Jump to content

M10 vs Sony A7RIII - ISO differences


indergaard

Recommended Posts

Advertisement (gone after registration)

I've started to compare an M10 with an A7RIII lately.

So far I've based my comparisons with these lenses: Summilux 50 1.4 ASPH and Sony/Zeiss 55/1.8 Sonnar.

 

Detail-wise, the M10 can't compete with the A7RIII of course. Color-wise they aren't that different. The M10 AWB is warmer and the A7RIII cooler. Match the white balance and the color between them are about identical, both in Capture One 11 and Lightroom 7.3.1.

 

But what surprised me is the ISO differences between these cameras.

 

To my eyes, 100 ISO on the M10 appears to be the same as 160 ISO on the A7RIII. Image for image, at the exact same settings and under the exact same lighting conditions, the files from the M10 are noticeably under-exposed, and needs a +0,60 exposure boost in Lightroom (7.3.1) to match the un-touched files from the A7RIII.

 

Is this normal? If so, is Leica doing basically what Fuji has been doing for years now? ISO cheating? or, as the Fujifans say, Fuji uses a different ISO variant than all the others. Regardless of what you want to call it, it's about being able to display cleaner files at higher ISO's by playing with numbers.

 

Or is the difference I am seeing a very big T-Stop difference between the Sonnar 55/1.8 and Summilux 50 ASPH lenses? If so, the Sonnar almost acts like a f/1.2 lens by comparison of light gathering capabilities. Bokeh-wise, they are about the same also. 50mm at f/1.4 and 55mm at f/1.8 looks almost identical.

 

Just wondering if anyone else has done similar comparisons and what your experience is?

 

Attached is two images I made at ISO 100, f/2.0, and 1/30 on the same tripod, under the same light, with both setups.

 

Uncorrected, straight from the camera:

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!

 

Corrected - added +0,60 exposure to the M10 file:

 

100% detail shot on point of focus after correction:

  • Like 3
Link to post
Share on other sites

Interesting. This kind of variation has been signaled regularly on this forum, root cause is the interpretation freedom built into the digital ISO norm.

You will see such differences between other brands as well.

  • Like 1
Link to post
Share on other sites

Interesting. This kind of variation has been signaled regularly on this forum, root cause is the interpretation freedom built into the digital ISO norm.

You will see such differences between other brands as well.

 

Yep, but that doesn't change the fact that ISO 6400 on the M10 is equal to ISO 10000 on the A7RIII if the images are adjusted for matching exposure. Turn all NR off and the M10's ISO performance all of a sudden isn't that impressing, especially considering the resolution differences.

 

I'm disappointed by Leica adopting this trend of manipulating numbers. Sony, Nikon and Canon seems to be the only manufacturers left that use the same variance of the ISO standard. Fuji. and now Leica, are playing with numbers to be able to be able to score higher. But side-by-side comparisons reveal the differences quite easily.

 

Also, the Sonnar 55/1.8 isn't the brightest E-mount lens. By comparison, at F/4, the Sony 24-105 F/4 zoom is actually about 0.3 stops brighter than the Sonnar 55/1.8 is at F/4 due to different T-Stop qualities. If I compared the huge consumer zoom at f/4 and the Summilux 50 ASPH at f/4 the difference would be bigger, almost a stop (+0.90 exposure) actually. This is not a small difference when we are talking about ISO values, noise, and image quality at higher ISO values. 

 

I like the M10, but it's really disappointing by Leica to adopt this play with numbers trend.

Edited by indergaard
Link to post
Share on other sites

That is exactly what was discussed in another thread: DxO Mark just measured exactly that and the M10 had there an especially big deviation from the real ISO. In my eyes that has less to do with the fredom of interpreting the ISO norm. There is no Canon or no Nikon or Sony that is as „bad“ according to DxO. Threre is probably no other no cost data base available that delivers us all these numbers.

Link to post
Share on other sites

Thanks for these. I'll respond to your questions about the 55/1.8 Sonnar FE in the other thread in a bit. But just upfront, yes, your're right, in terms of sharpness corner to corner wide open it's a stunner, still sharper in the center though, especially after applying the lens profile in LR but so is the Otus.  I hadn't bothered to compare it against Leica lenses so far and I don't own the 50 Summilux. I can probably borrow it.

Or is the difference I am seeing a very big T-Stop difference between the Sonnar 55/1.8 and Summilux 50 ASPH lenses? If so, the Sonnar almost acts like a f/1.2 lens by comparison of light gathering capabilities. Bokeh-wise, they are about the same also. 50mm at f/1.4 and 55mm at f/1.8 looks almost identical.

Judging a sensor's performance without a lens is meaningless to me. The M10 + 50 Noctilux stopped down a bit beats the pants off the α7R III + 50/1.4 Planar FE up to a certain ISO level in low light situations IMO. Look at the church pictures in the link below. The Nocitlux at f/2.8 on the M10 is a stunner. It would be even better on a Kolari modified Sony body but I don't have one. Also, detail-wise the M10 + 35 Summicron beats the α7R III + 35/2.8 Sonnar by a mile despite the Leica being only 24 MPx. See the last few pairs in the link.

 

Here's the link:  https://www.smugmug.com/gallery/n-Jfdr66/

 

And below my Sony gear and I'll do M10 + 50 Apo Summicron vs. α7R III + 55/1.8 Sonnar FE next. The only body that may be better suited to use Leica lenses on would be a Kolari modified Sony but Leica will be coming up with their own FF 40+ MPx BSI sensor, most likely for the L-mount, soon, I believe. Then we'll compare, again. In the meantime, the M10 is by far my favorite compact FF camera.

 

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 by Chaemono
  • Like 3
Link to post
Share on other sites

Thanks for these. I'll respond to your questions about the 55/1.8 Sonnar FE in the other thread in a bit. But just upfront, yes, your're right, in terms of sharpness corner to corner wide open it's a stunner, still sharper in the center though, especially after applying the lens profile in LR but so is the Otus.  I hadn't bothered to compare it against Leica lenses so far and I don't own the 50 Summilux. I can probably borrow it.

Judging a sensor's performance without a lens is meaningless to me. The M10 + 50 Noctilux stopped down a bit beats the pants off the α7R III + 50/1.4 Planar FE up to a certain ISO level in low light situations IMO. Look at the church pictures in the link below. The Nocitlux at f/2.8 on the M10 is a stunner. It would be even better on a Kolari modified Sony body but I don't have one. Also, detail-wise the M10 + 35 Summicron beats the α7R III + 35/2.8 Sonnar by a mile despite the Leica being only 24 MPx. See the last few pairs in the link.

 

Here's the link:  https://www.smugmug.com/gallery/n-Jfdr66/

 

And below my Sony gear and I'll do M10 + 50 Apo Summicron vs. α7R III + 55/1.8 Sonnar FE next. The only body that may be better suited to use Leica lenses on would be a Kolari modified Sony but Leica will be coming up with their own FF 40+ MPx BSI sensor, most likely for the L-mount, soon, I believe. Then we'll compare, again. In the meantime, the M10 is by far my favorite compact FF camera.

 

 

 

 

Thanks for your input. I haven't really put any effort into comparing detail and resolution, yet. I picked up an M10 yesterday, and already owned an A7RIII + the 28/2, 55/1.8, 24-105, and a Voigtlander 40/1.2 Aspherical (a lens that truly renders beautifully), and also own a Lux 50 ASPH and Summicron 35 ASPH. I'm waiting for the FE 35/2.8 to show up so I can compare it to the Cron.

 

I've owned the Noctilux 0.95 ASPH and used it on an M240 some years ago, but sold it because it is simply too big and heavy for my use. It's a magical lens though, but not Leica's sharpest. The Lux 50 ASPH and APO-Summicron 50 both beat the Noct in regards to resolving power at the same apertures. The Noct has far less vignetting though, which gives it it's beautiful "airy" quality at apertures up to f/5.6 compared to the Lux and Cron.

 

You can see the resolving power thoroughly tested by Lens-Rentals here: https://www.lensrentals.com/blog/2014/06/comparing-rangefinder-and-slr-50mm-lenses-version-0-7/

 

Can I ask how you compared the resolution between a 24mp and 42mp sensor? Did you upscale the 24mp images, or downscale the 42mp images?

I'm looking forward to your comparison between the 55/1.8 Sonnar and 50/2 APO-Summicron. I've considered the APO many times, but for me the Lux is simply two lenses in one. It offers a strong signature look at f/1.4 that doesn't exist on the APO, and looks almost identical to the APO at f/2 and beyond, except the fact that the APO has stronger contrast and better micro-contrast between f/2 and f/4, which makes subjects really stand out in the image.

 

I wish I could have my 35 Lux FLE back to compare with the 35/2.8 :) That would be interesting, as the Lux FLE really makes subjects stand out versus the 35 Cron. But overall I much prefer the size and ergonomics of the cron. It's such a small and compact bring everywhere kinda lens that performs great.

 

In the last two images on your link (the headlights of the car), I can see that you compare the FE 35/2.8 and Cron 35/2 ASPH, and I can clearly see that the Cron renders the image much more 3-dimensionally. From the EXIF info I can see that you shot both images at ISO 1000 with a shutter speed of 1/60, but since the M10 doesn't record the aperture used, I can only see F/4 on the Sony lens. Did you shoot the M10 at f/4 also? If so that's kinda interesting. Did you adjust the exposure in post?

 

I'm just curious, cause my testing so far shows that the M10 exposures need about +0,60 exposure adjustment to match the A7RIII's exposures with the same settings (ISO, shutter speed, aperture). Have you seen this as well?

Edited by indergaard
Link to post
Share on other sites

Advertisement (gone after registration)

Many questions and I am busy with other stuff right now. Some answers, already. Yes, I upscale the M10 pictures to 42 MPx.  :) I don't think that's cheating. The Summicrons stopped down a bit then kick butt.  :)

 

Yes, the Sony-Leica pairs were always shot at the exact same aperture.

 

On the engine pictures in the link, first an ARW-JPEG and a DND-JPEG, both not touched, are shown and then the two pictures that follow are with the exact same adjustments except, as you rightly pointed out, Exposure +0.37 for the Sony and +1.05 for the Leica. See the the post in this thread for the exact adjustments: https://www.l-camera-forum.com/topic/279882-dxo-mark-leica-m10-score/page-14.

 

I can provide the raw files. 

 

Again, the link to the JPEG pairs: https://www.smugmug.com/gallery/n-Jfdr66/

 

On the headlights of the car, the adjustments were as shown below (and Vignetting +30 on the Leica because the LR applied profile does not include this adjustment unlike for the Sony) . Again, I can provide the raw files.

 

Sony headlights adjustments

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!

 

M10 headlights adjustments

 

  • Like 2
Link to post
Share on other sites

That is exactly what was discussed in another thread: DxO Mark just measured exactly that and the M10 had there an especially big deviation from the real ISO. In my eyes that has less to do with the fredom of interpreting the ISO norm. There is no Canon or no Nikon or Sony that is as „bad“ according to DxO. Threre is probably no other no cost data base available that delivers us all these numbers.

It is not really a play with numbers thing. The ISO norm specifies standardized development and density values for film. However, as ISO values from sensors are influenced by the ADC conversion and  the amplification, plus the further settings in the pipeline like black-point and sensitivity curves, they cannot be more specific than an equivalence with film. That leaves a lot of leeway. I'm sure that Leica could give a semi-convincing argument why they are closer to the norm than others. On a side-note, a few years ago Canon was the brand that was out of step, and gave optimistic ISO values (Canon told the world that they were right and the others wrong - until they changed their methodology ;) ). It all boils down to "shooting a camera in" when new to the type, just like with film.

 

BTW, DXO is not always that reliable for practical purposes, as they publish the theoretical numbers, which may differ considerably from the actual usable numbers.

 

Another quibble - a lens will not transmit the same light intensity as another one of the same f-value. The reason is that the transmission factors of the various optical glasses used and the coating vary. You should use T-stops for this kind of comparison, as they denote the actual light transmission.

  • Like 1
Link to post
Share on other sites

Many questions and I am busy with other stuff right now. Some answers, already. Yes, I upscale the M10 pictures to 42 MPx.  :) I don't think that's cheating. The Summicrons stopped down a bit then kick butt.  :)

 

Yes, the Sony-Leica pairs were always shot at the exact same aperture.

 

On the engine pictures in the link, first an ARW-JPEG and a DND-JPEG, both not touched, are shown and then the two pictures that follow are with the exact same adjustments except, as you rightly pointed out, Exposure +0.37 for the Sony and +1.05 for the Leica. See the the post in this thread for the exact adjustments: https://www.l-camera-forum.com/topic/279882-dxo-mark-leica-m10-score/page-14.

 

 

 

 

 

 

I did a comparitive test shooting the M10 + Lux 50 ASPH at f/2.8 and the A7rIII + Sonnar 55 at f/2.8 today. I shot a canvas painting with tons of fine detail (canvas threads, edges and textures of paint, various areas of very thick paint, etc), and, I tried upscaling the image in Photoshop CC 2018 using the "Preserve Detail 2.0" upscaling method, which is by far the best method of upscaling: https://www.photoshopessentials.com/basics/upscale-images-photoshop-cc-2018/

 

The upscaled images from the M10 and Lux doesn't show nowhere near the same level of acuity and fine detail after upscaling to the exact same image dimensions from the A7rIII. I would post examples here, but the 500 kB file limit prevents me from uploading any image with meaningful detail. The jpeg compression needed just wipes out all the fine detail in the images.

 

So how exactly where you able to upscale with that level of clarity?

 

Also, the A7rIII files can also be upscaled to 80mp, and also lose about the same acuity as the upscaled M10 images. So...

Edited by indergaard
Link to post
Share on other sites

I did a comparitive test shooting the M10 + Lux 50 ASPH at f/2.8 and the A7rIII + Sonnar 55 at f/2.8 today [...] The upscaled images from the M10 and Lux doesn't show nowhere near the same level of acuity and fine detail after upscaling to the exact same image dimensions from the A7rIII.

 

I don't know how the 50 Summilux-M behaves stopped down a bit but in theory as with all lenses contrast should increase substantially resulting in better perceived sharpness. See here why that is: http://photography-on-the.net/forum/showthread.php?t=1039547&page=1.

 

The 35/2 Sonnar FE could be an exception to this as it's a bit crap wide open and stopped down. Also, see ramarren's comments in #38 about stopping down Leica lenses here: https://www.l-camera-forum.com/topic/282233-depth-of-field/page-2. He knows a thing or two about lenses, I believe.

 

So how exactly where you able to upscale with that level of clarity?

I simply tried to take two identically framed pictures with each of my favs (favorite compact FF combos) using the exact same settings on each, then imported the raw files into LR, made the exact same adjustments except for +0.60 to +0.80 more exposure adjustment to the M10 DNG, and used ON1 Resize to upscale the M10 DNG to 42 MPx to level the playing field.

 

Detail-wise and in terms of depth rendering, the M10 + 35 Summicron (one of my favorite lenses because of size and weight) beats the α7R III + 35/2.8 Sonnar FE by a mile. Link to the JPEGs that show this here: https://www.smugmug.com/gallery/n-Jfdr66/

 

Also, the A7rIII files can also be upscaled to 80mp, and also lose about the same acuity as the upscaled M10 images. So...

So, "tell me, Mr. Anderson, what good is a phone call if you're unable to speak?" Or to paraphrase Agent Smith, what good is upscaling if your lens is unable to resolve the details in the first place? See here and think of the α7R III + 35/2.8 Sonnar FE combo resolution as Neo's lips:

Edited by Chaemono
Link to post
Share on other sites

despite all of that advantage and disadvantage. which one you love to take outside to shoot? i wonder how many will choose a7r3 as their favourite  :)

 

I like both for different reasons. As my sunday driver and daily snapshot camera the M10 is great. But the MP does just as good a job as a sunday driver, if not better, except in the winter time when it's dark all the time.

 

When I want the best technical results (landscapes, architecture, etc) where the resulting image and print matters, the A7rIII is my favorite. It's a more precise and methodical tool, and has the feature set that I need for those types of photographs. Why Leica removed such a mandatory and basic thing as the horizon level on the M10 boggles my mind.

Edited by indergaard
  • Like 5
Link to post
Share on other sites

It is not really a play with numbers thing. The ISO norm specifies standardized development and density values for film. However, as ISO values from sensors are influenced by the ADC conversion and the amplification, plus the further settings in the pipeline like black-point and sensitivity curves, they cannot be more specific than an equivalence with film. That leaves a lot of leeway. I'm sure that Leica could give a semi-convincing argument why they are closer to the norm than others. On a side-note, a few years ago Canon was the brand that was out of step, and gave optimistic ISO values (Canon told the world that they were right and the others wrong - until they changed their methodology ;) ). It all boils down to "shooting a camera in" when new to the type, just like with film.

 

BTW, DXO is not always that reliable for practical purposes, as they publish the theoretical numbers, which may differ considerably from the actual usable numbers.

 

Another quibble - a lens will not transmit the same light intensity as another one of the same f-value. The reason is that the transmission factors of the various optical glasses used and the coating vary. You should use T-stops for this kind of comparison, as they denote the actual light transmission.

Hi Jaap, is there not a misunderstanding somwhere along the line: It is not true that the others are further away from the norm; it is Leica M10 that is far too optimistic. Same as about Fuji. I can not see that Canon is too optimistic. I think that several people here found out that there is a difference between M10 and others. The brightnes of the pictures should be the same with all cameras (of course lenses let more or less light through but this is not the reason). Just compare the M10 and Q and M typ 240. These are all Leicas but the M10 is very different.

 

I would frankly like to know more about that. Just to question the DxO database seems to me not to be the answer as the database says exactly what we can se ourselves. So the „test results“ of the members here correspond to the figures in the data base.

Edited by Alex U.
Link to post
Share on other sites

I think that several people here found out that there is a difference between M10 and others. The brightnes of the pictures should be the same with all cameras (of course lenses let more or less light through but this is not the reason). Just compare the M10 and Q and M typ 240. These are all Leicas but the M10 is very different.

 

I would frankly like to know more about that.

See #6 here: https://www.l-camera-forum.com/topic/280133-m10-vs-m240-dynamic-range/

Link to post
Share on other sites

 

He's not talking about how bright or not the shadows are. He is referring to the fact that Leica is cooking the ISO numbers to make the camera seem like it performs better at higher ISO values than it actually does. Fuji does the same. Great marketing gimmick and bragging rights, but to a person with cameras from other manufacturers, or a user who has a film Leica body with a meter, or a user who has an external light meter, it's very, very noticeable.

 

If anything, I would like consistency from Leica. When the meter on my Leica MP doesn't match the M10 with the same exposure values and lens, something is wrong. But it doesn't just stop there. The M10 doesn't match my Gossen Sixtomat F2 external light meter either. Nor does it match the light-meter from my A7rIII. Nor does it match the light meter on my wife's Canon 5DIII. This inconsistency that Leica has introduced in the M10 is very annoying, as it requires "special treatment", unlike the rest of my gear, and unlike other Leica cameras.

 

ISO 400 on the M10 is more or less equal to ISO 250-320 on everything else that I have. ISO 800 = 400-640. ISO 1600 = 1000-1250. ISO 3200 = 2000-2500. Etc...

 

It would be nice if Leica could be consistent between their camera models. Fuji does the same ISO number cooking, but the difference is that Fuji does the same consistently across all their cameras, and have done so from day one, which makes it easy to move between different Fuji cameras at least.

 

If I set any of my other cameras exposure to match my Gossen Sixtomat F2 light-meter, all of them get a correct exposure.

The M10 under-exposes about 0.6 EV consistently at the same settings on all scenes compared to my other equipment. It doesn't matter which lens is mounted. And it's not just the shadows, it's the highlights and the entire picture that is under-exposed.

 

Not cool Leica, not cool! Fix it in firmware!

Edited by indergaard
  • Like 2
Link to post
Share on other sites

 

 

Uh. That has absolutely nothing to do with these skewed ISO values. They are all software based choices, and has zero to do with the sensor or hardware itself. The code was probably implemented by reluctant engineers, but the decision was probably made by the product management team, with heavy influence and pressure from the marketing department, who had the backing of some clueless beancounter, who made some PowerPoint presentations about how much better ROI this would give. It's all about making the R&D look better than it is. It's quite common these days across the entire world of technology.

Edited by indergaard
  • Like 1
Link to post
Share on other sites

Hi Jaap, is there not a misunderstanding somwhere along the line: It is not true that the others are further away from the norm; it is Leica M10 that is far too optimistic. Same as about Fuji. I can not see that Canon is too optimistic. I think that several people here found out that there is a difference between M10 and others. The brightnes of the pictures should be the same with all cameras (of course lenses let more or less light through but this is not the reason). Just compare the M10 and Q and M typ 240. These are all Leicas but the M10 is very different.

 

I would frankly like to know more about that. Just to question the DxO database seems to me not to be the answer as the database says exactly what we can se ourselves. So the „test results“ of the members here correspond to the figures in the data base.

Well, the discussion of who is keeping to the norm depends on the norm itself. A discussion I will happily stay out of.

  • Like 1
Link to post
Share on other sites

nice article about ISO https://photography.tutsplus.com/articles/what-is-iso-a-technical-exploration--photo-11963

 

 

Uh. That has absolutely nothing to do with these skewed ISO values. They are all software based choices, and has zero to do with the sensor or hardware itself. The code was probably implemented by reluctant engineers, but the decision was probably made by the product management team, with heavy influence and pressure from the marketing department, who had the backing of some clueless beancounter, who made some PowerPoint presentations about how much better ROI this would give. It's all about making the R&D look better than it is. It's quite common these days across the entire world of technology.

Link to post
Share on other sites

I think that article is highly to the point when we are discussing variations in camera response to exposre, related to ISO values.

Skewed, however, iI find  rather charged word in a technical discussion. As long as the various manufacturers don't disclose how they set their ISO values and how they related their parameters to the ISO norm, we can only speculate.

The main thing is, IMO, to be aware of the bias of the camera one is using and expose accordingly. Something threads like this are quite useful for. The actual number the setting is labelled with is basically irrelevant.

Link to post
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

×
×
  • Create New...