Why no in-camera CA correction on Olympus bodies?

I've always been curious as to what the philosophy is with regard to not doing chromatic aberration correction in-camera?  Every m4/3rds Pansonic camera that I've had (4 so far) does it and every m4/3rds Olympus camera that I've had (just the OM-D, so far), doesn't.

I can understand the religious arguments for in-lens vs. in-body stabilization.  But I don't understand why you wouldn't want to get rid of CA in your JPEGs.

Perhaps the thinking is that if they do in-body CA correction, more people will buy Panasonic lenses over Oly lenses. The CA is the main downside of Panasonic lenses for Oly users.

What he said. I've read, but can't speak from experience, that Oly lenses are already corrected for CA but Pannys are not, so an Oly body in theory needs Oly lenses to avoid CA. Sounds a bit unlikely to me.

I can attest to the CA issue with Panasonic lenses on Olympus bodies.  I rented a Panasonic 12-35 for my OM-D and saw nocticeable CA on my shots.  It was easily correctable in post processing.  However, my Olympus lenses (14-42, 12-50, 10-150, 14-54 MKII) don't exhibit this effect as noticeably.  I guess it's a philosophical issue.  Olympus comes from a long history of making superb optics at a time when post processing was not much more than what was done in a darkroom.  Panasonic is an electronics firm that has no such history to draw on and sees nothing wrong with applying correction to the image.  For me, I tend to favor the Olympus approach - get it right at the source.  I am quite frankly a little put off by a $1100+ lens that doesn't effectively manage CA and relies on software to correct what I feel is a design flaw in the lens

Here are some thoughts about it:

http://forums.dpreview.com/forums/post/50641796

--
Henry Richardson
http://www.bakubo.com

My guess is that the CA correction process is Panasonic proprietary and not part of the m4/3 standard like the distortion correction.. Oly could reverse engineer it to work with existing lenses, but there's no assurance that this is future proof.

What if Panasonic invokes a new variable that now gets used in a new lens, and suddenly the Oly calculations get thrown off for that lens? Now they have to issue a firmware update.

Not sure about this. There must be some documentation, otherwise you'll need an update for every (correcting) RAW converter to support new lenses.

CA correction may simply be to hard to implement in Oly's JPEG engine. Does the Oly RAW viewer support CA correction?

Yes, but it's not automatic, there are 2 sliders to adjust but it's very effective with my Pany 14mm f2.5 which shows bad purple fringing under some conditions.

Rgds, rob

…why they don’t do it. I do NOT buy the notion that Panny lenses are inherently more subject to CA than Oly lenses. They aren't.

Is it difficult in PP? You adjust two sliders. How difficult is that? But why should we have to do it manually when Panny (and lots of others, don’t forget) clearly shows the way to do it in-camera?

Cheers, geoff

--
Geoffrey Heard
http://pngtimetraveller.blogspot.com/2011/10/return-to-karai-komana_31.html

And yet, they have no misgiving about putting out lenses with massive optical distortion that rely on in-body correction. Something as trivial as CA correction (which is a matter of course in nearly all WA lens PP-ing, regardless of maker or price) being left out is odd. It isn't as if Olympus completely removes it optically... though it is worth mentioning that even Panasonic does not completely remove its own through software (ACR will remove it completely from either, so it doesn't bother me as a non-jpeg shooter).

Nearly every Canon L lens is >$1000 and they ALL suffer from visible CA, including the 24-70/2.8 mk2 at 2x the price of the 12-35. Clearly this is not something easily addressed on the optics side. I am all for a PP approach if it keeps the size (or at least price) down and helps with the overall image quality (addressing CA optically makes other compromises). I am in the "ends justifies the means" camp when it comes to cameras... there is no merit to "purity of light transmission" so far as my prints are concerned.

--
-CW

Not saying that it never happens, just that it's far less prevalent with MZD than Panasonic lenses.

My guess is: Panasonic found that it was less expensive to design a small lens that has CA, and correct it after the fact, than design a small lens that doesn't exhibit CA. This is likely the reason that the PL25 costs quite a bit more than the 20 1.7 - Leica doesn't like to design a lens with CA, either (though both versions of the PL25 do have a bit of CA under certain circumstances).

Automatic CA correction sounds nice, but it does involve loss of detail, as adjacent areas are cloned into the CA. Better still that you don't have it in the first place.

Lateral CA correction is lossless, there is no loss of detail (if it is done on the raw data such as in camera or on raw files).

No, the explanation is not that it Panasonic propriterary. Lots of RAW converters do lateral CA correction and the simplicity with which it can be done makes me doubt that it even could be patented. Lateral CA arises because the three sub-images (red, green, blue) are not the same size. All it takes to correct is to scale the three relative to one another (enlarge one slightly and/or reduce another). It is very simple to do and not all computationally demanding.

In LR 4.x, which removes lateral CA automatically, without the help of a lens profile, the program first analyzes the image to determine the amount of CA correction required and then carryies out the actual correction. However, when the lens properties are known in advance, as they would be with an MFT lens on an MFT camera, the initial analysis is superfluous. The amount of correction required would simply be communicated by the lens to the body.

Indeed I wonder why not.

--
Simplicity is the ultimate sophistication.
God always take the simplest way.
Imagination is more important than knowledge.
God is the tangential point between zero and infinity.
Aleo Photo Site

I think that is not true. Based on tests I have read, both lenses have similar CA small issues.

--
Simplicity is the ultimate sophistication.
God always take the simplest way.
Imagination is more important than knowledge.
God is the tangential point between zero and infinity.
Aleo Photo Site

And the best Panasonic Lumix G lenses the same. I think they should include CA correction.

--
Simplicity is the ultimate sophistication.
God always take the simplest way.
Imagination is more important than knowledge.
God is the tangential point between zero and infinity.
Aleo Photo Site

CA correction is computationally very simple to do (see my reply to baxters above) so implementation difficulties is certainly not the explanation either.

The best short answer to this question is the one provided by kenw here.

His post is worth quoting more or less in extenso:

"The very expensive 12/2 has quite a bit of lateral CA which cleans up very nicely in LR but because Olympus is either too lazy, to inept, or too close minded to implement the corrections in camera it looks bad on even on their bodies when in fact it is an excellent lens.  The 9-18 improves greatly with lateral CA correction as well, as do their kit zooms.

Almost all manufacturers, even ones selling multi-thousand dollar lenses for professionals, have implemented lateral CA correction in camera and into their RAW file meta-data.

Panasonic has this one right while Olympus has their head up the location that the only profitable part of their company is best known for imaging."

I have been under the impression, for some time now, that CA correction parameters, like barrel distortion parameters, are actually stored in every [m]4/3rds lens and can be easily read by any camera (no need to obtain a lens profile elsewhere).

And it makes no sense for Olympus to not implement CA because they want to show how crappy Panasonic lenses are optically.  All the consumer is going to know is "this camera body produces worse pictures than this body."

There has to be some loss of detail.  You are either digitally zooming the lower wavelengths (i.e. red) or digitally compressing the higher wavelengths (i.e. blue).  Or both.  But you are either rendering the blue on fewer pixels than the 4/3rds sensor actually has or you are rendering the red on more pixels than it has.  There's loss either way.