M11 files – different resolutions have different crops

[hdmesa]

Advertisement (gone after registration)

After participating in the recent discussion of the M11 18mp files, I decided to test all three resolutions on my M11-D for more than just pixel level IQ (having previously tested IQ for landscape that way).

I took three shots in DNG & JPEG using these settings:

• Full manual
• Lens Detection set to Off / no manual lens profile applied
• Electronic shutter (to keep the shutter vibrations from affecting anything, though results were the same with the mechanical shutter)
• Remote shooting via the FOTOS app so that the camera was not touched during or between shots

Results viewed in Capture One (making sure no lens profile was applied to any of them):

There was one key difference between all three resolutions, and that is how much of the sensor each uses:

• 60mp – Slightly cropped compared to 36mp files
• 36mp – Uses the full available area of the sensor
• 18mp – Noticeably cropped compared to 36mp and 60mp files

Notes:

• The difference in image area between 60mp and 36mp is extremely minor, though I find it odd that 36mp shows more of the scene than 60. I would have expected the opposite.
• The difference in image area between 18mp and the other resolutions was notable. The 18mp images are noticeably cropped, though the crop amount is still small.
• DNG or JPEG, the results are the same.

Can someone else test this to confirm? I have not tested this by shooting straight from the camera, only by using FOTOS remote to both take the shots and to share them from the camera over to my MacBook.

[Advertisement]

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

[hmzimelka]

Yeah, I can confirm this with my M11.

Interestingly enough, when comparing the 36MP image with the 60MP image where I used "recovered edges", the comparison has identical coverage ...

However, receiver edges gains extra pixels for every DNG resolution setting. 

[hdmesa]

9 hours ago, hmzimelka said:

 

...recovered edges... receiver edges...

What program has those options? I don't see that in C1 or LRC. 

[hmzimelka]

1 hour ago, hdmesa said:

What program has those options? I don't see that in C1 or LRC. 

I use DNG Cleaner for Mac. There is, or used to be, a plugin for Lightroom Classic too, called RecoverEdges... 

https://www.mcguffogco.com/dngcleanermac.html

Edited November 1, 2024 by hmzimelka

[hdmesa]

47 minutes ago, hmzimelka said:

I use DNG Cleaner for Mac. There is, or used to be, a plugin for Lightroom Classic too, called RecoverEdges... 

https://www.mcguffogco.com/dngcleanermac.html

Ah, that makes sense because I don't think Capture One honors any crop data by default unless there's a mandatory correction flag as we see for the Q cameras.

[Photoworks]

this intrigued me, and I did a test myself.

I didn't use the fotos app, but capture one for iPad.

60mp and 36mp reproduced exactly the same image.

Furthermore, I have tested it on the SL3, exact same results.

[hdmesa]

Advertisement (gone after registration)

4 hours ago, Photoworks said:

this intrigued me, and I did a test myself.

I didn't use the fotos app, but capture one for iPad.

60mp and 36mp reproduced exactly the same image.

Furthermore, I have tested it on the SL3, exact same results.

Interesting. I wonder if the iPad version of C1 is somehow more consistent with regard to reading Leica's "recovered edges" instructions in the DNGs that @hmzimelka mentioned. Did you try 18mp? That's where the more noticeable difference was for me.

[SrMi]

In my book, it is a bug in the post-processor if the FOV differs from the one in the camera-created JPG. 

[hmzimelka]

16 hours ago, hdmesa said:

Ah, that makes sense because I don't think Capture One honors any crop data by default unless there's a mandatory correction flag as we see for the Q cameras.

Technically it's not crop data. All cameras, AFAIK, have an area of pixels that are always hidden from the resulting files. Usually, the gain of recovering edges is only 0.1 or 0.2 megapixels from 36-60MP cameras, so it's negligible. I think it has to do with preventing some rare interpolation artifacts as a result of the pixels being too close to the edge. I've noticed some cameras don't always produce the cleanest results in these recovered edges. 

I'm not sure what Leica is doing with the DNG files with their different size offerings, and whether what we observed is normal or not. 

I only use Lightroom Classic. Here I took some quick shots DNG+JPG, at the different sizes. Notice the pixel dimensions between Leica JPG and Lightroom files don't match. 

 

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 November 2, 2024 by hmzimelka

[SrMi]

55 minutes ago, hmzimelka said:

Technically it's not crop data. All cameras, AFAIK, have an area of pixels that are always hidden from the resulting files. Usually, the gain of recovering edges is only 0.1 or 0.2 megapixels from 36-60MP cameras, so it's negligible. I think it has to do with preventing some rare interpolation artifacts as a result of the pixels being too close to the edge. I've noticed some cameras don't always produce the cleanest results in these recovered edges. 

I'm not sure what Leica is doing with the DNG files with their different size offerings, and whether what we observed is normal or not. 

I only use Lightroom Classic. Here I took some quick shots DNG+JPG, at the different sizes. Notice the pixel dimensions between Leica JPG and Lightroom files don't match. 

 

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!

Interesting, but more importantly, the FOV is identical in all resolutions and formats.

[hmzimelka]

5 minutes ago, SrMi said:

Interesting, but more importantly, the FOV is identical in all resolutions and formats.

No, the JPGs do vary too.

Here are the Leica JPG files, the first from 18.2MP and the other at 60.1.

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 November 2, 2024 by hmzimelka

[SrMi]

9 minutes ago, hmzimelka said:

No, the JPGs do vary too.

Here are the Leica JPG files, the first from 18.2MP and the other at 60.1.

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!

 

You are right, though I would have not noticed it :). What about what you see in the EVF and LCD. Which resolution’s FOV does it match the closest?

[hdmesa]

52 minutes ago, SrMi said:

You are right, though I would have not noticed it :). What about what you see in the EVF and LCD. Which resolution’s FOV does it match the closest?

I only noticed the “issue” because I was getting more image area than what I was seeing in the EVF, so I imagine the Visoflex 2 is using the 18mp FOV. It was quite frustrating because I kept getting a small highlight over in one corner that I couldn’t see in the EVF. I’ll have to test it to be sure, or maybe @hmzimelka will beat me to it.

Edit: I only have an M11-D, so I can’t say if the LCD exactly mirrors the Visoflex 2.

Also, if you stack the 36 and 18mp images and switch back and forth between them like a GIF, the difference in FOV is a bit more dramatic than seeing the images side by side. I’d guess it’s the difference of about 2mm on a 50mm lens.

Edited November 2, 2024 by hdmesa

[Photoworks]

Yes, the Live View of the Visoflex frame is smaller than the sensor records, it is probably more dependent on the image stabilization in the viewfinder on the M11.

[hmzimelka]

1 hour ago, SrMi said:

You are right, though I would have not noticed it :). What about what you see in the EVF and LCD. Which resolution’s FOV does it match the closest?

Well, live view on the LCD and Visoflex both only show 96% horizontal FOV of the final image. I've changed the settings for L,M and S DNG on my phone while looking at Live View and no changes in the feed are seen. So to answer your question, the EVF and LCD live view probably represents most accurately the 18MP DNG option.

In very early firmwares, I was complaining to Leica about the live view feed being cropped and how precise framing was just not possible. They argued it was needed for stabilisation, which I understand is convenient when zooming in but not for the full frame feed. I do think they improved it marginally but I cannot recall for sure. 

Anyways, the differences in field of view are not big enough to annoy me. For instance, the focal length and FOV difference between my Summicron-M 50mm and my Planar 50mm are much larger than the different DNG resolution options on the M11.

[hmzimelka]

52 minutes ago, hdmesa said:

I only noticed the “issue” because I was getting more image area than what I was seeing in the EVF, so I imagine the Visoflex 2 is using the 18mp FOV. It was quite frustrating because I kept getting a small highlight over in one corner that I couldn’t see in the EVF. I’ll have to test it to be sure, or maybe @hmzimelka will beat me to it.

Edit: I only have an M11-D, so I can’t say if the LCD exactly mirrors the Visoflex 2.

Also, if you stack the 36 and 18mp images and switch back and forth between them like a GIF, the difference in FOV is a bit more dramatic than seeing the images side by side. I’d guess it’s the difference of about 2mm on a 50mm lens.

Yeah the LCD and Visoflex for me are pretty much the same. The live view feed is cropped slightly for stabilization purposes, so there is no such thing as perfect framing. I've asked Leica about it twice, and that I'm someone that would much prefer accurate framing when viewing the full image frame. 

[hdmesa]

39 minutes ago, hmzimelka said:

Yeah the LCD and Visoflex for me are pretty much the same. The live view feed is cropped slightly for stabilization purposes, so there is no such thing as perfect framing. I've asked Leica about it twice, and that I'm someone that would much prefer accurate framing when viewing the full image frame. 

I thought the digital stabilization only worked when magnified. At least it seems that way to me with non-coded lenses even with an M lens profile selected.

[hmzimelka]

3 hours ago, hdmesa said:

I thought the digital stabilization only worked when magnified. At least it seems that way to me with non-coded lenses even with an M lens profile selected.

Exactly, and that is how it would make sense to me too... however, image stabilisation is the reasoning Leica gave me to justify the 96% Live View feed... 

Edited November 2, 2024 by hmzimelka

[Advertisement]

Hallo,

Look here: Interesting Leica gear

[hdmesa]

2 minutes ago, hmzimelka said:

You are right, and that is how it would make sense to me too... however, image stabilisation is the reasoning Leica gave me to justify the 96% Live View feed... 

That makes me wonder why they don't stabilize the full view. Or maybe they do, but it's on a smaller scale that's only useful when zoomed. If that were the case, it might explain the flicking issue with the evaluative metering modes, just too many simultaneous processes running, and it causes a slowdown of the image feed.a