can the 24-90 be the best standard zoom Leica has ever produced

[scott kirkpatrick]

Advertisement (gone after registration)

No doubt, but nothing like as many variable corrections and work as would be required if the lens was to be used on a wider range of bodies (including film) like the Nikon and Canon equivalents. Without digital corrections, I'd imagine this SL lens shows very pronounced distortion towards the 24mm end and, possibly, lower sharpness at the other extreme.

I looked at the distortion corrections embedded in the dngs from this lens when it first came out.  There is quite a bit corrected at the 24 end, and the corrections for barrel distortion only go to zero at 50 mm.  Then there is some pincushion distortion (a smaller amount) from 70 to 90 mm.  I don't know about sharpness at the long end (where the resolution/contrast is definitely less, according to the mtfs).  Software can cover that up, but not restore it.

 

scott 

[Peter H]

I looked at the distortion corrections embedded in the dngs from this lens when it first came out.  There is quite a bit corrected at the 24 end, and the corrections for barrel distortion only go to zero at 50 mm.  Then there is some pincushion distortion (a smaller amount) from 70 to 90 mm.  I don't know about sharpness at the long end (where the resolution/contrast is definitely less, according to the mtfs).  Software can cover that up, but not restore it.

 

scott 

 

 

I understood digital correction in this context (as with the SL24-90) to be a matter of the lens being designed for the software (or whatever it's called) in the camera so that no restoration is required, so it's not a matter of covering things up, but designing for the best possible performance. It's different from the sort of correction you'd do in PS or LR.

 

Am I wrong?

Edited March 2, 2016 by Peter H

[Advertisement]

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

[farnz]

Barrel distortion and pincushion distortion can be corrected in PS and I correct for them in some of my more extreme lenses.

 

Pete.

[pgk]

I understood digital correction in this context (as with the SL24-90) to be a matter of the lens being designed for the software (or whatever it's called) in the camera so that no restoration is required, so it's not a matter of covering things up, but designing for the best possible performance. It's different from the sort of correction you'd do in PS or LR.

 

Am I wrong?

 

Well this is my thought too. But I don't think that any significant lack of 'sharpness' can be adjusted for. Distortions and some slight aberrations perhaps.

Edited March 2, 2016 by pgk

[jaapv]

I understood digital correction in this context (as with the SL24-90) to be a matter of the lens being designed for the software (or whatever it's called) in the camera so that no restoration is required, so it's not a matter of covering things up, but designing for the best possible performance. It's different from the sort of correction you'd do in PS or LR.

 

Am I wrong?

No, you are right.

[ramarren]

The difference between the kind of lens corrections that can be applied post-facto by folks external to Leica and the kinds of corrections that Leica can provide for a lens is that when Leica does it with corrections embedded in the lens firmware or raw files, they can adjust the lens' optical properties such that the post-exposure corrections they specify can be done with minimum losses. 

 

For example, consider the case (pretty common with modern zoom lenses) where a lens has "mustache" shaped rectilinear distortion caused by its optical design (which was created to minimize rectilinear distortion as a whole). Writing a correction algorithm to fix that is fairly complicated and often fairly lossy. However, if you design that lens to have simple barrel or pincushion distortion instead, even if the total uncorrected distortion is greater, the software correction algorithm post-exposure will produce cleaner results and is easier to implement. 

 

It's a matter of applying the right optical AND software correction designs in a coordinated fashion to produce the best possible results. You can't create details that weren't captured, so the optical design and focusing all have to be up to snuff in the first place, but you can minimize geometric and other aberrations that couldn't otherwise be avoided while losing as little data as possible. 

[FlashGordonPhotography]

Advertisement (gone after registration)

 

 

 

Lack of AF doesn't bother everyone, though evidently it bothers you.

 

As the owner of 3 M camera and a dozen lenses, not really, in general. But the SL is an AF camera. If I wanted manual focus in that range I'd use an M and M lenses.

 

I bought the SL as an AF camera that supports the occasional M lens.

 

Gordon

[cpclee]

I think you nailed it.  The idea is that when you design a digital lens + body system from ground up, as in the SL, you get to choose what kind of things to correct optically and what kind of things to leave for software to address.  You apply your optical correction "budget" to imaging faults that must be dealt with or can only be dealt with effectively optically, which leaves you imaging faults that can be dealt with at little or no loss via software.  So for a given price point, say $5000 for the 24-90, you apply a certain mix of optical and digital corrections such that you get the best resulting outcome you can achieve for that price point.     You would do worse if you try to address optically imaging faults that could be easily addressed in software (eg. wasting your optical correction "budget") or conversely if you attempt to digitally take care of imaging faults that weren't so effective for software to handle.

 

 

The difference between the kind of lens corrections that can be applied post-facto by folks external to Leica and the kinds of corrections that Leica can provide for a lens is that when Leica does it with corrections embedded in the lens firmware or raw files, they can adjust the lens' optical properties such that the post-exposure corrections they specify can be done with minimum losses. 

 

For example, consider the case (pretty common with modern zoom lenses) where a lens has "mustache" shaped rectilinear distortion caused by its optical design (which was created to minimize rectilinear distortion as a whole). Writing a correction algorithm to fix that is fairly complicated and often fairly lossy. However, if you design that lens to have simple barrel or pincushion distortion instead, even if the total uncorrected distortion is greater, the software correction algorithm post-exposure will produce cleaner results and is easier to implement. 

 

It's a matter of applying the right optical AND software correction designs in a coordinated fashion to produce the best possible results. You can't create details that weren't captured, so the optical design and focusing all have to be up to snuff in the first place, but you can minimize geometric and other aberrations that couldn't otherwise be avoided while losing as little data as possible. 

[pgk]

So for a given price point, say $5000 for the 24-90, you apply a certain mix of optical and digital corrections such that you get the best resulting outcome you can achieve for that price point.

 

One last point is that you also need to implement viable quality control so that you are able to guarantee that each lens that you produce is accurately enough built to allow software to optimise it as well as possible. From talking to a lens designer friend, I know that this in itself can actually be an expensive part of production because it will require testing which is time consuming and costly

[ramarren]

As the owner of 3 M camera and a dozen lenses, not really, in general. But the SL is an AF camera. If I wanted manual focus in that range I'd use an M and M lenses.

 

I bought the SL as an AF camera that supports the occasional M lens.

 

Gordon

 

 

To me, this points out how difficult it is to be a camera manufacturer: the broad diversity of the audience means that there will be some unresolvable conflicts in expectations no matter what you do. 

 

I bought the SL as a digital platform for my R lenses, primarily, and secondarily for its dedicated lenses. Although I think the 24-90 is a very very good performer, I've used my R lenses instead of it by a factor of 10 or more. 

 

We all have our predilections ... 

[cpclee]

My local Leica store tells me

 

- The SL has been outselling M cameras by quite a bit.  Oftentimes it is sold to a current digital M users for use with their M lenses.  Sometimes, but less frequently it seems, it is sold to be used with R lenses.

 

- A lot of SL users buy the SL without the 24-90 zoom, roughly 50% or more by their rough estimate.

 

Personally, I was interested in buying the SL only if I would buy the 24-90 too. I wanted access to a modern, versatile and productive setup for what I used to shoot with my R system.  The SL doubles as a great camera for using M lenses, but if I were only to use M lenses the M262 would be a better camera for me. I'm really happy with the choice and the opportunities that the SL + 24-90 opened up. 

 

To me, this points out how difficult it is to be a camera manufacturer: the broad diversity of the audience means that there will be some unresolvable conflicts in expectations no matter what you do. 

 

I bought the SL as a digital platform for my R lenses, primarily, and secondarily for its dedicated lenses. Although I think the 24-90 is a very very good performer, I've used my R lenses instead of it by a factor of 10 or more. 

 

We all have our predilections ... 

[ramarren]

My local Leica store tells me

 

- The SL has been outselling M cameras by quite a bit.  Oftentimes it is sold to a current digital M users for use with their M lenses.  Sometimes, but less frequently it seems, it is sold to be used with R lenses.

 

- A lot of SL users buy the SL without the 24-90 zoom, roughly 50% or more by their rough estimate.

 

Personally, I was interested in buying the SL only if I would buy the 24-90 too. I wanted access to a modern, versatile and productive setup for what I used to shoot with my R system.  The SL doubles as a great camera for using M lenses, but if I were only to use M lenses the M262 would be a better camera for me. I'm really happy with the choice and the opportunities that the SL + 24-90 opened up. 

 

 

We are truly all different in our predilections. 

 

For me, I already had the M-P typ 240. I love shooting with Leica Ms, but only really with a short list of focal lengths ... 35-50-75 most of the time. To me, that's where they sing best. 

The SL replaces my other versatile SLR and mirrorless cameras, and lets me put my R lenses (arguably my best performers) to use. I almost didn't buy the 24-90 but decided it would be good to have. It is worth it. 

 

Other than for testing purposes, I haven't had much occasion to want to fit the M lenses on the SL. 

[cpclee]

I agree with that. While on the large and heavy side (but really not too terribly so), it is an extremely versatile lens with totally modern performance and convenience features and benchmark opticalquality.  I can see my self going on the next vacation with just the SL + 35 FLE and 24-90.  And I'll probably have the 24-90 on most of the time except mounting the 35 FLE when I want compact or shallow DOF. (Low light is not a reason to use the 35 FLE because with OIS I think the 24-90 actually matches the 35 FLE in low light.)  Just these two lenses can cover 99.9% of all my shooting situations.  Back in the R days, I needed 3 lenses to cover this most frequently used focal length range of mine that the 24-90 alone now covers. (I typically brought with me the 21-35/3.5-4 ASPH, 50/2, and 80/1.4.)

 

 

I almost didn't buy the 24-90 but decided it would be good to have. It is worth it. 

 

Edited March 3, 2016 by cpclee

[cpclee]

I'm waiting to see what Leica does with the R to SL adapter to decide whether to buy back into the R lenses. If they give the adapter automatic aperture stop-down, then there is a strong reason to use R lenses over M lenses. In which case I'd snap up a 50 Summicron-R instantly as the alternative travel prime to go with the 24-90.

[scott kirkpatrick]

I understood digital correction in this context (as with the SL24-90) to be a matter of the lens being designed for the software (or whatever it's called) in the camera so that no restoration is required, so it's not a matter of covering things up, but designing for the best possible performance. It's different from the sort of correction you'd do in PS or LR.

 

Am I wrong?

 

There's a small bit of confusion here.  Distortion correction has always been possible in post processing (at least for pure barrel and pincushion distortion).  What Leica is now doing is to put the parameters for a sixth-order polynomial distortion curve (as a function of distance from the center of the image) for each color into the DNG files.  The parameters depend on the lens used (that's why we need usable R profiles!) and for the 24-90 zoom lens depend on the focal length in use as well.  A sixth order polynomial can describe mustache distortion.  Slight differences among the three colors capture certain color aberrations as well.  The whole structure is defined in Adobe's standards for TIFF and DNG, and all raw file developing programs can apply these corrections.  It shows up as the (optional but on by default) lens profile in Capture One.  For JPEGs, the same corrections are done as a final stage in camera, but since this correction is done after the R, G, and B signals have been demosaiced, for raw files it must be left to postprocessing, and is usually automatic.  Other vendors (Olympus, for example) also do this, but if they have a proprietary raw format, it takes a little longer after a new model ships before the controls are effective in third party software.

 

So if the correction is covered by general standards, it doesn't even matter if it is done in camera or by standards-aware software.  That's a big difference from the film era.

 

scott

Edited March 3, 2016 by scott kirkpatrick

[ramarren]

I don't know what confusion you're referring to, scott. Yes, the sixth order polynomial correction tailored per lens and stored in DNG files (adopted, by the way, by Adobe in collaboration with Olympus and the Micro-FourThirds consortium after they had implemented it in the Panasonic Lumix G and Olympus Pen line cameras) can do excellent correction. Putting it into DNG standardized the format and protocol and made it more accessible for others to take advantage of it. 

 

It still makes sense for lens designers to tune the lens designs to make the corrections less lossy and optimize performance. Digital cameras are low-power, specialized computing devices and can only do so much in-camera while providing the responsiveness and performance that photographers want in their hands. What they can do today is astonishing, but it isn't all encompassing. It's designing the system such that it has the right balance and coordination between all the parts that produces the best results. 

[scott kirkpatrick]

The confusion that results when people start distinguishing "in camera" (good) corrections from "post processing" (bad).

 

scott

[ramarren]

The confusion that results when people start distinguishing "in camera" (good) corrections from "post processing" (bad).

 

scott

 

 

I'm not sure that's what Peter H meant, although his statement is imprecise:

 

"I understood digital correction in this context (as with the SL24-90) to be a matter of the lens being designed for the software (or whatever it's called) in the camera so that no restoration is required, so it's not a matter of covering things up, but designing for the best possible performance. It's different from the sort of correction you'd do in PS or LR.

 

Am I wrong?"

 

 

The SL24-90 lens is designed to be used along with lens correction software for optimum results. That means that the lens' firmware communicates to the camera body, identifying the lens and delivering a set of parameters to be use for correction with each exposure. Characteristics like lens opening, focal length, and focus setting might all be dynamically represented by different correction parameters on an exposure by exposure basis. The camera body's imaging engine has software that interprets these parameters and uses them in the in-camera JPEG engine to optimize results. It also injects them into the raw data output (which in the case of the Leica SL are DNG files) for use by external raw converters that honor and implement the Leica (manufacturer's) lens corrections. Adobe's raw converter (both Camera Raw plugin and Lightroom) honor and implement these corrections when they are present in the raw files, as do some others.

 

Other raw converters (like DXO, for one) do lens correction as well but implement their own correction software and parameters based on identifying lens information supplied in the EXIF and their own lens testing, ignoring the manufacturer's correction parameters. Even Adobe's raw converter can use corrections of this sort as well: the "Lens Profiles" in Lightroom and Camera Raw are contributed by users, Adobe, or the lens makers after the fact of the lens  design, independently. 

 

I think what Peter meant by "It's different from the sort of correction you'd do in PS or LR." is these latter kinds of corrections, not supplied by the manufacturer or embedded in the lens design process. In other words, applying corrections based on the data presented by the lens which has had no influence on the lens design, rather than designing the lens with specific kinds of adjustments and correction algorithms in mind, and supplying parameters to run those correction algorithms with. 

 

So it's not the case of distinguishing "in camera (good)" corrections or "post processing (bad)" corrections. The digital correction design of the SL24-90 we've been talking about is the notion of optimizing lens designs for use with specific software corrections and embedding the parameters for those corrections with each exposure, rather than considering lens design to be wholly separate and decoupled from software corrections. 

 

It should also be clear that neither one nor the other is "good" or "bad". Each of these approaches to lens design and lens correction software has its plusses and limitations, and poses benefit to the users when used adroitly. 

[farnz]

While the matching of in-camera digital software correction with specific lenses is intended to improve the images that the camera and lens produce, and I can see the benefit in that, it also inexorably drives the lens to only be used with the camera (or cameras) that run that software.  When I bought my R lenses I knew that they would produce the same performance on a wide variety of cameras from a number of different brands so I object if the 24-90, for example, will give sub-par performance with other cameras because it's not optically as good as it might be because lens designers are content to allow software to 'fill in the blanks'.  Transportability is very important to me especially as I expected lenses to easily outlive digital cameras.

 

Pete.

[Advertisement]

Hallo,

Look here: Leica L-Mount

[cpclee]

See your point, but not sure if that's practical or feasible today.

 

Just think the M lenses.  Every digital M applies some digital correction to fix corner performance, color fringing, vignetting etc.  If you ever mount M lenses on other manufacturers' bodies you'd know performance loss is significant without these corrections.  

 

So whether a lens is designed with or without digital corrections in mind is almost of secondary importance at this stage, because you cannot get by without significant digital corrections with any lens or any camera today.  Unless you shoot film.

 

While the matching of in-camera digital software correction with specific lenses is intended to improve the images that the camera and lens produce, and I can see the benefit in that, it also inexorably drives the lens to only be used with the camera (or cameras) that run that software.  When I bought my R lenses I knew that they would produce the same performance on a wide variety of cameras from a number of different brands so I object if the 24-90, for example, will give sub-par performance with other cameras because it's not optically as good as it might be because lens designers are content to allow software to 'fill in the blanks'.  Transportability is very important to me especially as I expected lenses to easily outlive digital cameras.

 

Pete.