Nex Lens Quality/IQ Varience

Started May 28, 2012 | Discussions thread
Ron Parr Forum Pro • Posts: 13,785
Re: in-system "quality" not objective and measurable

RussellInCincinnati wrote:

But it appears to elude you that people could also be quite intelligent, non-murky, results-oriented types that give more weight to real world tests and subjective comparisons from trusted reviewers...because they don't want to fool around with memorizing, abstracting and predicting relationships between lab tests, and whether or not a certain lens is capable of looking pretty crisp all over or not.

It didn't elude me; I was being intentionally dismissive of that approach.

You also ignore that the entire project of "objectively" predicting a lens' field performance is doomed to inaccuracy no matter what kind of objective sub-tests are done. Because of such complex factors as flare resistance for a certain person's typical subjects, ease of manual focusing that affects how sharp the lens is ever gonna be in the hands of a real user, weight of the lens in combo with some person's kit and handling ability, could go on.

All of the things you mention above relate to objectively measurable quantities. If your point is that tests could be improved to include more of these quantities, then I would agree.

With this kind of terminology, in turn consider it naive to say that the quality that will be delivered by a lens to a given stranger's entire photographic system, is an objectively and accurately predictable quantity. That's about as true as saying that "how well someone is going to do on television quiz shows", is an objective and measurable thing because of perfectly repeatable DNA testing.

The physical properties of the lens will persist no matter who is holding it and it will perform deterministically. There's no prediction to make. It's not a random event. It's a complicated, deterministic one that involves many factors.

With a game show it is extremely difficult to characterize the distribution over questions that will be asked as well as the ability of contestant to respond correctly to questions and crucially the performance on the game show is a random event. The best you could hope to do for the game show is to predict the average performance, not the performance on a particular trial. (Of course, this assumes that the questions are drawn from some distribution and that the game is not rigged in some way.)

Yes, lab tests are measuring something accurately, but the person-kit-scene-moment-in-time photographic system has way too many variables in it. Too many variables for you to be so dismissive of people feeling that a few lab-testable qualities should be given less weight, than test photos and discussions of real scenes at wide and modest apertures.

This is the crux of the matter. You make a claim that there are too many variables. However, you can't have it both ways. If there are too many variables, then no reasonable number of "real world" shots would be sufficient for you to ascertain the quality of the lens. The fact that the images are snapped outside wouldn't change the underlying problem, and the lack of repeatability would would only aggravate it.

Since neither Teseg nor anyone in the world has gone or could go to user's houses and done MTF tests on all delivered lenses, can't imagine why you imply that such data "should" be available.

I would have been surprised to get an affirmative answer on that, but there have been some efforts to quantify lens variance.

Am going to buy the Alpha 35/1.8 some day, despite every lab report that describes the corner resolution as reprehensible wide open. Because of my subjective reaction to the Photozone test photo of a train engine. That hits me upside the head with what great images the $190 dollar plastic thing can make even at F/1.8.

Whether your personal criteria assign value to all of the things measured in objective tests (corner softness at f/1.8 in this case) is different from the question of whether objective testing is valuable in general.

Photozone has objectively, accurately measured that the Sony 16mm lens is optically "poor" on the Nex 7. How much clearer could the quite poor resolution charts be: " The nice build quality and the very attractive form factor of the pancake lens can't overshadow the conclusion that it should just be AVOIDED on the NEX 7. ".

Now what exactly, objectively, accurately does that lab test rating predict about the lens' usefulness, as is also divinable from this sample photo taken with a Nex C3?

This is an interesting case because the NEX-7 appears to have microlenses that are less forgiving than those in C3/5N sensor, which can influence corner performance. IMO the decision to adopt the NEX-7 as a single reference platform for e-mount lenses was a questionable one and he should probably include tests with two bodies.

The underlying issue is that there is a property of the lens that interacts with a property of the sensor and that current tests do not measure these independently. The appropriate reaction to this is to desire more and better tests since the alternative, relying upon subjective impressions, does not fix the problem.

In reality I suspect that most scientifically minded people do what I do: We first look at the objective tests, then think about what the tests don't show, then look for images that might give insight into the areas where tests have shortcomings. That last step is unfortunate. it is time consuming and error-prone, but sometimes necessary because of the limitations of current testing methodology.

Note that this is actually quite different from the subjective approach. Making a best effort to assess an objectively measurable quantity given the available data is not a subjective approach. Offering subjectivity as the alternative to the belief that the current suite of objective tests is complete is a false choice.

-- hide signature --
Post (hide subjects) Posted by
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow