23mm chromatic aberration

[julian m]

Advertisement (gone after registration)

I read quite a few reviews of the 23mm Summicron-T before picking it up, and most noted something along the lines of: "Lateral chromatic aberration remained within the negligible category for all aperture settings." (e.g., Photo Review)

 

But I see a lot of CA, both in the camera JPEGs and whatever Lightroom can do with the RAWs. I've attached downscaled corner crops from two here.

 

Am I just not using this thing right or is something wrong with mine? Most of what I see below does not qualify as "negligible" in my book. I know these are overexposed a bit (usually need to remember to bring the EV down a bit in automatic modes with the T) and have some harsh contrast, but it still seems excessive.

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!

Edited June 22, 2015 by julian m

[julian m]

Here are corner crops from some closer to properly exposed, to show it's not just an overexposure thing.

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!

[Advertisement]

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

[SirPiet]

Hi there,

 

i also had the same "issue" with my 23mm Summicron-T, so i traded it immediately for the Super-Vario-Elmar-T 11-23mm as soon as it was released. Here are two corner crops from my lenses, first one is Summicron, second one is the Vario-Elmar - both at 23mm, although the Vario is stopped down.

 

 

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!

[SirPiet]

Wide angle zoom sample:

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!

[bencoyote]

I saw this too when I first got the 23mm Summicron and I was shocked because it was my first Leica lens and I was expecting the flawless imaging that the lenses have a reputation for. Someone tried to tell me that it wasn't CA but I didn't understand the explanation. Many months later, I found an explanation of the phenomena that made it click. I can't seem to find the page right now but I'll do my best to explain it. Suffice it to say, it is not a flaw in the Leica lens, it is a property of optics and it cannot be fixed by lens design. However, you can quickly fix it in Lightroom and IIRC Leica fixes it in the JPG in the camera.

 

Good lens design will insure that at your focal distance all the wavelengths of light will arrive at the sensor plane together at the same location thus eliminating chromatic abberation. What you are seeing is a related effect. You have a 3D field that you took a picture of and you were using a fairly wide open aperature. Your focal point is such that while the palm trees, masts and leaves are within your depth of field and therefore look reasonably sharp they are toward the edge of that depth of field for your aperture setting and so the differential refraction of the various wavelengths of light begin to show.

 

Hopefully, I explained that well.

 

There are a ton of ways to deal with this:

1) just fix it in lightroom (what I tend to do)

2) use the jpgs rather than the dngs

3) It won't be as bad on a lens with a longer focal length. i.e. the effect is most pronounced on short focal length lenses. 23mm is pretty short. We tend to think of it as a 35mm because that is roughly the field of view but it really is a 23mm lens.

4) Use a smaller aperture i.e. bigger f-stop thus increasing your DOF and then fewer things will be near the edge of the range of distances which are in focus. (the easiest thing to do)

5) Pick a focus distance or a focal point which is closer to the same distance as the high contrast elements like the palm fronds, the masts, or the tree leaves. (probably the smartest thing to do)

6) Avoid high contast areas in your composition

7) If you must have high contrast areas, make sure that they are illuminated by monochromatic lights. The sun unfortunately is incandescent and a black body radiator and thus includes all frequencies of light.

 

and probably a few more which I didn't think of because it is late, I'm tired, and I've had a few beers. ;-) 

[jaapv]

Hi there,

 

i also had the same "issue" with my 23mm Summicron-T, so i traded it immediately for the Super-Vario-Elmar-T 11-23mm as soon as it was released. Here are two corner crops from my lenses, first one is Summicron, second one is the Vario-Elmar - both at 23mm, although the Vario is stopped down.

Which, at least, explains part of the difference.

[SirPiet]

Advertisement (gone after registration)

Sure it makes quite the difference, but who would shoot a landscape wide open anyway? At least at 23mm this makes no sense as there is no selective sharpness anyway.

Edited June 22, 2015 by SirPiet

[bencoyote]

Some approximate numbers to put some detail on this:

Summicron 23mm wide open - 23mm f/2 hyperfocal distance: 37'9" and at 25' DOF 15'-73.8'

Vario-Elmar 18-56 wide open at 23mm f/4 hyperfocal distance: 18'11" and at 25' DOF 10'9"-infinite

Super-Vario-Elmar wide open at 23mm f/4.5 - 23mm f/4.5 hyperfocal distance 16'11" and at 25' DOF 10'1"-infinite

 

So you can see at a guesstimate of your actual focal distance for the first image, the 3D field of actual distances to objects will be distributed around within the available DOF and if you do a plot of how much fringing you see vs distance you can that the it is directly related to how near it is to the limit of your DOF range for the distance at which your lens is focused.

 

Most landscape photographers operate primarily at smaller aperatures and longer shutter speeds to maximize the 3D DOF that is in focus. Hence the frequent use of tripods. I wouldn't actually think of 35mm as a landscape lens but if I were using it as such then I would consider starting around f/8 unless there was a good reason not to.

f/8 HFD 9'6" and focused at about 25' basically everything from 6'10" would be in focus

f/11 HFD 6'9" and everything from 5'3" would be in focus

f/16 HFD 4'10" and everything from 3'11"'would be in focus

 

So part of good landscape technique includes picking your f-stop and focal plane even when selective focus is not being used because secondary effects like this can crop up to the unwary.

[pgk]

This 'phenomenon' has been hotly debated and the causes pondered over, discussed and argued about ad naseum (including here on this forum - do a search - and I do mean ad naseum!). I note it every so often on a variety of lenses (old and new) and have still not found an explanation which covers ALL its occurrences and characteristics. But, it can be dealt with in software, usually to a reasonably satisfactory level especially if the shot is important and really must not show such colour fringing. What I would say though is that its not a 'flaw' in the lens design or an individual lens, in my experience it 'strikes' both good and less good lenses too, when all the varied parameters which influence its creation are obviously present. I've no interest in starting another debate about its cause, but would rather just say that we have to live with it and deal with it, which we can .

[julian m]

 

 

There are a ton of ways to deal with this:

1) just fix it in lightroom (what I tend to do)

2) use the jpgs rather than the dngs

3) It won't be as bad on a lens with a longer focal length. i.e. the effect is most pronounced on short focal length lenses. 23mm is pretty short. We tend to think of it as a 35mm because that is roughly the field of view but it really is a 23mm lens.

4) Use a smaller aperture i.e. bigger f-stop thus increasing your DOF and then fewer things will be near the edge of the range of distances which are in focus. (the easiest thing to do)

5) Pick a focus distance or a focal point which is closer to the same distance as the high contrast elements like the palm fronds, the masts, or the tree leaves. (probably the smartest thing to do)

6) Avoid high contast areas in your composition

7) If you must have high contrast areas, make sure that they are illuminated by monochromatic lights. The sun unfortunately is incandescent and a black body radiator and thus includes all frequencies of light.

 

and probably a few more which I didn't think of because it is late, I'm tired, and I've had a few beers. ;-) 

 

 

I appreciate the suggestions, but:

 

1) I've only been able to get it somewhat corrected in Lightroom, not enough for me to be happy with. And my usual workflow doesn't involve Lightroom.

 

2) Those crops are from the camera-processed JPEGs.

 

3) gotcha

 

4) those are from at least f/5, if not smaller.

 

5) Good call, will try

 

6) Pretty impossible in crazy-sunny California taking outdoor shots like this. It sucks because this camera isn't as good as my E-M10 in low light (hooray IBIS), and it creates more CA (or not-CA) in outdoor light.

Edited June 22, 2015 by julian m

[julian m]

Just to repeat again for everyone: These shots are from f/5 or smaller, not wide-open. The first two are definitely f/5. The second two are smaller.

 

My own experimentation hasn't resulted in radically reduced CA (or whatever this is) at even smaller apertures than that. But I'll try with much narrower and see what I get.

Edited June 22, 2015 by julian m

[bencoyote]

 

6) Pretty impossible in crazy-sunny California taking outdoor shots like this. It sucks because this camera isn't as good as my E-M10 in low light (hooray IBIS), and it creates more CA (or not-CA) in outdoor light.

 

 

6 was meant mostly as a joke. Of course you can't be outside with only monochromatic illumination. :-) 

 

I don't use JPG but thought that someone had said that this was fixed in the JPG. So scratch 2.

 

BTW though the T may benefit from IBIS or in lens VR like the Q I find that the T does equally well if not better than my E-M1 in low light. I will admit that I focus more on technique since I got the T and that may be part of it. The T just wants to be more conservative than the E-M1. The ISOs don't match up exactly. I get the feeling like the E-M1 kind of pushes it a bit. The T tries to stick to higher shutter speeds and lower ISOs but before I sold most of my m43 prime lenses, i tried to compare them. I could see that the to get the T to take the same shot I had to use a manual exposure setting and when I dialed in the same effect there was less noise on the T. Since you have both and my m43 kit is a shell of what it once was, it would be cool to see some comparative shots.

 

When I did it, I found that the T ended up being just under 1 stop faster for the same illumination and had less noise. 

e.g. E-M1 400 1/30s f/2.8 would end up being 400 1/30 f/4 or 400 1/60 f/2.8 or something like that.

it took some fiddling to get the exposure about the the same. I naively assumed that if one shot a particular set of parameters I could dial those into the other camera and get the same exposure. The difference seemed to be greater at higher ISOs.

 

BTW just I recalled the words I used to find the explanation about CA.

Lateral vs axial CA. The lateral CA is the one that can be designed around and happens mostly at the edge of the lens.

Axial is the one that happens anywhere in the frame and is caused when high contrast areas are too far from the focal distance.

Lateral is a lens designer's problem. Axial is a photographer's technique problem. 

 

When you compare the amount of CA between the T and m43 keep in mind that the DOF for a given aperture is about 1 stop wider for m43 than APS-C. So when you compare 23mm on the T to 17mm on m43 to make the comparison  f/1.8 on m43 would be more like f/2.5 on the T. The bigger the sensor, the more of a problem this is. I don't know how to compensate for the fact that 17mm is wider than 23mm but I know that wider lenses are more affected by this problem. M43 seems to do a lot more fixing in the camera even in their RAW files and in the camera and lens profiles than Leica which seems to have a very light hand.

[bencoyote]

Here are some examples of the T in very low light https://photos.google.com/album/AF1QipOz8jP65J_k-koVqxXM0LwsMYmKEd77_n4yldV5  Update: (darn the set from the E-m1 that I was going to post which is analogous was from a very private event and I do not feel at liberty to share publicly.)

 

Qualitatively the impression I got after playing around a bunch is the E-M1 could get approximately what I can see with my eyes. This was with a 25mm 1.4 or a 15mm 1.7. 

The T with the 23mm 2.0 can do a little bit better than my eyes but I have to be in a manual mode and set the exposure manually especially the ISO.

The M with a 50mm 1.4 does markedly better than eyes.

Edited June 22, 2015 by bencoyote

[Guest VVJ]

I have not yet experienced this with the T 23mm.  I usually shoot f8 outside.

 

I did experience a bad case of color fringing though recently shooting wide open inside a church with the Zeiss ZM 35mm and that is supposed to be one of the very best lenses for the M mount...

[julian m]

Here's f/9 vs f/3.5 (not that I'd use f/3.5 for this shot normally, just trying to compare). This tower is at least 300m away, focus point was on hills miles away. f/5.6 is somewhere in between (i.e., not that much worse than f/9).

 

There is a difference, but it's pretty minor when compared to seeing purple coloring at all vs not.

 

These are camera JPEGs.

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!

Edited June 27, 2015 by julian m

[tom0511]

the pruple fringing or Ca or whatever it is is pretty strong with the 23mm and one of the things I dont like about the T system, besides many other things I do like.

LR can reduce the effect.

[bencoyote]

Back to beating a dead horse.

 

I mentioned this to a friend of mine who happens to be a National Geographic photographer (http://www.jonathankingston.com) a couple of weeks ago and he mentioned two things in passing to me about it. I've been looking for corroborating sources but I haven't found anything yet. So I'm going to just toss it out there. 

 

1) Purple fringing as we see with the 23mm lens is worse on cameras without anti-aliasing filters. It is one of the trade offs that comes along with the increased sharpness.I wish I could cite a source to back this up but I haven't been able to find one.

2) It has a lot to do with the angle that the light hits the sensor, the further from perpendicular the more likely it is to occur. Considering the relative sizes of the lenses, and the size of the rear element, I bet that the light coming out of the 23mm is arriving at the sensor at a steeper angle. So when Leica designed the lens they optimized more for its compact size rather than the angle of incidence of the light when it hits the sensor. 

[julian m]

Just wanted to continue this thread with this. f/16. camera-processed JPEG.

 

Ends up giving large parts of this photo a slight purple hue.

 

 

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!

[CheshireCat]

Seems to me a mixture of lateral and longitudinal CA.

Impressive how much this lens sucks.

[Advertisement]

Hallo,

Look here: Leica L-Mount

[julian m]

I decided to get Lightroom and see what I could do. If I start with the DNG (again, above were camera-processed JPEGs at f/16), built-in lens profile, check Remove Chromatic Aberration.

 

In the first image, you can see some of the more obvious CA going away, though to my eye there's still a hint of a blue/purple hue left from what seems a bit like purple fringing.

 

You can see the remaining purple fringing in the second image. Using the defringe tool with that seems pretty tricky. Most of the samples I try end up causing parts of the sky to go grey. I'll have to try spending some more time with it.

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!

Edited October 31, 2015 by julian m