Crazy amounts of CA with Panasonic Lumix G 20mm f 1.7 II

L0n3Gr3yW0lf

Dad Pun Joke Master Over Nine Thousand Meme Lord
Joined
Jul 31, 2013
Messages
1,108
Location
UK
Real Name
Ovidiu
I know that Panasonic's 20mm f 1.7 is quite an ancient optical design (wonderful design still) with very heavy reliance on software CA correction AND that Olympus sensor design is very different from Panasonic's when it comes to optical correction both physical (as lens design) and software ... but sometimes it still boggles my Brian (aka brain) on the crazy amounts of CA it can still produce:
1617372532113.png
Subscribe to see EXIF info for this image (if available)

While some lenses measure from a few pixels of CA up to 10 or even 15 pixels wide ... this looks like over 25 pixels. It was so intense that I had to use +20 CA magenta removal AND an overlay of +100 Defringe with an Adjsutment Brush to remove it. And it did not come at little to no cost (as some people say about removing CA), there's quite a large gap of missing colour information now though it's only noticable on the tree trunk.

There has been plenty of well documented reports on Panasonic lenses producing very dificult or costly (sometimes even impossible) amounts of CA on Olympus cameras (lenses like Panasonic Lumix G Vario 7-14mm f 4 ASPH, Panasonic Lumix G 20mm f 1.7 I & II ASPH, Panasonic Leica DG Summilux 25mm f 1.4 ASPH). Personally I had some frustrations with even their highest of high end lenses: Panasonic Leica DG-Vario Elmarit 50-200mm f 2.8-4 ASPH Power OIS ... giving me very deep blue and yellow CA if the subject was not perfectly in focus and pin sharp as a needle.

I do hope Panasonic will take more consideration their lens design with Olympus camera compatibility (maybe stronger CA software profile correction or better optical design?) as OM-D Corp. has fewer shackless of the "old managment".
By no means does it make it not worth using Panasonic lenses on Olympus cameras, it's still a very per case situation of how much CA it can produce, for a lens that costs me 140 £ I won't be crying a river for having to deal with "this" ... on a lens that's 500 £ or even 1.500 £ might make me feel a bit differently though.

(PS: I can bring examples from Panny 50-200mm as well if peeps want.)
On a lighter note, at full view of the picture it's still noticeable but less "painful":
1617373409977.png
Subscribe to see EXIF info for this image (if available)
 
Joined
Jun 19, 2016
Messages
401
Location
New Jersey
Real Name
Mike Peters
That seems pretty extreme, and the situation is ideal for showing CA being heavily back lit against fine detail in the corner of the image. I'm not sure that there are many lenses that would fare much better.

I use the 20 quite a bit and CA or fringing has never been an issue with this lens at any aperture, for me. I've looked but cannot find any examples among my images that show CA to be a problem. That said, I shoot it pretty much exclusively wide open where CA is less of an issue than it is when stopped down.
 

L0n3Gr3yW0lf

Dad Pun Joke Master Over Nine Thousand Meme Lord
Joined
Jul 31, 2013
Messages
1,108
Location
UK
Real Name
Ovidiu
Never mind the CA, look at the contrast veiling with that lens wide-open. Do you have a filter on that lens? It's not an uncoated UV Tiffen, is it?
No filter on the lens, only a 46 to 52mm step up ring to fit a standard 52mm Panasonic lens cap (that's how the lens came with from CEX).
 

agentlossing

Mu-43 Hall of Famer
Joined
Jun 26, 2013
Messages
4,847
Location
Oregon USA
Real Name
Andrew Lossing
Looking at the before and after indicates really how much software in involved these days in getting us good IQ, thought I don't place the blame on the optics, I think a lot of it rests on the sensors. Lens design is always a series of compromises, maybe, but they have to make an awful lot (and then compensate for them after the fact) with MFT sensors.
 

L0n3Gr3yW0lf

Dad Pun Joke Master Over Nine Thousand Meme Lord
Joined
Jul 31, 2013
Messages
1,108
Location
UK
Real Name
Ovidiu
Looking at the before and after indicates really how much software in involved these days in getting us good IQ, thought I don't place the blame on the optics, I think a lot of it rests on the sensors. Lens design is always a series of compromises, maybe, but they have to make an awful lot (and then compensate for them after the fact) with MFT sensors.
Some optical corrections are happening alongside the sensor quality: vignetting is fixed in software by pushing the shadow exposure up around the edges where the optical profile has trouble with even exposure (but this requires a decent quality image from the sensor to make up for the extra noise in that area), as well as diffraction in lenses (which in-laws of physics it happens in the same in all the lenses when the conditions apply: very small apertures BUT) where cheaper lenses it can show up sooner because of higher compromise in build quality and design for lower cost (physical aperture design, less corrected light path that can scatter light easier at very small apertures for example) and on higher-end lenses it can be more resilient (but they will still be affected eventually) from fewer compromises made ... in this situation a sensor designed with low or no optical pass filter can help combat this issue (to a certain extent).

BUT for CA and LoCA it's almost exclusively optical lens design and software correction either on the processing unit of the camera (TruePic) or in editing software (Lightroom, Photoshop, etc). Over the years (from reading amazing articles from Imaging Resource and Rober Cicala as some of the best resources) I have learned more about the compromises it has to be done to design, build and sell lenses (the good as well as the bad). Lenses like Panasonic Lumix G 20mm f 1.7 has some serious limitations to keep the size this tiny, the cost is cheap through a simpler:
1617430346432.png
Subscribe to see EXIF info for this image (if available)

7 elements in 5 groups design with 2 ASPH elements for smoother rendition of the background but which introduces more CA in the lens. Olympus uses more ED lenses for less CA but these elements makes the background blur less smooth and more nervous. (A visible difference in this design can be seen between Olympus ED 40-150mm f 2.8 Pro and Panasonic X 35-100mm f 2.8 ASPH).
Panasonic decided to combat CA through a thicker sensor assembly design with a filter on the sensor for CA correction when the light is bent on the edge of the lens ... Olympus cameras lack this assembly so the issue is not corrected an shows more intense.

BY NO MEANS DOES THIS MAKE PANASONIC LENSES LESS WORTH USING ON OLYMPUS CAMERAS ... it's worth learning of such situations so users can work around this to make the best of the gear they bought with their hard earn money. It's a very per situation case as well as personal opinion of how much/bothersome it can be.
 

ivanbae07

Mu-43 Regular
Joined
Oct 9, 2020
Messages
42
Location
BKL, ID
Real Name
Ivan B
Yup, you're right about the panasonic 20mm f1.7 got a lot of CA (uncorrected pic). I always encounter them (CA) when i took a picture on a woodland, or when the sky is the background, and got some strong light or a really contrasty scene. When i'm on the peak of my laziness, i've always turn my back to that kind of scenes, so, i don't have to do some extensive CA cleaning later, haha.
But still, that lens is a really good one out there. For the price to performance, and the size, damn. I can't nominated other lens than the panasonic 20mm f1.7, so far.
ps: i have the mk1 version, nearly always mounted on my only milc (for time being), pen f
 

archaeopteryx

Gambian sidling bush
Joined
Feb 25, 2017
Messages
1,654
Hi Ovidiu, thought I'd chip in another round of observations.

I know that Panasonic's 20mm f 1.7 is quite an ancient optical design (wonderful design still) with very heavy reliance on software CA correction
As has been mentioned, that's veiling glare and not chromatic aberration.

The 20 1.7 measurements I'm aware of report 4.3 um of software chromatic aberration correction and a 2.2 um residual. For a 20 MP m43 sensor that's 1.7 pixels uncorrected.

Olympus sensor design is very different from Panasonic's when it comes to optical correction both physical (as lens design) and software
Olympus and Panasonic mostly buy the same 20 MP sensors from Sony (IMX269, IMX272). While I'm not aware of clear evidence confirming recent E-M1 series PDAF is a customization of the IMX272 it seems pretty likely. Thickness of the optical stack above the sensor is standardized in the m43 specification as that's required for basic lens interoperability. Within lens corrections of aberration are properties of the lens design, not the sensor. Software correction is applied to pixel values after ADC, again not the sensor.

I may be wrong about this but didn't Olympus address the purple blob issue by changing the hot mirrors in their more recent bodies to a sharper UV cutoff? (Regardless of whether they did or didn't, the hot mirror is a filter in the optical stack above the sensor and not part of the sensor itself. And the purple blobs are a form of lens flare, not chromatic aberration.)

less corrected light path that can scatter light easier at very small apertures
Please review the mathematics as this isn't how Fraunhofer diffraction works. Note that for any finite aperture the size of the Airy disc is greater than zero.

I think what you're developing is more an argument lower cost lenses can be more prone to veiling glare. I see it from time to time in images from the Panasonic 12-32 and 35-100 f/4-5.6 as well but it's unclear how much of this is attributable to baffling versus coating versus just the layout of the elements. I'm less familiar with the Olympus 12-50 and 40-150 R but haven't noticed the phenomena with them. For low cost Olympus primes I don't know. So, while the hypothesis seems plausible, I'm not aware of an adequate body of data to support or reject it.

for CA and LoCA it's almost exclusively optical lens design and software correction either on the processing unit of the camera (TruePic) or in editing software (Lightroom, Photoshop, etc)
Well, not quite. Definitely both optics and software for lateral chromatic aberration. I'm not aware of any software which intentionally corrects longitudinal chromatic aberration, though focus stacking can mitigate it.

Olympus uses more ED lenses for less CA but these elements makes the background blur less smooth and more nervous.
I'm not aware of any evidence use of low dispersion glasses has a consistent effect on bokeh.

My current tabulation of elements in the lenses for which the two manufacturers provide information gives an average of 13.6 elements in Olympus lenses (n = 29) and 12.7 for Panasonic (n = 33). I suspect this mostly just reflects Panasonic's overall tendency towards somewhat slower and lower cost lenses, though it's also Olympus not giving element counts for their body caps. Neither manufacturer provides element types for all of their lenses but first order corrected estimates are the average Olympus lens has 4 special elements and Panasonic 0.7. Since there's no apparent difference in things like MTF and CA between the two manufacturers some of the difference is likely due to Olympus marketing a number of lens element types—DSA, EDA, HR, HD, E-HR, SHR, UHR, and IIRC also UED—not found in Panasonic nomenclature as well as differences in design targets for aperture, size, weight, and price. While it's also possible Panasonic gets normal lens elements to do things Olympus doesn't there's no evidence to support this.

I don't know of any measurements of bokeh so it doesn't appear practical to quantify differences between manufacturers or within specific classes of lenses. This makes it quite difficult, I think, to argue Olympus and Panasonic have distinct lens design strategies which produce differences in bokeh. Anecdotally, Fuji X lenses do seem prone to nervous bokeh but, from what I can tell, this is more Fuji being an outlier compared to Olympus, Panasonic, Sony, Nikon, and Canon. Anecdotally, there also seems to be a tendency for zooms to have more nervous bokeh than primes. But this varies substantially on a lens to lens basis, as well as with zoom position, and I know of some modern zooms which compare favorably with some modern primes for smoothness of bokeh. Also anecdotally, there appears to be a positive correlation between OIS activity and nervous bokeh. So Olympus versus Panasonic claims around bokeh probably need to be controlled for Panasonic having more OIS lenses as well as body to body differences in IBIS since, at least on the G9, Panasonic has stated dual IS gives IBIS priority before engaging OIS.

Panasonic decided to combat CA through a thicker sensor assembly design with a filter on the sensor for CA correction when the light is bent on the edge of the lens
Citation needed. I suspect this might be an attempt to say something about the hot mirror but, if so, it's incorrect. And changing the height of the stack would violate the m43 specification.
 
Last edited:

Latest posts

Links on this page may be to our affiliates. Sales through affiliate links may benefit this site.
Mu-43 is a fan site and not associated with Olympus, Panasonic, or other manufacturers mentioned on this site.
Forum post reactions by Twemoji: https://github.com/twitter/twemoji
Forum GIFs powered by GIPHY: https://giphy.com/
Copyright © Amin Forums, LLC
Top Bottom