Summilux 24 Hyperfocal problem at infinity

[newnew]

Advertisement (gone after registration)

I have acquired a Summilux 24 mm lens and have been testing its capabilities under different conditions. The lens is perfect to expectation in shapness, close or distant when set to the right focus distance. However , when I set it to hyperfocal , with the infinity sign coinciding to for example F5.6, the infinity at corners is murky (see below at 5.6 second picture). The center at the image borders are fine sharp. However when I set the focus to infinity without using the hyperfocal sharpness is fine (first picture @F5.6). I have made the same test with Elmarit 24 2.8 at all the apertures and this does not happen. Could anybody explain why that might happen ? Summiluxe's diaphragm blades seem to open round but this was just a visual check.  Has anybody else experienced that ? I thought the sharpness in hyperfocal would only depend from aperture and if the lens is sharp otherwise why would that happen ? Has the Floating element structure any influence ?

Thanks for your comments.

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

Edited July 25, 2020 by newnew

[Advertisement]

The Leica Q3 is available for pre-order at B&H Photo Video and Adorama

[a.noctilux]

Hello Ali,

I'd consider curvature of field, if everything is fine.

Features of so many lenses (mainly "light Giants" that I use ).

Another thought : circle of confusion in use in digital world is not 0.03mm anymore.

Out of curiosity, I had a look at the MFT published by Leica, the Summilux-M 24mm at f/5.6 shows strong curvature of field:

have a look, here 

( 30% to 15% of contrast for 40 lp/mm may be the answer )

 

...

or atmospheric haze ?

Edited July 25, 2020 by a.noctilux

[newnew]

Thanks Arnaud, but look at the same taken (picture below) by Elmarit 2.8 at hyperfocal with infinity set at 5.6 , there is a difference that is not the haze. Could the field curvature be exaggerated with a hyperfocal setting ? Not that I am a pixel peeper and the Summi lens is great for street or as an overall  use. I want to understand why the hyperfocal behaviour is different than other lenses like the Elmarit, and also if that is just my copy or is it its natural behaviour ?

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

Edited July 25, 2020 by newnew

[adan]

Two points.

1) The DoF scales on Leica M (and most other) lenses define what will "look sharp" in a standard 25cm x 20cm print (the old standby 8" x 10" glossy). Pixel-peeping at 100% grossly violates that assumed standard. It won't work. DoF is always about what will "look" sharp at a fixed image enlargement, not what will "be" sharp in an absolute sense.

To use hyperfocal focusing and actually have infinity sharp at high viewing magnifications (digital or high-res film scans on-screen, or large prints), the rule of thumb is to stop down at least two stops beyond where you set the infinity mark. I.E. set infinity to f/5.6, and then actually set aperture to f/11 or f/16. (At which point you may get a little overall loss of sharpness due to diffraction - but no-one ever said the laws of physics are kind to photographers. )

Or limit your viewing size on-screen, and printing size, to 25cm x 20cm. In which case the lens markings will be about right in predicting apparent sharpness.

2) The 24 Summilux is well-known for weak corner/edge performance in any case (I've used one!). It is a photojournalist's lens, not a landscape lens. And your results show pretty much exactly what Leica's own MTF charts predict for f/5.6. Very low MTF (~17%-25% at 40lpmm and the edges) in the tangential direction (the side-to-side blurring/streaking of the buildings and other details in your example, as represented by the ••••••• line at 18-21.0 mm out from the center of the picture (right edge of the graphs)).

By comparison, the 24 Elmarit MTF (lower chart) at f/4 is nearly 60% at the picture edge (18mm - long dimension) in the tangential direction - much higher acutance or perceived sharpness, not counting DoF/hyperfocal effects. It is substantially "sharper" to begin with.

In general, do not expect an f/1.4 lens to equal a slower (f/2.8) lens at the same smaller aperture (e.g. f/5.6) - unless there has been a major intervening improvement in lens technology (e.g. 2006 50mm Summilux-M ASPH vs. 1980 50 Summicron-M v.4/5 - 26 years makes a difference). And even than, not at the edges, necessarily. Lens designers have to pick their battles.

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

   

Edited July 25, 2020 by adan

[pgk]

Firstly, as a rule of thumb I tend to use the depth of field scales for two stops wider open than the stop I am actually using - even on the M9. 

Secondly I'd guess that the Summilux has field curvature and this combined with lower corner performance is what you are seeing. Its not a lens that I would use for landscapes in good light levels.

Fast lenses, especially wide angles, are still, to some degree, a trade off between speed and performance. I've used them (but not Leica) and have found much cheaper, slower lenses to be as good if not better at medium stops. I would use the Elmarit in preference for landscapes myself (I had one and liked it well enough, but I'm not really a 24mm shooter and much prefer 21 or 35 lenses).

[newnew]

Thanks Andy , Thanks Paul for your detailed explanations. I indeed knew while acquiring this lens that it is not the perfect landscape lens and that the field curvature existed. I bought it mostly for "in context" portraits and street . Actually in my tests , at lower apertures it renders crisper than the Elmarit. Set to infinity ,  I find its performance pretty honorable at corners, pretty useful for some "moody" landscapes".  What really puzzled me was the difference between the direct focus setting and the hyperfocal setting. I will make more tests taking your suggestions of +2 f stops for infinity in hyperfocal setting.

The important is to understand how the lens behaves and adjust accordingly to its nature. I was also trying to see if it would make the Elmarit obsolescent in my lens portfolio.  I might only use the SEM 21 for the landscapes.

I also understand that what I see is not my copy related but seems to be a general property of this lens design. It would be interesting to know if other users have observed the same effect.

[adan]

Advertisement (gone after registration)

Just for the record, I would note that fuzzy corners or poor corner MTF can be caused by things other than "curvature of field." Such as coma, or astigmatism.

Curvature of field = fuzzy corners, but fuzzy corners ≠ curvature of field, necessarily.

The strongly-diverging MTF curves for tangential ...... and sagittal _____ rays in the 24mm Summilux plots at f/5.6  - and the mono-directional blur in newnew's samples, looks to me a lot like astigmatism (lines at 90° to each other do not focus in the same plane).

To demonstrate curvature of field, one would have to shoot a test picture of a three-dimensional "field" or array of points, and show not only that the corners at, say, 3m from the lens are fuzzy when the center is sharp - but that they are also sharp at some other distance (2m or 4m). If they are just fuzzy everywhere, that is not a curved field of focus. Just soft corners.

[lct]

Seems like the 24/1.4 asph (first chart) has sharper corners than the 24/2.8 asph at f/5.6. I am no techie enough to explain this but i wonder if the 24/1.4's floating element plays a role in the OP's problem.

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

 

[adan]

Gotta include the dotted lines as well.

[lct]

7 hours ago, adan said:

Gotta include the dotted lines as well.

Would you mind to explain? Just curious.

[adan]

The solid and dashed/dotted lines show the MTF in two different orientations. The sagittal orientation (resolution lines oriented to pass through the center of the image) and in the tangential orientation (resolution lines that are perpendicular to the sagittal lines).

They are solid and dashed not because one is more important that the other, but simply to distinguish them on the graph (since they often overlap). They are equally important in graphing how the lens reproduces (or fails to reproduce) detail. Some companies even swap which are dotted and which are solid - or just use different colors or thicknesses.

This is why classic lens-resolution targets always show the resolution line targets in pairs, in both horizontal and vertical orientation. And in the best targets, scattered them around the center in various orientations. To detect directional differences in resolution or MTF, caused by aberrations with a directional bias (coma, astigmatism, even lateral (out from the center) chromatic aberration).

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

Here is a real-world example of why MTF must be considered in both orientations. Astigmatism in a C/V 35 Nokton v.1 at f/1.4. The details are much blurrier up-and-down (or, in the whole image, around the center) than left-to-right (or, in the whole image, out from the center) and get progressively more so as one gets farther from the image center (image center is out of frame to the left).

If one only used the solid (or dotted) lines on a graph to judge this lens, one would get a false impression of how it performed.

[lct]

Thank you much .

[newnew]

Thanks Andy for these nice clarifications.

I have actually tested the lens with 2 stops more for infinity in hyperfocal and the results are much better. I am just wondering if , in that case, the focus point is not out of FLE activation zone such that the situation disappears. Could we have the effect of FLE to interfere with the hyperfocal setting if the hyperfocal distance is set to less than 2.5 m ?

All this said, there is really a very small difference in terms of the sharpness range if the lens is set to full infinity without using the hyperfocal. In practical terms I am not sure that it will make a big difference in any photo. But one just needs to be aware of that feature of the lens. Another question would be if other Lux lenses with FLE exhibit that behaviour in hyperfocal setting ?