How does focus shift manifest itself?

[froyd]

Could somebody explain the issue of focus shift to me? So far as I understand, lenses affected by this issue will exibit a shift in the focus point when the diaphragm is adjusted from an f-stop to another. Is this correct? If so, does it follow that focus shift won't ruin a photograph as long as the photographer remembrs to set the aperture before focusing? If the photographer fails to follow this protocol and changes the aperture after focusing, with the rangefinder patch move in an OOF position? Also, are we talking about a shift of fractions of inches or something more severe?

 

[Chriscrawfordphoto]

You're right about what it is, but setting the aperture first will not fix it. On a rangefinder camera, the focus of the lens is not connected in any way to the aperture. The lens doesn't know what aperture you set. The lens's focus accuracy can only be set to be accurate at a single aperture if the lens has focus shift, and that has to be done by the manufacturer when they design and build the lens (a repair tech can reset a lens to be accurate at another aperture too).

I have the 50mm C-Sonnar. They were sold either set for f1.5 or for f2.8; the ones set for f2.8 accuracy are also accurate at smaller apertures but the front-focus when set to f2 or f1.5 while the ones set for f1.5 are only accurate at f1.5. The only way to compensate is only use it at apertures where focus is accurate, or guess how much to set the focus 'off' what the rangefinder in the camera says is the focus point.

 

[mtargz]

I wonder if it's technically possible to carefully test a lens for focus shift, and then create a focusing-mount where the rangefinder coupling moves slightly according to the aperture selected, thus allowing perfect focusing.

Financial viability is another story, though. I guess at the prices it'll demand, it'd be easier to redesign using floating elements.

 

[Roger Hicks]

I wonder if it's technically possible to carefully test a lens for focus shift, and then create a focusing-mount where the rangefinder coupling moves slightly according to the aperture selected, thus allowing perfect focusing.

Financial viability is another story, though. I guess at the prices it'll demand, it'd be easier to redesign using floating elements.

Yes, it's comparatively easy, but VERY time-consuming, as the RF cam must be profiled to give 'a bit more here, a bit less there'.

Probably a lot cheaper if done at the time of production, CNC milled, but it would still bang up the price of a lens detectably. And an awful lot of photographers are at one extreme 'careful with their money' and at the other, 'pathologically mean'.

Cheers,

R.

 

[Roger Hicks]

You're right about what it is, but setting the aperture first will not fix it. On a rangefinder camera, the focus of the lens is not connected in any way to the aperture. The lens doesn't know what aperture you set. The lens's focus accuracy can only be set to be accurate at a single aperture if the lens has focus shift, and that has to be done by the manufacturer when they design and build the lens (a repair tech can reset a lens to be accurate at another aperture too).

I have the 50mm C-Sonnar. They were sold either set for f1.5 or for f2.8; the ones set for f2.8 accuracy are also accurate at smaller apertures but the front-focus when set to f2 or f1.5 while the ones set for f1.5 are only accurate at f1.5. The only way to compensate is only use it at apertures where focus is accurate, or guess how much to set the focus 'off' what the rangefinder in the camera says is the focus point.

Dear Chris,

You don't really need to guess; it's quite easy to measure.

Cheers,

R.

 

[ampguy]

If you have a focus shifting lens

If you have a focus shifting lens

and many do, you'll likely want to focus it at your taking aperture with either an SLR, or live view camera.

Or, focus with the live view camera, at preferred aperture, then keep the lens and subject in place, then replace the live view camera with the legacy back.

Could somebody explain the issue of focus shift to me? So far as I understand, lenses affected by this issue will exibit a shift in the focus point when the diaphragm is adjusted from an f-stop to another. Is this correct? If so, does it follow that focus shift won't ruin a photograph as long as the photographer remembrs to set the aperture before focusing? If the photographer fails to follow this protocol and changes the aperture after focusing, with the rangefinder patch move in an OOF position? Also, are we talking about a shift of fractions of inches or something more severe?

 

[john_s]

I wonder if it's technically possible to carefully test a lens for focus shift, and then create a focusing-mount where the rangefinder coupling moves slightly according to the aperture selected, thus allowing perfect focusing.........

Yes, it would be similar to a floating element (such as exists in the Summilux 50mm ASPH) except that it would move the whole optical system instead of a floating element. It would make a lens more expensive. Theoretically it might have to adjust for aperture and distance, since AFAIK the focus shift is a function of distance as well as aperture.

 

[Rob-F]

Yes, it's comparatively easy, but VERY time-consuming, as the RF cam must be profiled to give 'a bit more here, a bit less there'.

Hmm, but how would the cam know what aperture the lens is set to? Seems to me the focus is still going to shift as we stop down.

 

[Chriscrawfordphoto]

Dear Chris,

You don't really need to guess; it's quite easy to measure.

Cheers,

R.

When shooting, you don't carry a ruler, so you have to guess how much to change the focus if you want to get it right, 100% of the time, perfect, no exceptions ever. I can't tolerate guesswork, and if a photo is even half a centimeter out of focus, its ruined for me and cannot be used. For the work I do, precision is a non-negotiable requirement, which is why I don't use mine at larger apertures than f2.8

A lot of people on RFF post images from their Leicas with fast lenses (not just this one...with lenses that don't focus shift a lot of people are just sloppy workers or have cameras and lenses needing repair) that are slightly out of focus. Why bother; if you 're not going to do it right, don't waste your time. For such people, the focus shift 'doesn't exist' or can be compensated by leaning in or out a bit'. The stone cold fact is, that for the careful worker (and if you are going to spend $1000+ for a lens, you damned well ought to be quality conscious enough to do it right) this lens is an f2.8 lens. The two largest openings are not for use, they're to allow the lens to focus shift down to where it is accurate at the real max aperture of f2.8 (if they made the lens as an actual f2.8 lens, it would likely not be 'usable' until f5.6!)

 

[Freakscene]

if they made the lens as an actual f2.8 lens, it would likely not be 'usable' until f5.6!

Focus shift is a property of all optical systems. All lenses display focus shift; the question for practical purposes is whether the point at which the focus system indicates is within or outside the acceptable depth of field. Problems - out of focus photos - occur when the point at which you focus using the rangefinder in an RF camera or the ground glass in an SLR is outside the depth of field. The image projected rearwards has some room to shift if you use a recording medium with depth like film, but not on a digital sensor.

If the CZ Sonnar 50 was an f2.8 lens the design would probably be sufficiently conservative, and would control spherical aberrations sufficiently well, that there would probably be negligible focus shift. The residual aberrations and the focus shift they cause, magnify as the lens design is 'stretched' - i.e. the further from the native relative focal length for the design you go and the faster you make it.

Mechanically linking the aperture to the RF mechanism is possible but would be impossibly complicated. The best current strategy for managing focus shift but retaining wide-open focus would be to use AF cameras where the lens and camera exchange information and then for the focus to be corrected by the camera for the aperture. The Konica Hexar AF did this and was designed more than 20 years ago.

I am a little more tolerant of focus shift and softness than Chris is, but only a little. M7, Neopan 400, Noctilux, showing shift in the Hermitage, St Petersburg.

Marty

 

[rxmd]

For such people, the focus shift 'doesn't exist' or can be compensated by leaning in or out a bit'. The stone cold fact is, that for the careful worker (and if you are going to spend $1000+ for a lens, you damned well ought to be quality conscious enough to do it right) this lens is an f2.8 lens.

Precision fanatics are of course free to use their lenses only at f/2.8 or lower, but I don't see anything wrong with getting to know your gear well enough that you can actually use it at higher apertures and be happy with the results. Fast lenses have been exhibiting focus shift for ages. Somehow those people in the 1930s to 1950s still managed to get pictures in focus with their Sonnars - by knowing how far you have to lean in or out, by knowing how to compensate with the rangefinder, or whatever more or less accurate method would give them pictures they were happy with.

Incidentally the idea that a lens has two extra fast stops just so that a slow stop can be more accurate is a bit misleading; without the design decisions to make the lens fast, there wouldn't be much focus shift in the first place. A f/1.5 lens has that aperture for taking pictures at f/1.5 and nothing else. If the user finds f/1.5 shooting too inaccurate for their taste, it's basically the user's own decision.

 

[Roger Hicks]

Hmm, but how would the cam know what aperture the lens is set to? Seems to me the focus is still going to shift as we stop down.

Red face -- brain fade! You are of course absolutely right and I was absolutely wrong. All I can plead is that it was late after a good dinner with too much wine.

Cheers,

R.

 

[Roger Hicks]

When shooting, you don't carry a ruler, so you have to guess how much to change the focus if you want to get it right, 100% of the time, perfect, no exceptions ever. I can't tolerate guesswork, and if a photo is even half a centimeter out of focus, its ruined for me and cannot be used. For the work I do, precision is a non-negotiable requirement, which is why I don't use mine at larger apertures than f2.8

A lot of people on RFF post images from their Leicas with fast lenses (not just this one...with lenses that don't focus shift a lot of people are just sloppy workers or have cameras and lenses needing repair) that are slightly out of focus. Why bother; if you 're not going to do it right, don't waste your time. For such people, the focus shift 'doesn't exist' or can be compensated by leaning in or out a bit'. The stone cold fact is, that for the careful worker (and if you are going to spend $1000+ for a lens, you damned well ought to be quality conscious enough to do it right) this lens is an f2.8 lens. The two largest openings are not for use, they're to allow the lens to focus shift down to where it is accurate at the real max aperture of f2.8 (if they made the lens as an actual f2.8 lens, it would likely not be 'usable' until f5.6!)

Dear Chris,

The other 'stone cold fact' is that most people can lean forward/back with sufficient precision to get a sharp image. Although the d-o-f at f/1.5 and 1 metre is tiny, as rxmd said, an awful lot of people have successfully used lenses with focus shift for many decades. To dismiss them all as 'sloppy workers' is stretching credibility a bit.

Also, you're just plain wrong about 'The two largest openings are not for use', again for the reasons rxmd and Marty gave. My lens is optimized for focus at f/1.5, so it's a sharp as it can be at f/1.5 (which isn't very, because f/1.5 Sonnars aren't at full bore), and as it stops down, the d-o-f increases quite rapidly -- your 'half a centimetre' is something of an exaggeration at f/2.8 -- with a SMALL shift backwards. Leaning forward 2-4" takes care of it.

Cheers,

R.

 

[maddoc]

Hmm, but how would the cam know what aperture the lens is set to? Seems to me the focus is still going to shift as we stop down.

That might be possible with a double-helicoil, one for the focus and a second governed by the aperture setting mechanism. Not like the FLE mechanism (shifting lens elements gradually with focus) but shifting focus gradually with f-stop.

 

[dave lackey]

Focus shift is a property of all optical systems. All lenses display focus shift; the question for practical purposes is whether the point at which the focus system indicates is within or outside the acceptable depth of field. Problems - out of focus photos - occur when the point at which you focus using the rangefinder in an RF camera or the ground glass in an SLR is outside the depth of field. The image projected rearwards has some room to shift if you use a recording medium with depth like film, but not on a digital sensor.

If the CZ Sonnar 50 was an f2.8 lens the design would probably be sufficiently conservative, and would control spherical aberrations sufficiently well, that there would probably be negligible focus shift. The residual aberrations and the focus shift they cause, magnify as the lens design is 'stretched' - i.e. the further from the native relative focal length for the design you go and the faster you make it.

Mechanically linking the aperture to the RF mechanism is possible but would be impossibly complicated. The best current strategy for managing focus shift but retaining wide-open focus would be to use AF cameras where the lens and camera exchange information and then for the focus to be corrected by the camera for the aperture. The Konica Hexar AF did this and was designed more than 20 years ago.

I am a little more tolerant of focus shift and softness than Chris is, but only a little. M7, Neopan 400, Noctilux, showing shift in the Hermitage, St Petersburg.

Marty

Hmmm...please point out where this focus shift is in this image.🙂

 

[john_s]

.......... All I can plead is that it was late after a good dinner with too much wine.

Cheers,

R.

It's not the best time for pontification, as I know well!

 

[Freakscene]

Hmmm...please point out where this focus shift is in this image.🙂

Focused on the dude, got the painting....Noctilux at 2.8 or 4, I think.

Marty

 

[dave lackey]

Hmmm...is this something that occurs with every image or is it infrequent?😕 The reason I ask is not knowing which camera, focus system, etc. I don't know what to think but it seems that the focus occurred right in the center of the image but the top half is more in focus than the lower left portion. Almost like a zone of out-of-focus area when compared to the rest of the image. Weird.

If it happens all the time, what can you do to fix it? If it is infrequent, what causes it? Questions that I would like to hear solved as I have never encountered this problem.:angel:

 

[Benjamin Marks]

To just return to the OP for a moment and to provide a gloss: focus shift refers to the behavior of certain lenses as the aperture is changed from some optimal aperture. On an SLR, you are generally focusing through the lens when it is wide open. On an RF you are using the gradient of an RF-cam adjusted to provide the best focus at a given aperture using the RF system. The "feature" is inherent in the lens's design and it doesn't matter when in the process you stop the lens down (unless you are using an SLR with the lens already stopped down or camera with an EVF and focus after you stop down). Remember, with an RF camera you are just using a complex mechanical device to translate the position of a cam inside the lens to the camera's internal RF mechanism (e.g. the rangefinder patch). Since the RF mechanism inside the camera is generic, it has no way of knowing which lens you've attached or what its aperture is. Similarly, the lens's cam has no way of knowing what aperture you have chosen. So with an RF system you are really left with learning the behavior of the lens and adapting your shooting style to it in the situations where the issue is likely to cause a problem.

If is not difficult to see the behavior in action, but you have to have the right gear (horse before the cart here). For instance, if you put the 50 C-Sonnar that Chris described on a M-4/3 camera, place your camera on a tripod and focus on some nearby subject where it is easy to see these changes (say a ruler), you can actually see the zone of best focus "walk" a little forward and then backwards around the point of focus you have chosen as you change the lens's aperture. By f:5.6 or so the depth-of-focus from your chosen aperture will cover the change in location of best focus enough so that you won't be able to notice.

How much does it matter? Well I have two of those focus-shifting C-Sonnars that Chris describes: one for a Leica M and one for a Nikon-S mount, and other lenses that do it too (C/V Nockton 35/1.4, C/V 28/2). I love 'em and learning to use them is not too tricky.

Ben

[Edit: Dave: I'd put it this way: It is always there. It is a feature inherent in certain lens designs that tends to be an issue as you stop down from f:1.5 through f:4 or so. Seems to be more of an issue with fast lenses than with slow ones in my thoroughly unscientific experience. The feature is always present in a particular lens design at the problem apertures. My 50 C-Sonnar in M mount is optimized at f:2.8 -- that is to say, at that f-stop the focus indication provided by the camera's RF patch is always accurate. Used wide open, there's a bit of shift and stopped down to f:4 there's a bit of shift in the other direction. Keeps you on your toes. BTW, I think the picture of the guy in front of the painting is engaging. At the resolution presented, I would not have noticed that the man was slightly out of focus. I think maybe we worry about this more in the current pixel-peeping age than we did in the world of silver halides . . . what do you think?]

Ben

 

[john_s]

........but I don't see anything wrong with getting to know your gear well enough that you can actually use it at higher apertures and be happy with the results. Fast lenses have been exhibiting focus shift for ages. Somehow those people in the 1930s to 1950s still managed to get pictures in focus with their Sonnars - by knowing how far you have to lean in or out, by knowing how to compensate with the rangefinder, or whatever more or less accurate method would give them pictures they were happy with.........

True, and an argument for not having too much gear: just get to know a small amount of gear really intimately.

(I don't always practise what I preach, but I'm trying!)