MFlenses - デジタル時代の新しさ
MFLenses 2026

マニュアルフォーカスレンズフォーラム

修理ノート、実際のテスト写真、画像豊富な詳細アーカイブスレッドから、ヴィンテージレンズの世界を探検してください。

修理ノート、実際のテスト写真、画像豊富な詳細アーカイブスレッドから、ヴィンテージレンズの世界を探検してください。
スポットライト

フォーラムトップページのおすすめパートナー掲載。

AF bad, MF worse! (in lens testing)
前のトピックへ :: 次のトピックへ  


Reply

Focus Fallibility: Lens Test Fallacies

Interesting article on the difficulty of getting optimum focus in lens testing.

引用:
We've come to say around the lab that the actual performance of modern AF systems is the "dirty little secret" of the camera industry. (Typical AF performance is probably good enough for most users most of the time with most lenses, but AF systems achieve truly optimal focus relatively rarely.)

and
引用:
As this graph shows, manual focusing through the viewfinder was indeed a complete non-starter of an idea. (Note, by the way, the expanded scale on this graph, the vertical scale is twice that of the previous chart: Variations on this chart are thus twice as large per unit of height as those in the AF example above.)


Reply

Interesting article

but with enough bullshit to confuse reader
the only solution they find to the problem was bracketing
I would rate this article 90% bullshit
they rely everything on DxO Analyzer, only trying to get better result from it, ignoring why, how & what
a lens is not better focused when dxo numbers are higher
when someone only see the tree and not the forest, it is called bullshit


Reply

Uhm. isnt that how a person would work? If the photographer was the "submarine" Wink.


Reply

poilu 氏の書き込み:
Interesting article

but with enough bullshit to confuse reader
the only solution they find to the problem was bracketing
I would rate this article 90% bullshit
they rely everything on DxO Analyzer, only trying to get better result from it, ignoring why, how & what
a lens is not better focused when dxo numbers are higher
when someone only see the tree and not the forest, it is called bullshit


Interesting response, but I don't see any technical argument from you - just statement of opinions with nothing to back them up.

Since the article gives their reasoning and their results while you provide no reasoning and no results, and no suggestion for a better method, then I would say it is rather your response which is crap, actually.

But perhaps you can explain your statement that "a lens isnot better focussed when DxO numbers are higher".


Reply

zewrak 氏の書き込み:
Uhm. isnt that how a person would work? If the photographer was the "submarine" Wink.


Yes (especially in macro photography, when using focussing rails), although a person would not spend that much time getting a subject in focus in nrmal use. But this is not normal use - its ensuring that the test charts they are photographing are in the sharpest possible focus, so that they are indeed testing the lens and not their focussing on that shot on that day with that person.

Actually its the first time I saw the beginnings of a scientific study of the sources of variation in this type of testing.

I also found it interesting that for the zeiss lens they used, they could get good focus across the field or they could get crispest centre focus at the expense of soft corners. Both shots would have fallen within the range of 'in focus' but the results would be written up quite differently. (I assume this is because of field curvature?)


Reply

Chris 氏の書き込み:
then I would say it is rather your response which is crap


In the slrgear forum, someone tried to explain some elementary but I doubt the writer can understand it
touristguy87 氏の書き込み:

Very few if any lenses are flat across the field of view, corner to corner...
You either BS people or you tell them the truth.
Otherwise you get bogged down in BS and lose your way...

slrgear forum

this 'test' actually test curvature of the lens.
by defocussing they change dof to bring border in focus
the article try to evaluate AF and manual focus based on this defocussing technique -> bullshit
A manual focus photographer see the focus on the viewfinder and adjust based on that, not by bracketing focus

I know of course the solution to obtain 'focus' without bracketing in lens testing
A simple convolve filter on liveview can be used to visualize exactly the dof
But I would lose my time trying to explaining it