interpretation of ISO by the manufacturer you can not compare the absolute values of the curves without taking this in to account (try comparing the various Leicas to an Olympus E-M1.2 for example - I think that one of the E-M1.2 charts shows the camera exceeding the theoretical performance for a u4/3 sensor!).

 

What is more useful from the Photons to Photons analysis is that the M10 sensor is more or less ISO invariant above ISO800 - so if the M10 tends to underexpose that is probably a good thing. Heck, if Leica did not mess with the RAW sensor data, the best strategy would to always limit the max ISO to ~800 and lift exposure when post processing - just like pushing film.

 

Frankly, I find all this high ISO angst pointless. In really low light you are limited by the light itself. All modern sensors are astonishingly good at high ISO and very close to the theoretical performance limits for Bayer sensors. But while the scope for high-ISO improvements from newer sensors is limited, low ISO is another story...

[indergaard]

Let's go back to our reliable DR source: photonstophotons, which gives us the usable dynamic range (as opposed to DXO, which is the measurable DR, less relevant) The jumps indicate digital manipulation of the response curve, the shadows get noise reduced.

 

The M10 curve has no such trickery. Disregarding electronic enhancement, the dynamic ranges are virtually identical.  

 

 

The jump you are seeing on the Sony at ISO 640 is the dual conversion gain stage activating. This means that at ISO 640 the camera switches to a secondary high gain amp/capacitor, which in effect gives higher dynamic range at higher ISO's. Two completely separate capacitors are used for low-gain (ISO 100-500) and high-gain (ISO 640-25600).

 

For example, it is better to shoot the A7rIII at ISO640 than ISO 200-500. ISO 640 will give better dynamic range. If shooting above ISO 100 on the A7rII and A7rIII, it is recommended to switch directly from ISO 100 to ISO 640. Anyone shooting the A7rIII at ISO 200-500 is actually getting higher noise and lower DR than shooting at ISO 640.

 

This is not digital manipulation or trickery. It is a licensed technology that was developed by Aptina (a sensor R&D and manufacturer), which PhotonsToPhotos hosts a white-paper of, which goes in depth to describe how it works and the real advantages of using it: http://www.photonstophotos.net/Aptina/DR-Pix_WhitePaper.pdf

 

Sony and Fujifilm has incorporated this technology in their cameras for a few years now. Maybe others too.

 

Think of it as the camera/sensor having two separate base ISO's: ISO 100 and ISO 640. Which is also why the read noise goes down at ISO 640 as seen here:

 

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!

 

So if anyone is shooting their A7rII or A7rIII at ISO 250-500, they are just getting worse DR and more noise than if they just shot at ISO 640.

Edited September 16, 2018 by indergaard

[Advertisement]

Hallo,

Look here: Leica M

[jaapv]

I meant the peak at ISO 125. preceded by a pull- up-and-down  -which is not on the Riii- and the noise reduction at the low end..
 
Leica uses dual-gain as well, according to Bill Claff (AKA Photonstophotons) in a mail to me:
 

Jaap,
Regarding the SL and the Q; they appear to be using dual conversion gain
technology.
This gives a performance boost at ISO 200 and ISO 400 respectively.

 

Either Leica did not use it in the M10, or they implemented it in a way that smooths out the graph.