Does LR Super Resolution obviate the need to upgrade to M10-R?

[derleicaman]

Advertisement (gone after registration)

With the new Super Resolution feature available in Light Room Classic, theoretically a 24MP image from my M10-P becomes a 96MP image. Why would I want to upgrade to the M10-R at this point, aside from the increased dynamic range that the R offers? I was torn between acquiring an M10-R, or grabbing a NIB M10-P at a great price. I really think for most of my work, 24MP is more than enough. I was dreading having to upgrade my computer to handle 40+ MP files. 

I went with the M10-P, which will cut down on lens changing now that I have two M10-P bodies, one chrome and one black. This will be very welcome for travel when I can keep the WATE on one camera and a 35 or 50 on the other camera, or maybe even the MATE.

Jono was trying to convince me to go with the R, but I’m happy with my choice. What do you all think?

[Advertisement]

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

[setuporg]

As a computer scientist, I believe the “superresolution” is a particular form of sharpening through deep learning algorithms.  It uses patterns to enrich the texture.  However, the enrichment is only as good as the training data.  Eg if you photograph a rare chameleon whose photos were never used in training the algorithms, it would never be able to imagine its tiniest details.  In digital domain data is king.  If the higher resolution sensor is not adding noise, it’s always better to start with more original data.  I had the original M10, skipped the P, but upgraded to R as soon as it came out.  I’m seeing a great improvement for my use cases, and having just finished a road trip with both M10R and SL2, I prefer the M shots in every case, even with M APO 50 vs SL2 APO 50.  The highlights are much better OOC.

[otto.f]

After a thorough try-out of both camera’s I chose for the M10-R because of the quality of the images, noise is nicer, colors are better manageable to my taste. I’m not a great fan of cropping, but sometimes there’s no choice and with 40mp you really have a serious photo after cropping. But it is another way of making photo’s, you really don’t need to underexpose anymore, as in many cases you had to be aware of that in former M’s. 

Edited July 12, 2021 by otto.f

[thighslapper]

How well 'super resolution' enhances images depends to some extent on the subject matter. I do mostly landscape photography and found it makes close random textured stuff like foliage on trees look a bit odd and mushy when it's background detail. The other setting 'DNG enhance' appears to do precisely zero to my files. 

[Guest Nowhereman]

7 hours ago, otto.f said:

...But it is another way of making photo’s, you really don’t need to underexpose anymore, as in many cases you had to be aware of that in former M’s. 

Otto - This is an interesting point, and I never know what to make of it because it may simply reflect the contrast curve that Leica has written into the M10-R firmware, compared to that of the M10 firmware, which apparently has more contrast. Specifically, looking at the photonstophotos.net charts, the M10-R and the M10 have virtually the same dynamic range but it appears, as stated by some testers, that the distribution of this dynamic range is allocated differently, with the M10 having more of it in the shadows and less of it in the highlights, while the M10-R has relatively more of the dynamic range in the highlights than the M10. This, then, would be the reason that, to protect highlights, one needs to underexpose so much with the M10 than with the M10-R, and raise shadows in post-processing.

Actually, there could be something more general underlying this. Recently, to deal with an old M9 image — one that I only processed in B&W previously because the colors were so "difficult" —  I went back to the shareware Raw Photo Processor (RPP), which is often good for difficult to process images and which I haven't used for several years. The first thing that I saw was the my (correctly exposed) image was very dark when opened in RPP. The RPP manual states that images in RPP look dark: Because they really are underexposed. By default RPP shows images as they were captured by your camera in Raw with only essential tonal range adjustments. They may therefore look darker than you expect. Usually this happens because your camera’s light meter is calibrated to some low gray point value by the camera vendor. Most (if not all) modern DSLR cameras do this to preserve more highlights, and most other converters quietly apply compensation to your image to correct that. There are two ways to resolve this - apply exposure correction in RPP or adjust the camera’s light meter.

In my thread on the new Cobalt-Image camera profiles, I mentioned the Repro version of their basic camera profiles: Repro is their linear profile without dynamic compression, to be used when you want to get the maximum out of the DNG file, and is the equivalent of the dark view you get in RPP. Thinking about this and the RPP dark view of files makes me wish that Leica had not applied such a strong contrast curve to the M10 DNGs. Presumably they did this to get as close as possible, SOOC, to a higher contrast look then would otherwise be the case.

You also wrote above that you prefer the M10-R colors. In RPP images also appear to have less saturation than with other raw developers: the explanation is that, Other converters set the defaults to increase contrast and saturation to provide colorful result by default. For example, Adobe Lightroom by default sets the Contrast to +25, Black Point to +5, Medium Contrast Curves, etc. Increasing the contrast automatically leads to an increase in saturation. When using RPP you start with a more “honest” (in terms of information contained in the file) pictures.

The upshot of all this is that I would have preferred Leica to have applied less contrast to the M10 DNGs, and presumably this could be fixed through a firmware upgrade, but that is not likely to happen.
________________________
Frog Leaping photobook

Edited July 12, 2021 by Nowhereman

[pedaes]

11 hours ago, derleicaman said:

With the new Super Resolution feature

Another interesting question is how does it compare to Topaz AI Sharpen https://www.topazlabs.com/sharpen-ai?utm_source=google&utm_medium=cpc&utm_campaign=brand_search&attribution=true&gclid=Cj0KCQjw0K-HBhDDARIsAFJ6UGhTMNgyBmiWEmluSqt25bmlyEyeL4mg0yUimt0M_joNZIAUlwRZc5saAjDvEALw_wcB

 

[pedaes]

Advertisement (gone after registration)

11 hours ago, derleicaman said:

What do you all think?

I think a good decision. 24mp seems to be plenty for general photography upto A3+/A2 prints. The only advantage of more mps for me would be if I wanted to crop heavily, which I very rarely do. I use Topaz AI sharpen if I want/need to really max the detail, but few images need this. 

It would be interesting to test a enhanced 24mp M10 image against a R  image, but this is impossible to be meaningful at  resolution limits here. There will no doubt then be the question of an enhanced R image, but where do you stop with 35mm format. There is a bit of debate on another thread about tripods, although most of us see 35mm as a hand-held format with odd exceptions, which I guess could be about a photographers technique, as even at 24mp with the latest lenses your technique needs to be good to achieve best possible outcomes.

Edited July 12, 2021 by pedaes

[Jeff S]

I think the best way to answer such questions is to test the gear for yourself, both shooting and processing/output, and decide.  We each have different needs, standards and preferences, as well as different processing tools and methods, etc. Jono and others suggest that the strength of the M10-R lies in highlight protection/recovery, not necessarily resolution, and still others will warn about handholding techniques with high MP (much debated).  Only you can decide if these matters affect you, or even whether your computer can handle the files. 
 

For my print sizes, I have no image quality concerns whether using my standard M10 or my SL2.  Results are up to me; camera choice depends on other needs and goals. And my 11 year old Mac Pro (upgraded) handles all the files easily.

Jeff

Edited July 12, 2021 by Jeff S

[Photoworks]

after 2 years of 10M-P I have upgraded to M10-R. Best sedition for me for M camera use.
To me the sensor quality is what makes the difference on this camera. the sensor comes from Leica S3 and cut to 35mm.

few key points to consider.

No more burned out clouds and need to underexpose many stops. expose like you would any other camera. Finally !

better colors and best reds. Skin tones much nicer. the photos that come out of this camera are so true to the reality I had infant of me.

Noice didn't get worse with High MP, the new pixel design is helping getting cleaner images and any ISO.

 

[SrMi]

1 hour ago, Nowhereman said:

Otto - This is an interesting point, and I never know what to make of it because it may simply reflect the contrast curve that Leica has written into the M10-R firmware, compared to that of the M10 firmware, which apparently has more contrast. Specifically, looking at the photonstophotos.net charts, the M10-R and the M10 have virtually the same dynamic range but it appears, as stated by some testers, that the distribution of this dynamic range is allocated differently, with the M10 having more of it in the shadows and less of it in the highlights, while the M10-R has relatively more of the dynamic range in the highlights than the M10. This, then, would be the reason that, to protect highlights, one needs to underexpose so much with the M10 than with the M10-R, and raise shadows in post-processing.

Actually, there could be something more general underlying this. Recently, to deal with an old M9 image — one that I only processed in B&W previously because the colors were so "difficult" —  I went back to the shareware Raw Photo Processor (RPP), which is often good for difficult to process images and which I haven't used for several years. The first thing that I saw was the my (correctly exposed) image was very dark when opened in RPP. The RPP manual states that images in RPP look dark: Because they really are underexposed. By default RPP shows images as they were captured by your camera in Raw with only essential tonal range adjustments. They may therefore look darker than you expect. Usually this happens because your camera’s light meter is calibrated to some low gray point value by the camera vendor. Most (if not all) modern DSLR cameras do this to preserve more highlights, and most other converters quietly apply compensation to your image to correct that. There are two ways to resolve this - apply exposure correction in RPP or adjust the camera’s light meter.

In my thread on the new Cobalt-Image camera profiles, I mentioned the Repro version of their basic camera profiles: Repro is their linear profile without dynamic compression, to be used when you want to get the maximum out of the DNG file, and is the equivalent of the dark view you get in RPP. Thinking about this and the RPP dark view of files makes me wish that Leica had not applied such a strong contrast curve to the M10 DNGs. Presumably they did this to get as close as possible, SOOC, to a higher contrast look then would otherwise be the case.

You also wrote above that you prefer the M10-R colors. In RPP images also appear to have less saturation than with other raw developers: the explanation is that, Other converters set the defaults to increase contrast and saturation to provide colorful result by default. For example, Adobe Lightroom by default sets the Contrast to +25, Black Point to +5, Medium Contrast Curves, etc. Increasing the contrast automatically leads to an increase in saturation. When using RPP you start with a more “honest” (in terms of information contained in the file) pictures.

The upshot of all this is that I would have preferred Leica to have applied less contrast to the M10 DNGs, and presumably this could be fixed through a firmware upgrade, but that is not likely to happen.
________________________
Frog Leaping photobook

There is no dynamic range of shadows or highlights. Dynamic range is the ratio between the largest signal and smallest detectable signal.

How do you define correctly exposed? I am not arguing what it should mean, but wondering what it means in the context of your post.

IMO, when people evaluate a camera, they should evaluate it in combination with their favorite post processing tools.

BTW, thanks for writing about Cobalt-Image camera profiles. I bought a bunch from them :).

[pedaes]

1 hour ago, Photoworks said:

No more burned out clouds and need to underexpose many stops

I am pleased you are happy with your M10-R, but I have never ever had an issue with M10 and balancing exposure. Also, not sure what 'better colours and best reds' means - do you shoot jpg?

Anyway, as I said, pleased you are a happy bunny.

Edited July 12, 2021 by pedaes

[Mark II]

I find high-contrast scenes with clouds tend to blow out the sky on my M10 too. It is definitely worse at ISO100, where I need about -2/3 stop exposure compensation to avoid the problem in a typical wide-angle landscape scene shot with classic metering.

Curiously, the photonstophotos.net dynamic range chart suggests the DR of M10 and M10-R are almost exactly the same, while several online tests of resolution etc have noted that the M10-R exposure times are almost shorter than for the M10 with the same aperture and ISO. My interpretation is that there probably is not much difference between the cameras other than the M10-R effectively doing the equivalent of shooting the M10 with a slight under-exposure and perhaps applying a different luminosity curve to compensate.

[Photoworks]

57 minutes ago, Mark II said:

My interpretation is that there probably is not much difference between the cameras other than the M10-R effectively doing the equivalent of shooting the M10 with a slight under-exposure and perhaps applying a different luminosity curve to compensate.

the M10-R is a difference sensor.

When I meet dr. Kaufman in Paris he ask what I like to see improved on the M10-P. my answer was a like a camera that does not blow highlights so quickly.
and he said: we have the new SL2 .

Few years make a difference in sensor tech. and for sure Leica learned and improved their in camera software.

1 hour ago, pedaes said:

Also, not sure what 'better colours and best reds' means - do you shoot jpg?

I shoot Raw+J, almost never use the JPG, but it is interesting to learn what the leica look is, how much and what they correct, how much hi lights recovery and clarity is added .

the color differences are in the RAW and on JPG.

I high ISO the is more color retention then the M10-P often the more that you amplify the signal the more color purity you loose, modern sensors deliver better results. It is not all about the noise.

[SrMi]

5 minutes ago, Photoworks said:

the M10-R is a difference sensor.

When I meet dr. Kaufman in Paris he ask what I like to see improved on the M10-P. my answer was a like a camera that does not blow highlights so quickly.
and he said: we have the new SL2 .

Few years make a difference in sensor tech. and for sure Leica learned and improved their in camera software.

I shoot Raw+J, almost never use the JPG, but it is interesting to learn what the leica look is, how much and what they correct, how much hi lights recovery and clarity is added .

the color differences are in the RAW and on JPG.

I high ISO the is more color retention then the M10-P often the more that you amplify the signal the more color purity you loose, modern sensors deliver better results. It is not all about the noise.

In almost all cases, blown highlights are caused by metering, not by sensor itself.

[Guest Nowhereman]

2 hours ago, SrMi said:

...How do you define correctly exposed? I am not arguing what it should mean, but wondering what it means in the context of your post...

Essentially that the mid-tones fall on Zone V, if that there is where I want them; and for shots into the light, or with strong sidelight, when I'm exposing for the high-lights, that the main subject is exposed where I want it, after I lift shadows and mid-tones to where I want them. My issue with the M10 is that it's necessary to underexpose too much when exposing for the highlights, compared to many other cameras (the Ricoh GRIII in my case), and that this results from the contrast curve that Leica applied to the M10 DNGs, which allocates too little of the available dynamic range to the highlights.

Sorry that I wrote such a long off-topic post above, but it's an issue that I'm interested in and @otto.f's post touched on that. On the topic of this thread itself, I agree with @pedaes's thought that it would be interesting to have a comparison of Lightroom Super Resolution to the Topaz AI tools; and also agree with @Jeff S's statement that the necessary sensor resolution really depends on what the photographer wants. This has a long history. On the old Leica forums, people would say that, if you wanted print's larger than 11x14 — or even 8x10, though some said 16x20 — you needed to go to medium format film. All this was blown out of the water for me when, around 2006, I saw the Daido Moriyama retrospective at the Gallery of New South West in Sydney, with sixty prints of 100x150 cm printed on an Epson wide-format inkjet printer from 35 mm Twi-X negatives. The prints were dazzling, but clearly this type of high-contrast work requires less resolution than certain types of landscape.

On a future M11, I would like to see a sensor something like the one that Leica uses in the SL2S.
________________________
Frog Leaping photobook

[Jeff S]

I’d love to have a dollar for every time the term ‘dynamic range’ is used in lieu of the more appropriate ‘exposure latitude’.

Jeff

[pedaes]

57 minutes ago, Photoworks said:

I high ISO the is more color retention then the M10-P often the more that you amplify the signal the more color purity you loose, modern sensors deliver better results. It is not all about the noise.

I don't know what this means.

I try and use native ISO as much as possible. The effect i think you are talking about is true, but when it becomes visibly apparent varies between all cameras. 

[pgk]

11 hours ago, setuporg said:

As a computer scientist, I believe the “superresolution” is a particular form of sharpening through deep learning algorithms.  It uses patterns to enrich the texture.  However, the enrichment is only as good as the training data.

Imagine photographing a scene with a newspaper in it, as they did in old movies. Would using superresolution reveal the date on the page if the data isn't in the original image? No, of course not. Its a statistical way of interpreting what missing data and as such cannot produce 'original' data which is missing. Although it may give the impression of more data, its extrapolated data, not 'original' data, which is why a higher resolution sensor, combined with a good lens and good technique, will always provide more 'original' data and a correctly detailed image.

[Advertisement]

Hallo,

Look here: Leica M

[SrMi]

4 minutes ago, Jeff S said:

I’d love to have a dollar for every time the term ‘dynamic range’ is used in lieu of the more appropriate ‘exposure latitude’.

Jeff

I understand exposure latitude in the context of the film, but what does it mean in the context of digital sensors?

For me, dynamic range means how much noise is there in the shadows.

[Guest Nowhereman]

8 minutes ago, Jeff S said:

I’d love to have a dollar for every time the term ‘dynamic range’ is used in lieu of the more appropriate ‘exposure latitude'...

In this case you'd be trying to earn your money through pedantry, i.e., false pretenses, because I'm talking about dynamic range. But, hey, knock yourself out.
________________________
Frog Leaping photobook