SL 24-90 digital corrections

[dgktkr]

Advertisement (gone after registration)

After reading some discussions about digital corrections on SL lenses, I was motivated to do an experiment to see for myself how large those corrections might be. So I took my SL 24-90 and made an image of a sheet of graph paper with Siemens stars. I shoot raw and the software I like is Iridient Developer. The image was made with a focal length of 24mm and near the close focus limit -- 0.3m. The camera was on a sturdy tripod and the graph paper was cut to just barely fit into the view in the viewfinder. The graph paper is lying on an oak footstool and, as you can see, it is very slightly rotated.

 

Here is a screenshot of the Iridient window that has the DNG embedded lens corrections enabled + 8 more units of distortion dialed in because the embedded correction wasn't quite right.

 

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!

 

dgktkr

Edited February 11, 2017 by dgktkr

[dgktkr]

Here is the uncorrected image (obtained by simply unchecking the "Enable lens profile" box and unchecking "Enable distortion correction"). It is pretty clear that the sensor sees more that is shown by the program when the image is corrected. You can also see that the image gets pretty dark in the corners.

 

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!

 

dgktkr

[Advertisement]

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

[dgktkr]

Here are 100% crops from the corrected and uncorrected images. Blue (purple?) fringing is prominent in the uncorrected one, while even the corrected one shows some color at the edges of the black features.

 

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!

 

 

dgktkr

[Alo Ako]

Hm, fine. And what is your conclusion?

[dgktkr]

Hm, fine. And what is your conclusion?

 

Sometimes the embedded corrections are not quite right and can be tweaked for better results.

 

I'm happy to allow digital corrections because it reduces the constraints in the physical lens design (and fabrication) and the end result is better than if they weren't allowed.

 

Without digital corrections, I wouldn't be using this lens.

 

I'm not convinced we should blame Leica (and Adobe) for forcing us to use digital corrections against our will.

 

dgktkr

[tom0511]

The 24-90 is continuing to impress me. Over the months I have compared it with several M and R primes and almost allways the result was the zoom was just as good if not better. As long as the software corrections do not lead to mushy corners or other visible image degradation (I am not talking about pixel peeping but real world images) I am happy to have this function. 

[Alo Ako]

Advertisement (gone after registration)

The 24-90 is continuing to impress me.

 

Same to me. I love the results.

[Guest VVJ]

+1.  I simply do no get these never-ending threads about digital corrections.  What does it matter if the end result is excellent?

[ramarren]

+1.  I simply do no get these never-ending threads about digital corrections.  What does it matter if the end result is excellent?

 

 

The answer is: It doesn't matter at all. 

[zlatkob]

+1.  I simply do no get these never-ending threads about digital corrections.  What does it matter if the end result is excellent?

 

It's very simple.  If you photograph a group of people with some people near the edge of the frame, those who are near the edge of the frame get their heads stretched — made more oblong by digital correction of barrel distortion.  Does that sound "excellent" — s-t-r-e-t-c-h-e-d heads?  For some professionals, that result is not excellent; rather, forced digital correction is a form of forced digital distortion with a distinctly unappealing result.  It is something anyone can see if they took a little time to educate themselves.  But it seems that a few of those who don't understand the issue just assume that it doesn't exist.  If it doesn't matter to them, it can't matter to anyone else — A distinctly self-centered view.  Thankfully, a few people do understand that if you correct for one form of distortion, you increase another form of distortion.  Those who understand it tend to agree that the optimal solution is to give the photographer full control of the image, allowing for uncorrected barrel distortion when it suits a particular image.

[ramarren]

zlatkob,

 

You've said this same thing four or five dozen times. 

I've hunted through thousands of exposures looking for this 'stretchy head' effect and have yet to find it. And I have TONS of frames of people filling the frame from side to side with all kinds of lenses. 

 

I do find 'stretchy head' (and arm, and body, and whatever) with wide angle lenses (even 35mm can produce it): it's part of the issues of using a wide angle lens with a crowded row of people due to wide angle perspective distortion. This has NOTHING to do with digital corrections and EVERYTHING to do with the geometry of a 2D planar projection of the three dimensional scene. 

 

The SL24-90 and SL50 digital corrections are part of the lens design. They result in the lens producing results IDENTICAL to the results of a finely tuned lens of the same focal length without ANY digital corrections. Period. 

 

You will likely go on and on and on about this anyway, but the notion that some "few of those who don't understand the issue" is just a condescending way of saying "I'm convinced I'm right about this and the rest of you are too stupid to see it." Thanks, we get it. 

[IkarusJohn]

It's very simple.  If you photograph a group of people with some people near the edge of the frame, those who are near the edge of the frame get their heads stretched — made more oblong by digital correction of barrel distortion.  Does that sound "excellent" — s-t-r-e-t-c-h-e-d heads?  For some professionals, that result is not excellent; rather, forced digital correction is a form of forced digital distortion with a distinctly unappealing result.  It is something anyone can see if they took a little time to educate themselves.  But it seems that a few of those who don't understand the issue just assume that it doesn't exist.  If it doesn't matter to them, it can't matter to anyone else — A distinctly self-centered view.  Thankfully, a few people do understand that if you correct for one form of distortion, you increase another form of distortion.  Those who understand it tend to agree that the optimal solution is to give the photographer full control of the image, allowing for uncorrected barrel distortion when it suits a particular image.

 

 

Leaving the not so thinly veiled suggestion that those who haven't spotted this problem are "uneducated", if what you say (so many times) is true, then surely it would be apparent in the first image posted above?  If a head is going to be s-t-r-e-t-c-h-e-d near the edges, those perfect little squares near the edges of dgktkr's photo above would be r-e-c-t-a-n-g-u-l-a-r?  The squares seem to be perfectly formed all over the image ...

 

I can't remember if you have an SL with either zoom or the 50, but I would have to say your argument would carry more weight if you could produce an image taken with the SL and either the 24-90 or the 50 showing the effect you claim.  I guess, it's "put up or shut up" time - otherwise, your statements seem to be "assumption" rather than anything evidence based.

 

Don't get me wrong, I am genuinely interested.  Perfect square at the edge, taken at 24mm looks pretty remarkable to me ...

[Guest VVJ]

It's very simple.  If you photograph a group of people with some people near the edge of the frame, those who are near the edge of the frame get their heads stretched — made more oblong by digital correction of barrel distortion.  Does that sound "excellent" — s-t-r-e-t-c-h-e-d heads?  For some professionals, that result is not excellent; rather, forced digital correction is a form of forced digital distortion with a distinctly unappealing result.  It is something anyone can see if they took a little time to educate themselves.  But it seems that a few of those who don't understand the issue just assume that it doesn't exist.  If it doesn't matter to them, it can't matter to anyone else — A distinctly self-centered view.  Thankfully, a few people do understand that if you correct for one form of distortion, you increase another form of distortion.  Those who understand it tend to agree that the optimal solution is to give the photographer full control of the image, allowing for uncorrected barrel distortion when it suits a particular image.

 

 

I didn't think there was a need to get patronizing or arrogant...

 

I honestly have not seen that effect in over 20K pictures.

 

And if you don't want to sound like a broken record proof that the issue exists and then people will take you seriously.

[scott kirkpatrick]

There are a number of excellent lenses that I have used or seen used in which the designers chose to let software corrections add the final corrections.  Besides the SL's 24-90 below 50 mm focal length and the SL 50, there are the Panaleica 25/1.4 for micro 4/3, and the Fuji compact 23 and 35/2.0 lenses for the X-Pro2.  In the latter cases the manufacturer's recommended corrections are buried inside a proprietary raw format.  As a result, I understand that Adobe does the corrections by default, and Capture One permits them to be overridden or tweaked, but prefers to ignore them completely unless users complain and cause Phase to develop their own corrections.  Curiously, the SL now outputs LCA and distortion corrections for some classic Leica M and R wide angle lenses and R zooms in its DNG files, using a non-proprietary opcode format, but the M10, using the same DNG 1.4 format, does not.  Instead the M10 provides a template indicating the Adobe proprietary files in which any corrections will be found, which is no help to Adobe-avoiders and probably a loss of some useful information.

 

I don't care for lines which curve but should have been straight at the edges of the frame, but I have seen examples where people at the corners (not the edges) of my PL 25/1.4 look more natural with the distortion retained, so I would prefer to always have the correction overridable. The other argument against software correction that you will hear is that since SDC involves interpolating the as-captured pixels at the edges to get the true, corrected image, the corrected image will have lower resolution.  I've done experiments to test this, and don't see it.  I think there are two reasons:  the as-captured pixels are not matched in any way to high frequency components of the actual image, so a smart interpolation scheme will introduce minimal additional artifacts; and the chromatic aberration corrections (different distortion corrections for each color) may actually sharpen the image by a sort of software APO correction.

 

scott

[zlatkob]

zlatkob,

 

You've said this same thing four or five dozen times. 

I've hunted through thousands of exposures looking for this 'stretchy head' effect and have yet to find it. And I have TONS of frames of people filling the frame from side to side with all kinds of lenses. 

 

I do find 'stretchy head' (and arm, and body, and whatever) with wide angle lenses (even 35mm can produce it): it's part of the issues of using a wide angle lens with a crowded row of people due to wide angle perspective distortion. This has NOTHING to do with digital corrections and EVERYTHING to do with the geometry of a 2D planar projection of the three dimensional scene. 

 

The SL24-90 and SL50 digital corrections are part of the lens design. They result in the lens producing results IDENTICAL to the results of a finely tuned lens of the same focal length without ANY digital corrections. Period. 

 

You will likely go on and on and on about this anyway, but the notion that some "few of those who don't understand the issue" is just a condescending way of saying "I'm convinced I'm right about this and the rest of you are too stupid to see it." Thanks, we get it. 

You do accept that there is stretching of heads inherent in wide angle photos.  But you don't seem to accept that digital correction of barrel distortion worsens the stretching of heads near the edges of the frame.  Add barrel distortion and it improves — people near the edges look better (at the cost of bending straight lines).  Remove barrel distortion and it worsens — people near the edges look worse.  It's a simple thing that anyone with Lightroom can test by moving a slider on an appropriate image.

 

About my tone in the previous post — it's a direct response to the condescension I've received.  I've objected to forced digital corrections because I feel that photographers should have the option to turn them on and off based on how they want to express a particular image.  That's a very a defensible photographer-centered position, one that I would think photographers would support.  And Sean Reid expressed the same point of view for the exact same reason (people near the edges of the frame).  But for this I've been treated like some know-nothing in other threads.  

 

Leaving the not so thinly veiled suggestion that those who haven't spotted this problem are "uneducated", if what you say (so many times) is true, then surely it would be apparent in the first image posted above?  If a head is going to be s-t-r-e-t-c-h-e-d near the edges, those perfect little squares near the edges of dgktkr's photo above would be r-e-c-t-a-n-g-u-l-a-r?  The squares seem to be perfectly formed all over the image ...

 

I can't remember if you have an SL with either zoom or the 50, but I would have to say your argument would carry more weight if you could produce an image taken with the SL and either the 24-90 or the 50 showing the effect you claim.  I guess, it's "put up or shut up" time - otherwise, your statements seem to be "assumption" rather than anything evidence based.

 

Don't get me wrong, I am genuinely interested.  Perfect square at the edge, taken at 24mm looks pretty remarkable to me ...

That's a fair comment.  I don't have an SL so I'm not going to post samples (and wouldn't post people photos if I did have one).  The SL is just a camera that I've considered buying.  I feel that the issue of forced corrections transcends this particular camera and raises a broader question of design philosophy.  There are inevitable tradeoffs when manufacturers use software to enforce perfection.  I don't think Sean Reid owns an SL either, but that doesn't stop him from expressing reasoned design preferences.  For my photos of people, I don't want forced digital corrections from any camera.

 

The problem with the perfect squares is that if you photograph a group of people, the people near the edges of the frame actually look better, more realistic, in the uncorrected version — the version in which the squares are slanted and curved.  Strange but true.  In a group photo, people near the edges look a little wider than those in the center.  With digital correction, they look even wider, i.e. fatter.  Add (or restore) some barrel distortion, and that effect is reversed.  At 50mm the effect is subtle.  At wider focal lengths it is more obvious.

Edited February 13, 2017 by zlatkob

[IkarusJohn]

Hmm, so it's all assumption then.

 

Me? I'm not worried about your philosophical reservations or your theories; I have the camera, and I don't seem to be experiencing what you describe. What I don't understand is why you post so much about a problem you don't have with a camera you don't own; a problem which no one who has the camera seems to have experienced.

 

As I say, I'd sit up and take notice if you could produce a single image taken with the SL and the SL 50 or the zoom between 24 and 50 which displays the problem. So far, we have the digitally corrected image taken at 24 showing perfect squares - which would suggest undistorted heads.

 

Walks like a duck, quacks like a duck ... what do you think? Sounds like bullshit to me.

[scott kirkpatrick]

Sounds like feeding a troll to me.  I was out of town and didn't realize how repetitive and circular this discussion had become, spread over several threads.

 

I think there is an issue in the difference between how the SL and the M10 are handling lens corrections.  

 

scott

[tom0511]

Overall I also would welcome if it was possible to de-activate distorsion correction. I doubt that I would deactivate the function often, but then everybody could choose himself.

By the way for some M lenses and S lenses I often leave coorection switched off, not because I see degradation in corrected images, but because in many case/subjects I dont really see any problems without correction.

[pop]

Questions to all those ladies and gentlemen who prefer deactivating one kind of distortion control over properly correcting the lens and applying the desired effects later:

 

What makes that particular distortion more desirable than other lens-induced artefacts such as pin cushions, chromatic aberrations, coma or flare?

How often does the arbitrary omission of one kind of correction exactly match your photographic needs?

What makes you think that applying two transformations in a row has a negative effect on your image?

If you think applying two transformations in a row is bad, why don't you blame the picture processing software for failing to combine the two transformations into one?

[Advertisement]

Hallo,

Look here: Leica L-Mount

[zlatkob]

Hmm, so it's all assumption then.

 

Me? I'm not worried about your philosophical reservations or your theories; I have the camera, and I don't seem to be experiencing what you describe. What I don't understand is why you post so much about a problem you don't have with a camera you don't own; a problem which no one who has the camera seems to have experienced.

 

As I say, I'd sit up and take notice if you could produce a single image taken with the SL and the SL 50 or the zoom between 24 and 50 which displays the problem. So far, we have the digitally corrected image taken at 24 showing perfect squares - which would suggest undistorted heads.

 

Walks like a duck, quacks like a duck ... what do you think? Sounds like bullshit to me.

 

You're missing the point.  The squares are perfect squares because they are on a flat plane, not a 3-dimensional object.  Photograph a building with a perfectly corrected lens and all of the straight lines will all be straight, assuming the plane of the sensor is parallel to the face of the building (and shifted if needed).  Now stand a group of people in front of the same building, photograph them in the 24 to 35mm range with people filling the frame edge to edge, and the people by the edges of the frame will be wider than the people near the center.  Finally, remove the correction (for barrel distortion), and you improve the appearance of the people, while distorting the appearance of the building.  Again, anyone with Lightroom can test this with the distortion slider.  And again, Sean Reid understands what I'm talking about and addresses it specifically — contrary to your claim that "no one" has experienced this "bullshit".  It's a fundamental problem of optics that exists in every camera system, and there is no perfect solution.  However, manufacturers take different positions with respect to the photographer's degree of control. 

Edited February 13, 2017 by zlatkob