Focus shift musings (35 Lux pre-FLE, etc.)

[pklein]

Advertisement (gone after registration)

This post might be called Dr. Strangelens, or, How I Want To Stop Worrying About Focus Shift.

I've decided to focus (pun intended) on my Leicas after a several years favoring another camera. I love the rendering of my pre-FLE Summilux ASPH (bought new in early 2000s). But it has noticeable focus shift. I realized that I could compensate by focusing slightly closer than the intended subject at middle apertures. But I didn't want to deal with that in fast situations. I also picked up a used 50/1.2 Nokton v.1. It had no appreciable focus shift, but has a duller rendition than the 'Lux, weighs a ton, and blocks too much of the viewfinder. So I only use the Lux at f/1.4, 2.0 and 8. And I've ended up with a set of "available light" lenses and an other set for outdoors.

Fast forward to today. I retested the 'Lux on my current cameras (M10p and the M(9)M "Henri" Monochrom). I taped a tape measure to the floor, and set up a tripod 1.1 meter away, looking down diagonally. Nothing had really changed, of course. And then I decided to test all my lenses.

Both my 35mm and 50mm 1980s Summicrons showed front focus at full aperture, especially the 35. At f/2.8 and smaller, they are both fine for all practical purposes. Then I used my Visoflex to focus them. It turns out that they are not soft wide open as I had thought. They are actually only a little softer than at f/2.8, but they are focusing closer than the RF indicates. It turns out most of my f/2 or faster f/2 or faster lenses have some focus shift. Except for modern Voiglander Noktons. And one surprise: my 1947-vintage Summitar. Go figure.

The question is, how much does it really matter? I need a real-world benchmark. My experience tells me that if I can't see unsharpness when a file is blown up to 50%, or can just barely see it, it doesn't matter at the size that most of us display or print. Is this a reasonable standard?

Years ago, conventional wisdom was that Leica lenses were perfect and all unsharpness was user error. That might have been believable with film, which has thickness and grain. But not with digital. And we now have the ability to pixel peep beyond all reason.

Thoughts and advice welcome.

--Peter

[Advertisement]

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

[elmars]

I hate focus shift. Before I buy a lens, I check it for this. According to my findings, almost all older lenses with a wider aperture have a greater or lesser degree of focus shift. This is (one of the reasons) why I prefer modern designs.

Whether the blurring caused by focus shift is actually distracting depends, of course, on the magnification scale, but also on the subject being photographed and the message of the image. In most cases, it doesn't matter if the plane of focus is a few centimeters further forward or backward. The only really critical cases are portraits. 

I think it's a good idea to view the image at 50% magnification. If you're satisfied with it then, the fact that the sharpness isn't quite perfect won't matter (the same applies to noise).

[adan]

I have never found focus shift to be a problem.

For the simple reason that I cannot imagine any reason at all on Bog's Green Earth to use a lens "stopped down just a little."

Either I need maximum aperture for low-light journalism or maximum action-stopping shutter speed (the f/1.4 or f/2.0 I paid for).

Or I want the best overall performance across the whole image (corners plus sufficient DoF). Traditionally, at 3-4 stops down = f/4.8-f/6.8.

The one exception to that was using the non-ASPH M-mount 35 f/1.4s (Summilux or Nokton), which I used as pseudo-Summicrons at f/2 in low light because they were far too "dreamy" at f/1.4. And fortunately the copies I had of those were calibrated for f/2.0 rangefinder focusing anyway.

[costa43]

5 minutes ago, adan said:

I have never found focus shift to be a problem.

For the simple reason that I cannot imagine any reason at all on Bog's Green Earth to use a lens "stopped down just a little."

Either I need maximum aperture for low-light journalism or maximum action-stopping shutter speed (the f/1.4 or f/2.0 I paid for).

Or I want the best overall performance across the whole image (corners plus sufficient DoF). Traditionally, at 3-4 stops down = f/4.8-f/6.8.

The one exception to that was using the non-ASPH M-mount 35 f/1.4s (Summilux or Nokton), which I used as pseudo-Summicrons at f/2 in low light because they were far too "dreamy" at f/1.4. And fortunately the copies I had of those were calibrated for f/2.0 rangefinder focusing anyway.

I am the opposite and do not like focus shift, I typically use just the rangefinder so it’s a pain on lenses that are prone. I also use the entire aperture range on a lens.

Sometimes it’s nice to have some separation whilst still retaining some context to the scene. In lowlight, with today’s ISO capabilities, I also sometimes stop down a little for more depth of field and ride the ISO a little. For portraits up close, f2/8 and f/4 are useful for getting the ears and nose in focus but still having some dreamy fall off. I can think of a lot of examples where the middle apertures are useful but we all approach things differently. 

[david strachan]

I have many lenses that focus shift at various apertures.

Many i know to just focus a tad closer in the viewfinder, which brings my subjects to critical focus. 

Not too much of a problem...really just another of the foibles of rangefinder focusing.  Sometimes I prefer focusing off the sensor. Or even autofocus..  

Edited 19 hours ago by david strachan

[Kiwimac]

I gather that, at least in the context of the 35 Lux pre-FLE, if you send them for CLA (and coding if you have one that isn’t) then Leica can sprinkle secret sauce on the lens and significantly reduce the issue. 
 

I’m in the process of buying one and I asked the selling dealer (who I know to be on the level and trustworthy) about this. He told me that the lens I’m looking at was CLA’d by Leica 2 years ago and coded. He tested the lens himself and couldn’t see any shift. 
 

If we go ahead, I’ll be testing it myself and returning it if it doesn’t perform. 

[TomB_tx]

Advertisement (gone after registration)

Every fast lens I've tested carefully on digital shows some shift in focus as it is stopped down. I seldom noticed in actual pictures even with lenses that show significant shift - like the Nokton 35 1.4 ver 1. The 50 Summicrons from the 1950s-70s I've tested tend to best match the RF focus at about 1 stop closed, which I suspect was intentional to compensate for the shift, like Zeiss did with the 50 Sonnar ZM,

I was fully satisfied with all the lenses before I started testing for shift. So my advice is just enjoy the lens and don't obsess about it.

[jankap]

I don´t use rangefinders. And I hate rulers in cm or inches. An LCD screen gives me the real world. No double light lines.

Edited 12 hours ago by jankap

[lct]

I like neither rulers nor focus shift so i try to avoid the latter. A good way could be to have your lens calibrated by Leica. My 35/1.4 pre-FLE has been calibrated this way together with 6bit coding and i can't seem to find any focus shift out of it. Another way could be to use an EVF but this is another story.

[Vanillasludge]

Focus shift is a characteristic of the core lens design and can not be eliminated by rangefinder calibration.  At best we are setting a point of calibration that averages out the effect in a way that is less noticeable.  

Be aware that when people say it’s only a problem at close range they are misunderstanding the nature of the issue. 

Focus shift is expressed as a percentage of the total distance to the intended target.  This is because the film plane of focus is shifted by a distance regardless of the distance to the object. (The change in aperture determines the amount of shift)

That means that at close range it may only be a focus shift of a few centimeters from the subject, but as the distance grows, so does the length of the shift.  

As long as it stays within an amount that our eyes accept as “sharp” it will not be noticeable. Beyond that point, even at longer distances, a very “particular” person will see the plane of focus shift, if their equipment allows sufficient magnification.

On the King Of Shift, Sonnar CM for instance,  you can see that the distant background becomes quite sharp when focusing at middle distances and stopping down to say, 5.6.  

The point of intended focus however will actually lose a slight amount of sharpness. 

Focusing the Sonnar at infinity and setting the aperture to 5.6 will result in focusing beyond infinity, and introduces slight blur.  (This is for lenses calibrated to be sharp at full aperture. Those calibrated for 2.8 will front focus quite severely at wider apertures). 
 

In my experience the greatest example in the Leica stable is the King Of Bokeh, v4 Summicron 35.  I would suspect many of the older fast double gauss designs would be the same. 

Luckily, as a relatively wide angle lens, the effect is not a destructive as it is in a fast 50.  

All photography is an acceptance of a certain amount of blur. The limits of our eyes and the amount of magnification our equipment can achieve are what makes any of it acceptable. 
 

Best wishes 
 

 

[pippy]

22 hours ago, pklein said:

...The question is, how much does it really matter?...

I'm happy to say that this is a question which only you can answer.

For myself it doesn't really matter in the slightest but that is primarily due to the sort of subject-matter I like to shoot and how I decide it should be shot.

After having noticed that one of my own favourite (and most used) lenses was the subject of much focus-shift criticism hereabouts - a phenomenon of which I had previously never been aware - I undertook a long-winded (and very tiresome) torture-test in the studio with c. a dozen lenses and discovered that whilst the extent to which those lenses exhibited focus-shift varied across all lenses it was ALWAYS there to a certain extent but only truly noticeable at minimum-focus or thereabouts.

Normally wide-open and Min. Focus everything is fine. Stopping-down between 1 to 2 stops would introduce 'shift'. Beyond a 2 stop decrease focus shift ceased to be a problem because the increase of depth-of-field would 'disguise' residual warts. Focussed much beyond 1.5m at all apertures everything is also fine.

Pixel-Peeping is something I do when ridding my snaps of the ubiquitous sensor-dust blight but it has no place (in my opinion) when viewing of Photographs in the real world.

Philip.

Edited 1 hour ago by pippy

[rcusick]

The lens can be adjusted to determine where at which aperture the focus shift happens.  Most people choose to optimize for wide open and f5.6 and greater with the most focus shift at f4.