Image distortion in JPEG

[Susurs]

Hi!

Today I noticed that my Sony SEL18200 produces barrel and pincushion distortion (See the attached image @100mm). Reviews of the lens state that it is "supposed to be like this", respectively it is not a defect of a particular lens but it is as it is for all SEL18200 lenses.


http://www.imaging-resource.com/lenses/sony/e-18-200mm-f3.5-6.3-oss-sel18200/review/


What I was thinking is, however, that the lens compensation settings in camera would take care of this. I thought that the review is done without the compensation settings ( ? )


My question - Is this the reality what I can expect after the in-camera image compensation, or there is something wrong additionally with the specific lens (image compensation is not working - it seems to be enabled).

 

[dotnet]

In-camera conversion is usually the best place to correct for distortion, CA and possibly other lens characteristics, as long as the lens is from the camera manufacturer and/or known to the camera firmware.

There are plug-ins for image processing software that can do a similar job. They usually cost extra, and you have to make sure that whatever you buy supports your camera/lens combinations.

If you use in-camera correction the results should be very obvious when your compare the RAW and JPEG images from the camera, especially with wide-angle lenses. If they're not then something isn't working.

 

[kenoh]

Also, in real world shooting i.e. as long as you are not photographing grids and brick walls, then it should not be noticeable in the vast majority of shots. You then have Adobe Lightroom profiles for lens corrections too.

I wouldnt sweat this one.

 

[Susurs]

Thanks! Yeah, it most probably will not be noticable in real life outdoors. Still ... is it supposed to be normal that a JPEG looks like this after in-camera image correction and there is some uncorrected geometry?

 

[Stefan johansson]

The slight distortion is caused by the optics,and it's a typical thing that get worse with shorter lenses,all wide angle lenses have more or less distortion. The "unsharp background" seen in the pic,can easily be changed if you adjust aperture setting.

 

[kenoh]

Thanks! Yeah, it most probably will not be noticable in real life outdoors. Still ... is it supposed to be normal that a JPEG looks like this after in-camera image correction and there is some uncorrected geometry?

It is a compromise like everything else. This image was probably taken as it was the one that showed the distortion the most - to make a point. Remember the optical correction is a computer algorithm so it isnt infallible. It is a case of "close enough".

The uncorrected geometry if there is some is because you are trying to make a lens group optimised for one focal length give good performance at another and all points in between.

With all due respect though, the lens wont be the issue with an image you take. Understand I am not saying you are a bad photographer, just that most of us are not printing at billboard size or professional editorial so it is just not that big a deal for us. If it is a big deal you need to look at the high end lenses, the likes of G-Masters as this is the kind of thing they are corrected for and carry a price tag to match. You will also see that they dont have ranges of focal lengths like this as it is nigh on impossible to get the optical perfection expected - until we invent glass for a lens that works like the lens in your eye and can consistently dynamically change shape.

 

[Susurs]

Thank you all for helping me! I really appreciate it!
The test image was just a fast random shot not thinking of any other settings - so it came out as it is (bad quality)

I went to a local shop and the excellent staf gave me a SEL18200LE, and A6000 body. I tested 4x - 2x with 2 lenses on my camera, 2x with both lenses on the A6000. And the result was that there is the same distortion within all 4 images! A bit of difference with LE, but still the same. Also it does not matter whether the image was in RAW or JPEG.

Hmm... I was thinking that a lens which is not a cheap lens, will not have any such problematic aspects. To tell the truth, I like my SEL50F18 much more , but sometimes I want to get closer to wildlife as well...

 

[TheDrift-]

If you want something with extremely low distortion be prepared to pay for it.

This lens has extremely low distortion it's a 50mm that only goes to F2 it also costs around $7000 dollars.

https://diglloyd.com/prem/s/LEICA/LeicaM9/lens-Leica50f2-APO-ASPH-distortion.html

Generally the bigger the zoom range the slower/distortion/poorer IQ you get. Spending big $$$$ can mitigate some of the issues.

Most of us correct in Lightroom or another converter when needed...in real life it's less of an issue.

 

[Susurs]

For 50mm I have SEL50F18 which is an excellent lens and have no noticable distortion at all.

 

[TheDrift-]

For 50mm I have SEL50F18 which is an excellent lens and have no noticable distortion at all.

Primes will tend to have a better IQ as they are a fixed focal length.

Zooms tend to have more optical compromises but have the connivence of a zoom.

You might find the distortion varies along the focal length with some focal lengths faring better than others

 

[dotnet]

All lenses will show either rectilinear distortion (RD), angular magnification distortion (AMD) or a mix of both. As a lens designer you can only correct for one at the expense of the other, but not for both at the same time.

RD is the familiar pincushion distortion. With AMD all lines in a brick wall image are straight, but the bricks in the centre are larger than those at the edges. Both RD and AMD can be corrected for in software, and with knowledge of the lens characteristics that's exactly what the in-camera conversion does.

In general, lenses with longer focal lengths show less RD/AMD since their image plane is inherently less curved. The compromise between RD and AMD is therefore mostly a problem for designers of wide angle lenses.

 

[dwig]

...

RD is the familiar pincushion distortion. With AMD all lines in a brick wall image are straight, but the bricks in the centre are larger than those at the edges. Both RD and AMD can be corrected for in software, and with knowledge of the lens characteristics that's exactly what the in-camera conversion does.
...

The distortion in some lenses is quite complex. While in theory, such distortion can be corrected in software, there is a vastly higher need to make the camera work quickly, be compact, and take a large number of pictures on a single charge. Given the limited processing power of a camera's microprocessor (primarily limited by the need to be have low power demands) and this need to process the image very very quickly, the correction algorithms are "quick and dirty" compromises. They do a very good, but not perfect, job. Generally, they correct only the average barrel/pincushion distortion (which sometimes leaves a "mustache distortion" component uncorrected as seen in iPhone images) and not dealing with any AMD.

 

[dotnet]

The distortion in some lenses is quite complex. While in theory, such distortion can be corrected in software, there is a vastly higher need to make the camera work quickly, be compact, and take a large number of pictures on a single charge. Given the limited processing power of a camera's microprocessor (primarily limited by the need to be have low power demands) and this need to process the image very very quickly, the correction algorithms are "quick and dirty" compromises. They do a very good, but not perfect, job. Generally, they correct only the average barrel/pincushion distortion (which sometimes leaves a "mustache distortion" component uncorrected as seen in iPhone images) and not dealing with any AMD.

Well, that depends on the camera, with DSLRs the image quality is a prime concern. They use custom chips designed to do exactly these kinds of image transformations, based on a bunch of parameters describing the lens characteristics. And yes, they do correct AMD. One of my favourite wide-angle zooms is well corrected for RD but shows AMD. This is not immediately obvious in typical scenes, but when I flick between the raw and JPEG versions of an image it pops out.

 

[Stefan johansson]

Well, that depends on the camera, with DSLRs the image quality is a prime concern. They use custom chips designed to do exactly these kinds of image transformations, based on a bunch of parameters describing the lens characteristics. And yes, they do correct AMD. One of my favourite wide-angle zooms is well corrected for RD but shows AMD. This is not immediately obvious in typical scenes, but when I flick between the raw and JPEG versions of an image it pops out.

All optical lenses has some distorsion. Of course a shorter focal length has more,but overall there is always some. This side effect has been known ever since mr. Daguerre invented the first photography equipment.
Most professional photographers know how to use the lens distortion to get more interesting pictures.

 

[dotnet]

All optical lenses has some distorsion. Of course a shorter focal length has more,but overall there is always some.

Yes, that's what I said.

This side effect has been known ever since mr. Daguerre invented the first photography equipment.

In fact, this knowledge goes back to the 17th century, following the invention of the telescope and microscope, and predating the invention of photography.

 

[Stefan johansson]

Yes, that's what I said.



In fact, this knowledge goes back to the 17th century, following the invention of the telescope and microscope, and predating the invention of photography.

Yes, that's what I said.



In fact, this knowledge goes back to the 17th century, following the invention of the telescope and microscope, and predating the invention of photography.

Exactly,and mr Daguerre invented the first "photography" in 1807.

 

[dwig]

Exactly,and mr Daguerre invented the first "photography" in 1807.

Quite incorrect. Daguerre did not "invent photography", nor did he make any significant contributions as early as 1807, which is over 15 years before Daguerre even began his very first experiments with creating a camera produced image.

Actually, Joseph Nicéphore Niépce took the first "permanent" (it still exists) camera produced photographic image in 1826 or 1827 (date is a bit uncertain). He later partnered with Louis Daguerre and the two worked to create a reliable process. Nicéphore Niépce died before they were successful, and Daguerre continued on his own, taking full credit for the result when he published is work in 1839.

That same year, just months later, Henry Fox-Talbot published his work. At that time, he had been successfully producing photographic images for a number of years, building on earlier work by Thomas Wedgwood and others. Several of his images dating from the early to mid 1830s still exist.

The Daguerreotype was quite successful for a number of years, but the process left no technological descendants. Fox-Talbot's process, on the other hand, is the "father" of all future silver based photographic processes up to and included today's remaining film processes.

 

[monokakata]

Unless I missed it, no one has talked about post processing, and that might be the key for you.

I have a variety of lenses (Nikon, but that's unimportant) and I've downloaded lens corrections from Nikon and installed them in Lightroom CC. I don't have the lens corrections set to operate on import, which means that when I call for them, I can see the results immediately, and there are almost always some visible corrections.

I suggest that, if you're not using Lightroom, that you download a trial, then explore what correction plugins Sony has available, install them, and see whether your distortions disappear or are at least minimized.

I'm suggesting that there's another level of correction that you haven't explored.