Resolution of M43 lenses

Started Mar 31, 2013 | Discussions thread
Anders W
Forum ProPosts: 19,064Gear list
Like?
Re: Clarification & Correction
In reply to texinwien, Apr 2, 2013

texinwien wrote:

Anders W wrote:

texinwien wrote:

Anders W wrote:

texinwien wrote:

Anders W wrote:

texinwien wrote:

axlotl wrote:

The reviewers at Photozone appear not to have fully accepted that post capture lens corrections are now a normal part of the image making process.

Normal according to whom? Not all lenses require as much software correction as others. Some require very little. Some require much more.

So they RAW convert with RAW Therapee, demonstrate 5.8% barrel distortion at 12mm and decry that as "excessive".

It is pretty excessive for a $1300 lens.

But if, like the rest of us, you simply use the camera as designed and allow the distortion to be corrected automatically in camera then all you will see is a mild amount of residual distortion. In addition the Photozone reviewers say that the automatic distortion correction is "lossy" , however those of us who have actually used the lens for several thousand photos, often critically inspected for sharpness, are unaware of this supposed lossiness.

Whether you have noticed it or not, correction is, in fact, a lossy operation. There is nothing 'supposed' about that statement, and the Photozone reviewers are 100% correct.

They also criticise corner shading which is present but which I hardly ever notice in actual photos and if I do, it is easily removed in a RAW converter.

Relying again on software correction for a problem that should probably be optically corrected in a premium lens.

In other words they are complaining about lens attributes which are of no concern in the real world.

Not true. The attributes are of concern, which is why you must apply software correction in order to fix them. The point is, at what price should we expect a lens to be optically excellent? It costs less to build a lens with poor optics, then correct resulting problems in software.

As long as the manufacturers pass on some of the savings they realize from these cost-cutting measures, I think most are OK with the idea. If, however, a manufacturer builds a lens with poor optics and tries to sell it as a premium quality lens, there are plenty of discriminating buyers who won't appreciate the trick.

On what grounds do you leap to the conclusion that designing lenses to be software-corrected in one specific regard (geometric distortion) is done only or primarily for the purpose of cutting costs?

I never said what you're asking me to defend. Software correction is practically free, however.

If you are not saying that it is done for the purpose of cutting costs, how then should we interpret what you say in the two paragraphs immediately above the question I raised?

There's not much to interpret, is there? It's a simple declaration. I'm not sure which words lead you to your interpretation. The statement, "it costs less to build a lens with poor optics, then correct resulting problems in software" doesn't require much interpretation, I don't believe, and your 'restatement' relies on something other than the words I wrote, it would seem.

I thought that what you said was meant to apply to the Panasonic 12-35/2.8, which, according to you, has "excessive [distortion] for a $1300 lens" and relies "on software correction for a problem that should probably be corrected optically in a premium lens". If what you said was not meant to apply to that lens, what would be its relevance in the present context?

It was meant to apply to that lens, but the restatement of my words in your original reply to me still does not follow.

Allow me to expound: I never, at any time or in any place claimed or implied that manufacturers design lenses to be software-corrected "only or primarily for the purpose of cutting costs". Those are your words, and not mine.

I said, "it costs less to build a lens with poor optics, then correct resulting problems in software." Your restatement does not follow from my words.

OK. I don't think we have to rehash this any further. We have both made our points and I suggest we leave it at that.

I could probably add an 'all things being equal' to the front of the statement to make it a little less arguable. I might also replace 'problems' with 'imperfections' then add 'than it does to build a lens with excellent optics in the first place' to the end.

And what makes you think that the end results (after correction) must necessarily be worse than they would have been for a lens corrected by optical means only?

I have also not made that claim.

If you are not saying that the results are worse, how then should we interpret what you say in the two paragraphs immediately above the question I raised?

This is how you should 'interpret' them: all things being equal, it costs less to build a lens with poor optics, then correct resulting imperfections in software, than it does to build a lens with excellent optics in the first place.

I find no reason to think along such lines. If the end results, after correction, are good, as they are for those MFT lenses with strong barrel distortion prior to correction that I own and use (12/2, 7-14/4, 14-45/3.5-5.6), what's wrong with it?

Nothing is wrong with it if the end results are as good. I never claimed otherwise. As a matter of fact, I specifically said that most are probably fine with the idea if the cost-savings are shared with the consumer.

So you are suggesting that it is a matter of cost-savings after all?

I'm suggesting that, all things being equal, it costs less to build a lens with poor optics, then correct resulting problems in software, than it does to build a lens with excellent optics in the first place.

I think of this lens-design strategy as an asset rather than a liability for MFT and I am sure the main reason that some others refrain from taking advantage of it is that it runs into trouble with an optical viewfinder.

I don't have a problem with the strategy.

When you compose, you want to know what you will get once the image is properly corrected. Possible with an EVF, not possible with an OVF.

Interesting theory- sounds plausible.

If you have any links to good reading on the subject, I'd be happy to have a look. Might save you some effort trying to retype what you've already written elsewhere.

There are lots of threads on the subject where this has been debated at quite some length. Here is one of them. See particularly the contributions by kenw.

http://www.dpreview.com/forums/post/39680143

Will brush up, thanks for the link.

tex

Thanks for supplying the informative link. It's worth a read for anyone who's interested in the subject.

Correction: After reading some of the posts at your link, as well as information elsewhere on the Internet, I've learned that optically-correcting barrel or pincushion distortion often (always?) causes a lens to suffer from mustache distortion. So it's not a question of whether we wish to correct all distortion optically, or not. It seems we'll have one sort of distortion or the other, if we rely on optical correction, only.

Manufacturers making lenses for OVF cameras tend to correct the generally more visible pincushion and barrel distortion in the optics, so the photographer can see in his OVF how the resulting photo will look. This, in turn, implies that these lenses are more likely to suffer from mustache distortion.

Pincushion and barrel distortion are easily corrected with software. Mustache distortion is much more difficult to correct with software. For this reason, manufacturers making lenses for EVF cameras tend to leave barrel and pincushion distortion uncorrected in the optics, relying instead on software correction, both in the camera and out. This implies that these lenses are less likely to suffer from mustache distortion, which, as noted above, is difficult (sometimes approaching impossible?) to correct via software.

So, I was wrong to imply that a premium m43 lens should have barrel or pincushion distortion corrected optically. Rather, by leaving such distortion uncorrected in the optics and relying, instead on software, manufacturers are able to make lenses that deliver images that, once processed, suffer from neither pincushion, barrel nor mustache distortion, something that may be either very difficult or even impossible to achieve with lenses manufactured for OVF cameras.

Interesting stuff.

Certainly. And you are perfectly right, as far as I know, about what you say above. What might be added is that the choice isn't, and has never been, between full optical correction (with only a little mustache left behind) on the one hand and strong software-corrected distortion on the other. WAs not designed to be software-corrected usually show a fair amount of residual barrel distortion aside from the mustache. Typically, the amount left is greater than what a corresponding MFT WA lens delivers after software correction but smaller (say about half) of what it delivers before correction.

A case in point here is the very premium Nikkor 14-24/2.8, with a barrel distortion of about four percent at its wide end, i.e., not a whole lot more than the Panasonic 7-14 at 5.2 percent. While the difference may be smaller in this particular case than it usually is, you get the picture.

Of course, before the digital era, optical correction was the only realistic option. Even in the digital age, systems relying on OVFs still face the problem that software-correction of distortion implies that what you see is no longer what you get. MFT doesn't face this problem and is thus the first system that is free to take full advantage of software correction. While I am sure that such correction is no panacea, adding alternatives to the available design options is rarely a bad idea.

Clarification: I also didn't wish to start a firestorm here. To be clear, I think the final result is the important one, no matter how it is reached. That is, the resolution and distortion tests should look at the processed images. That's all that counts.

I do think that manufacturers should share the savings with consumers where such savings are realized by using software rather than hardware correction. In other words, as a consumer, I would prefer that the manufacturers from whom I buy base their prices at least partly on their costs of manufacturing. If a manufacturer finds a way to build an equivalent widget for 1/10th of his previous costs, then offers the widget to me at twice the previous price, I'm likely to feel cheated and avoid the manufacturer's products in the future, when possible.

Lastly, I may be personally biased against the Panasonic 12-35mm and 35-100mm. I feel like these two lenses may be overpriced, but I'm not an expert on the matter of lens quality, construction costs and/or pricing. It could be due to a slight bias toward Olympus on my part coupled with a bias toward primes over zooms, at the moment.

tex

 Anders W's gear list:Anders W's gear list
Panasonic Lumix DMC-G1 Olympus OM-D E-M5 Olympus E-M1 Panasonic Lumix G Vario 14-45mm F3.5-5.6 ASPH OIS Panasonic Lumix G Vario 7-14mm F4 ASPH +21 more
Reply   Reply with quote   Complain
Post (hide subjects)Posted by
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark post MMy threads
Color scheme? Blue / Yellow