Anyone else liking their Q images better with SDC turned off or applied at less than 100%?

[Luke_Miller]

Advertisement (gone after registration)

I subscribe to Sean Reid's (ReidReviews) site and he is a big proponent of photographers deciding whether to allow the imaging processor to apply the camera's Software Distortion Corrections (SDC) or not.  Many of his lens tests show side-by-side results of SDC on and off.  He shows the Q lenses (28mm and 43mm) to both have barrel distortion that is corrected by SDC.  The barrel distortion of the 28mm is significant.

When shooting events with my Q2 I tried to compose the shots so that my subjects were not close to the edges of the frame, due to the effect (I assumed) of perspective distortion when shooting at 28mm.  People there were distorted and looked unnatural. Often I would activate the 35mm frame lines in the EVF to help me keep people away from the edges.  Since I have Capture One I have the option of disabling the SDC corrections.  Once I did this the problems of distorted subjects at the edge of the 28mm frame went away. The problem was entirely due to the SDC correction being applied.  In most of my shots the uncorrected barrel distortion is not an issue.  Occassionaly in intereior shots there are features in the room that reveal the barrel distortion.  Then I use the distortion slider to apply less than 100% correction, to find a compromise setting that works for both the room and the people.

Unfortunately Lightroom automatically applies 100% of the correction with no option to disable it. However ACR in the Optics Panel does and allows the user to set the amount to be applied.  The ability to disable, or moderate, the SDC corrections resolves the biggest frustration I found with my Q2.

[pgh]

Not Q but I generally do not apply lens corrections to my M’s. I won’t say never, but it’s rare and only in specific circumstances.I tend to only shoot 35mm and 50mm so the effect is minimal, and my reasons are as much philosophical as anything, but to your point - I generally see no real benefit to using them. I’d imagine it’s quite different with other lenses and cameras and have seen as much (for instance, Olympus M/43 lenses have what I felt was a staggering amount of distortion before the lens profile was applied - one reason among a few that I sold that kit). The biggest issue I get with what I currently use (M10 iterations, 35, 50 crons mostly) is vignetting, but sometimes a natural vignette is no problem for me or even welcome. 

[ckern]

1 hour ago, Luke_Miller said:

Unfortunately Lightroom automatically applies 100% of the correction with no option to disable it.

The Aspect control in Lightroom's Develop module offers some control over wide angle distortion at the left and right edges of the frame.  It's a blunt instrument, though.

If you're comfortable using command-line tools, the following should do what you want more cleanly:

   exiftool -overwrite_original -OpcodeList3= L???????.DNG

Note the space after the "-OpcodeList3=" argument.  Use this at your own risk.  Since you are modifying the metadata for the image, it's obviously advisable to do this on a renamed copy of the original file.

This technique works well on a Q3 raw.  I have not tested it on a file from a Q2.

It's possible to batch-process multiple files, but this is the safest way to test this approach and decide whether it works for you.

Edited June 4 by ckern

[UliWer]

vor 2 Stunden schrieb pgh:

Not Q but I generally do not apply lens corrections to my M’s.

Well, digital lens correction for the Q is not the same thing as for an equivalent M lens.

M lenses - which have to deliver also for film when no digital correction is possible - are optically corrected against distortion. So if you apply digital lens correction for them it doesn't make any significant difference.

The lens of a Q can only be used with a digital camera. So digital correction of distortion is an option and Leica chose to make use of it. Without the digital correction you get an image of an uncorrected lens. This means the optical faults which naturally result from the form of the lenses are fully visible.

Though distortion caused by a lens is not the only reason for an image which looks distorted. The scene you take a photo of is naturally always three dimensional. The photo has only two dimensions. So there is also some sort of "natural" distortion, which becomes more obvious if the field of view gets wider.

To get an impression one might compare the results of the same scene taken with an M-lens which is optically corrected against distortion and a Q lens - with digital correction applied in one case and shut off in the other. In all three cases the "natural" distortion caused by the projection of a three dimensional scene to a two dimensional photo will be visible. Perhaps the uncorrected version with a Q lens will even show the more pleasing results. Though I think this is merely accidental as two forms of distortion - the natural one from the scene and the optical one from the lens - neutralize each others. In other cases the optical distortion of an uncorrected lens will look like what it really is: an optical fault.   

[Le Chef]

For my Q's it would make no sense to turn off SDC. The camera was designed around it so I see no point in not using it. Leica have explained how they correct lenses and their approach makes sense to me. 

It's all why, as Uli implies, that M lenses are so much more expensive.

[ckern]

20 minutes ago, UliWer said:

. . . distortion caused by a lens is not the only reason for an image which looks distorted. The scene you take a photo of is naturally always three dimensional. The photo has only two dimensions. So there is also some sort of "natural" distortion, which becomes more obvious if the field of view gets wider.

Indeed: as far as I know, volume distortion (a.k.a. "edge stretching") is an inherent optical characteristic of wide angle lenses that map the real world onto a flat sensor.  If you want to eliminate it completely, I suspect you would need a camera with a curved sensor.

[pgh]

Advertisement (gone after registration)

31 minutes ago, UliWer said:

Well, digital lens correction for the Q is not the same thing as for an equivalent M lens.

M lenses - which have to deliver also for film when no digital correction is possible - are optically corrected against distortion. So if you apply digital lens correction for them it doesn't make any significant difference.

The lens of a Q can only be used with a digital camera. So digital correction of distortion is an option and Leica chose to make use of it. Without the digital correction you get an image of an uncorrected lens. This means the optical faults which naturally result from the form of the lenses are fully visible.

Though distortion caused by a lens is not the only reason for an image which looks distorted. The scene you take a photo of is naturally always three dimensional. The photo has only two dimensions. So there is also some sort of "natural" distortion, which becomes more obvious if the field of view gets wider.

To get an impression one might compare the results of the same scene taken with an M-lens which is optically corrected against distortion and a Q lens - with digital correction applied in one case and shut off in the other. In all three cases the "natural" distortion caused by the projection of a three dimensional scene to a two dimensional photo will be visible. Perhaps the uncorrected version with a Q lens will even show the more pleasing results. Though I think this is merely accidental as two forms of distortion - the natural one from the scene and the optical one from the lens - neutralize each others. In other cases the optical distortion of an uncorrected lens will look like what it really is: an optical fault.   

I’m aware of all of this. What I can’t figure out is what is prompting the lesson.

The digital correction of optical flaws is the exact thing I prefer to avoid - if the lens is itself flawed to the point of requiring significant digital correction (which is why I sold the aforementioned M4/3 setup, and one reason I’ve not purchased a Q) that’s a trait I personally find off-putting. But that’s just me, and I have my reasons. In short, the file you get is less representative of what the camera actually saw. That’s an academic issue for many and for me one of substance. (Let’s not let this devolve into what a true photograph is, I know there’s no good answer - especially with digital equipment - but I’ve drawn my lines where I see fit).

If a Q user goes against the preferences of Leica and finds the optical flaws something they can live with, there might be reasons to do so. 

[UliWer]

Well, one has to cope with the problem that a  lens cannot project a three dimensional object to a two dimensional medium without distortion. Optical correction of this fault is one way, digital correction is another. I don't see a clear distinction that only the optical correction is a "legitimate" way and the digital way "illegitimate".

There is a more complex relation between optical correction of distortion and correction of different optical faults. It is generally accepted that the Carl Zeiss Jena 50mm Sonnar lenses were much better than the contemporary Leitz lenses. They were "sharper", more contrasty etc. But they showed much more distortion. Their "better" performance was not alone but to a certain degree achieved by allowing more distortion.

Modern lens design seems to follow this concept. They seem to say: let's ignore distortion and concentrate on the other aspects, as distortion can be corrected digitally. 

It may be interesting to read what the modern lens designers from Zeiss say about this: https://lenspire.zeiss.com/photo/app/uploads/2022/02/technical-article-distortion.pdf

On p. 23 of this article they show examples of photos with converging lines which are not the result of lens aberrations but just of the fact that a three dimensional object is reduced to a two dimensional medium. It may be useful to think about why we don't see the objects this way when we look at them in reality. Are our eyes so much better corrected than the most perfect lenses? No, it's just our brain which does the correction. You might call this "digital" correction as well.   

Edited June 4 by UliWer

[zlatkob]

Yes, It makes perfect sense to sometimes turn software corrections off. When a wide angle lens is fully corrected, people near the edge of the frame can look distorted. Turning off software correction brings out any barrel distortion, but this can have the excellent effect of making people near the edge of the frame look much more natural. Of course, without corrections, straight lines in architecture start to look curved, but this can be a good compromise if the people subjects are the priority.

[pgh]

2 hours ago, UliWer said:

Well, one has to cope with the problem that a  lens cannot project a three dimensional object to a two dimensional medium without distortion. Optical correction of this fault is one way, digital correction is another. I don't see a clear distinction that only the optical correction is a "legitimate" way and the digital way "illegitimate".

There is a more complex relation between optical correction of distortion and correction of different optical faults. It is generally accepted that the Carl Zeiss Jena 50mm Sonnar lenses were much better than the contemporary Leitz lenses. They were "sharper", more contrasty etc. But they showed much more distortion. Their "better" performance was not alone but to a certain degree achieved by allowing more distortion.

Modern lens design seems to follow this concept. They seem to say: let's ignore distortion and concentrate on the other aspects, as distortion can be corrected digitally. 

It may be interesting to read what the modern lens designers from Zeiss say about this: https://lenspire.zeiss.com/photo/app/uploads/2022/02/technical-article-distortion.pdf

On p. 23 of this article they show examples of photos with converging lines which are not the result of lens aberrations but just of the fact that a three dimensional object is reduced to a two dimensional medium. It may be useful to think about why we don't see the objects this way when we look at them in reality. Are our eyes so much better corrected than the most perfect lenses? No, it's just our brain which does the correction. You might call this "digital" correction as well.   

We all need to make decisions somewhere along the line. When you get a photograph from an optically corrected lens, it is still a result of what the lens sees, yes? Just like bokeh is an artifact of optics and not an accurate representation of a three dimensional space. This is commonly recognized by viewers of photographs. There’s an inherent conceit that of course a lens is not rendering reality as it actually is.
 

Digital correction isn’t just different in kind it’s different in spirit. It’s more about an engineer’s idea of what “correct” is which is generally based on things like consistent exposure and the way that lines are depicted near the edge of the frame. 

As “digital” correction being likened to how our brain resolves 3d space into images I think maybe the more analogy would be more accurate if it was discussed as “someone else’s brain” doing the correction and not one’s own. 

I made no claim to what’s legitimate or not - I made a point to say I didn’t want to go there because it an irresolvable argument I’ve no interest in litigating. I’ve my own ethics about that I won’t impose on anyone else, but am only here to say that there are reasons it makes sense to avoid this feature that go beyond aesthetics.