Comparison of Leica M11, Sony a1, and Sony a7r4 at ISO 12,500

[onasj]

Advertisement (gone after registration)

A quick test:

Leica M11, 50 APO, f/2, ISO 12500

Sony a1, 50 GM, f/2, ISO 12800 (scaled up to the same image size of 9,528 pixels wide in Photoshop 2021)

Sony a7r4, 50 GM, f/2, ISO 12800

RAW files were opened in Photoshop 2021, auto corrected, then enlarged to 100% and screencapped.

 

First the a1:

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!

 

Then the a7r4:

 

And finally the M11:

 

Findings:

- Colors with the M11 are indeed rendered differently—more accurately and more vibrantly—than with the M10/-P/-R, which has suffered a bit shooting this test scene in the past at high ISOs.

- The high ISO performance of M11 is outstanding, with less luminance and chroma noise than the Sonys, at least with this workflow. (I couldn't use Capture One 20 because the color profile for the M11 image ended up very green, which I assume will be fixed over time as the M11 profiles are released).

- Wow that Sony 50/1.2 GM is sharp and very well corrected.  It is actually sharper and more apochromatic at f/2 than Leica 50 APO (which I realize sounds like heresy here, but which has been confirmed by others too). Of course it's a giant lens, it's stopped down here, and it's not nearly as fun to use, but it's an optical masterpiece.

- Overall, I'm pleasantly surprised that Leica seems to have put a fairly state-of-the-art sensor into the M11, rather than their past history of being about a generation behind in sensor releases. Of course the a1 has, effectively, a global shutter than can churn out 20-30 RAW frames per second at 50 MP, but in terms of single image quality, M11 owners have a sensor that is truly world-class, as of now at least.

 

RAW files:

a1: https://www.dropbox.com/s/xqh6urb2kdih9zw/Sony a1 ISO 12800 50 GM f2DSC00449.ARW?dl=0

a7r4: https://www.dropbox.com/s/h1yqvvl3b3zm666/Sony a7r4 ISO 12800 50 GM f2 DSC07838.ARW?dl=0

M11: https://www.dropbox.com/s/ns52lftfg83t91y/Leica M11 ISO 12500 50 APO f2 L1000105.DNG?dl=0

 

Edited January 14, 2022 by onasj

[Steve Ricoh]

Rumour has it the sensor in the M11 is from Sony; emphasis on rumour.

[Advertisement]

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

[onasj]

9 minutes ago, Steve Ricoh said:

Rumour has it the sensor in the M11 is from Sony; emphasis on rumour.

I suspect the sensor has to be from Sony, because no one else in the world is making a ~60 MP BSI sensor and one doesn't just set up such a capability without the industry knowing.  Sony provides at least the starting sensors for many other brands' cameras.

[1400700226]

Could you please share the dng comparison of the full frame 60 36 18mp mode? Thanks so much

[pgk]

Look like three very capable cameras to me.

[onasj]

5 hours ago, pgk said:

Look like three very capable cameras to me.

They are definitely three outstanding cameras. 

[gotium]

Advertisement (gone after registration)

less noise in the M11 photo, but don’t you think also less detail? as in some NR “baked” into the RAW file?

 

Also there’s blue in place of black in much of the M11 image. Which is the real color?

Edited January 15, 2022 by gotium

[Chaemono]

vor 17 Stunden schrieb onasj:

Findings:

There is false color in the M11 picture. It must have to do with the way the DNG is rendered. It could make sense to wait until Adobe/ACR supports the camera with it own profiles before doing the comparisons. I find that once LR supports a new camera these false colors disappear. PS and LR are both using ACR, I believe. 

Edit - I just opened the files in LR. There is Adobe support for the M11 already. There must be NR baked into the M11 DNGs. I turned off all NR in LR. The rendering of the DNG has less noise than the Sony files. With color NR set to zero, they all show a lot of color noise but the Sony files visibly more. However, ACR seems to do a much better job at rendering the Sony RAW files with color NR at +25. The M11 image turns blue with color NR set to +25. It looks to me that it's an issue of NR being baked into the DNGs and the way color NR is then applied by ACR to the M11 files.

Edited January 15, 2022 by Chaemono

[Al Brown]

So to summarize, Leica M11 is on par with A1 or A7R4 or slightly better at high ISO.

[opera207]

Yes I have sony 50gm and m50apo, in terms of optical, 50gm is better.

[Chaemono]

Sony files look sharper but it’s a different lens. Also, the APO 50 Summicron-M can very a bit in sharpness. It’s tough to make such a small lens that is equally sharp for all copies. I used to have two. One was visibly softer at 100%. I sold it. The copy that I kept is ultra sharp wide open.

[Chaemono]

vor 5 Minuten schrieb opera207:

Yes I have sony 50gm and m50apo, in terms of optical, 50gm is better.

I agree in these. I bet my copy of the APO could match your Sony 50gm in terms of sharpness wide open.

Edit - Wait, did he stop down the Sony f/1.2 to f/2? That’s not fair then, not even for color detail.

Edited January 15, 2022 by Chaemono

[Chaemono]

Am 14.1.2022 um 15:55 schrieb onasj:

A quick test:

Leica M11, 50 APO, f/2, ISO 12500

Sony a1, 50 GM, f/2, ISO 12800 (scaled up to the same image size of 9,528 pixels wide in Photoshop 2021)

Sony a7r4, 50 GM, f/2, ISO 12800

Did you stop down the Sony lens?

[John Smith]

8 hours ago, Chaemono said:

There is false color in the M11 picture. It must have to do with the way the DNG is rendered. It could make sense to wait until Adobe/ACR supports the camera with it own profiles before doing the comparisons. I find that once LR supports a new camera these false colors disappear. PS and LR are both using ACR, I believe. 

Edit - I just opened the files in LR. There is Adobe support for the M11 already. There must be NR baked into the M11 DNGs. I turned off all NR in LR. The rendering of the DNG has less noise than the Sony files. With color NR set to zero, they all show a lot of color noise but the Sony files visibly more. However, ACR seems to do a much better job at rendering the Sony RAW files with color NR at +25. The M11 image turns blue with color NR set to +25. It looks to me that it's an issue of NR being baked into the DNGs and the way color NR is then applied by ACR to the M11 files.

Isn't this what Panasonic did with the S1R, bake noise reduction into its DNGs? And that's why DXO ranked it at 100? Could that be what @Oswalthas discovered with pushing the files (Image thread)? 

[Chaemono]

vor 5 Minuten schrieb John Smith:

Isn't this what Panasonic did with the S1R, bake noise reduction into its DNGs? And that's why DXO ranked it at 100? Could that be what @Oswalthas discovered with pushing the files (Image thread)? 

I just went back to Oswalt’s post. Exactly. The S1R applies NR to base ISO 100 and it got a much better rating at DXO than the SL2 with the same sensor and more malleable files. 

[SrMi]

1 hour ago, John Smith said:

Isn't this what Panasonic did with the S1R, bake noise reduction into its DNGs? And that's why DXO ranked it at 100? Could that be what @Oswalthas discovered with pushing the files (Image thread)? 

S1R applies noise reduction up to ISO 200 and then from 6400 on. Canon R5 applies noise reduction up to ISO 600. Canon R3 applies noise reduction all the time. We need measurements by PhotonsToPhotos to confirm that Leica applies NR to M11.

In-camera NR may or may not be damaging, though I prefer to do it in post.

[Stevejack]

8 hours ago, opera207 said:

Yes I have sony 50gm and m50apo, in terms of optical, 50gm is better.

I’ve found this too. I mean the Leica is tiny, so it’s in no way fair fight, but wow is that GM a good lens for the price. 

[Rollin]

@onasj Thank you for doing this test, posting and writing some conclusions as high ISO is important to me. M10 was defiantly a generation, or more, behind. I saw odd artifacts as low as ISO 8,000. Are you finding the M11 solid to 25,000 or so?

At one point there was a rumor the M11 had the same sensor as Sigma's Fp L's.

[onasj]

7 hours ago, Rollin said:

@onasj Thank you for doing this test, posting and writing some conclusions as high ISO is important to me. M10 was defiantly a generation, or more, behind. I saw odd artifacts as low as ISO 8,000. Are you finding the M11 solid to 25,000 or so?

At one point there was a rumor the M11 had the same sensor as Sigma's Fp L's.

Yes, in terms of practical usefulness I think 25,000 is usable and possibly a bit more depending on how dark the photo is and the viewing or printing size.

There are very few (one) known manufacturer that makes a ~60 MP BSI sensor so it is a good assumption that it began life in a Sony foundry, like the chip for the a7r4 and fp-L.

[Advertisement]

Hallo,

Look here: Interesting Leica gear

[SrMi]

7 hours ago, onasj said:

Yes, in terms of practical usefulness I think 25,000 is usable and possibly a bit more depending on how dark the photo is and the viewing or printing size.

There are very few (one) known manufacturer that makes a ~60 MP BSI sensor so it is a good assumption that it began life in a Sony foundry, like the chip for the a7r4 and fp-L.

@jdlaing wrote that he has a confidential source knowing that it is not a Sony sensor. We only know for sure that it is a BSI sensor. On the other hand, it does not matter who made the base silicon layer of the sensor. It matters what Leica does with it (sensor toppings, firmware).