Jump to content

24-90mm Focus Shift (Diglloyd)


agencal

Recommended Posts

Advertisement (gone after registration)

Is there a FAQ on focus shift vs front/back focus?  Sorry, I didn't check.   If not, there should be.....confusion arises regularly, it seems.

 

Jeff

 

 

I'm not aware of a FAQ, but here are the basics...

 

Front/Back focus is where the camera consistently sets focus slightly in front of or slightly behind the actual point of focus.  It would be visible at all apertures in the same amount.  It is most commonly an issue for cameras that use phase detect autofocus systems such as DSLR's since the sensor doing the focusing is not the same physical sensor as the one taking the images.  You can have a mechanical mis-match.  Some newer cameras allow user adjustment to address this issue.  It's not unlike rangefinder calibration that can be an issue with 'M' cameras and, particularly, longer focal length lenses.

 

Focus shift is an attribute of the lens, not an attribute of the camera.  It's where the point of best focus varies with f-stop.  It is generally caused by residual spherical aberration in the lens design.  If the outer portion of the lens elements themselves have a slightly different point of best focus than the central portion of the lens (spherical aberration), you will see the focus shift slightly as you stop down the lens.  Since most cameras--including the Leica SL--always do their auto focusing with the lens wide open, you can see the point of best focus shifting as you stop down.  Most lens designers try to ensure focus shift is small enough that the increasing depth of field as you stop down keeps the focus shift invisible.  I would say that is the case with the 24-90 based on my initial, quick test.  The numeral "7" is clearly the point of best focus at f/4, but by the time you stop down to f/11 the focus point appears to have shifted slightly closer to the camera.  The right-hand side of the image is in better focus overall than the left-hand side in the f/11 sample above, but the "7" is still well within the range of critical sharpness.

 

Cameras using contrast autofocus systems such as the Leica SL should not have an issue with front or back focusing since the main imaging sensor is used to detect focus and to capture the image.  There are no mechanical tolerances involved (as long as the imaging chip is square to the optical axis, but that's a different can of worms).  However, both phase detect and contrast autofocus systems can be susceptible to focus shift since focusing is performed with the lens wide open.  Lenses for range finder cameras are often calibrated at a particular f-stop that represents a good compromise.  I believe Zeiss, for example, calibrates their lenses at f/2.8 whereas Leica calibrates them wide open.  Some owners prefer the Zeiss system, others the Leica (depending on what percentage of their use is with the lens at maximum aperture).

 

Does that explain the difference?

 

- Jared

Edited by Jared
  • Like 7
Link to post
Share on other sites

Leica may not have made aperture shift a priority on this lens, given that it was auto-focus. Similarly with it being parafocal, although it actually seems pretty good on this aspect, if you set it to manual focus and zoom, the focus does not change significantly. All lenses are a compromise between various things and if say you can improve the resolution of the lens or say the focus speed at the cost of some aperture shift, that might be a compromise worth making on an auto focus lens. 

 

Wilson

 

PS I am far from convinced the lens is focusing wide open and then stopping down. You can hear the IS motor running but you don't hear the aperture motor fire up when you take a shot, if you hold your ear against the lens. 

 

W.

 

 

It definitely focuses wide open.  Just set it to f/16, point it at yourself, put it on AF-C mode (so you get shutter priority rather than focus priority release), and partially depress the shutter.  You can see the lens shifting as it tries to focus.  Now press the shutter all the way and you will see the aperture diaphragm close when the shutter releases.

 

- Jared

Link to post
Share on other sites

Amongst the first things I did with the SL and 24-90mm lens once it was delivered was to set up and test the lens for parfocal and focus shift behavior. I tested all focal lengths at all lens openings. I found my example of the lens to be parfocal and to exhibit no significant focus shift. I say "significant" because it is unheard of that any lens exhibit 'absolutely no' focus shift, particularly any fast-ish zoom lens; I couldn't see any without inspecting at 4:1 pixel magnification. That's close enough to "none" to satisfy me.  :rolleyes:

 

Whether some other examples do show significant focus shift, well, unit to unit variability applies to the output of any production line. The Vario-Elmarit-SL 24-90mm is no exception. 

  • Like 2
Link to post
Share on other sites

It definitely focuses wide open.  Just set it to f/16, point it at yourself, put it on AF-C mode (so you get shutter priority rather than focus priority release), and partially depress the shutter.  You can see the lens shifting as it tries to focus.  Now press the shutter all the way and you will see the aperture diaphragm close when the shutter releases.

 

- Jared

 

The SL with 24-90 always tries to keep the aperture wide open for focusing and framing, but will occasionally close the lens down a little to help modulate display/evf brightness. However, if your need is absolutely critical focus and you suspect some small amount of focus shift may be influencing the focus performance, press the FN button once to go into 'aperture simulation' (aka DoF preview) mode and check focus at the set aperture. That's what aperture simulation is there for. 

  • Like 1
Link to post
Share on other sites

Amongst the first things I did with the SL and 24-90mm lens once it was delivered was to set up and test the lens for parfocal and focus shift behavior. I tested all focal lengths at all lens openings. I found my example of the lens to be parfocal and to exhibit no significant focus shift. I say "significant" because it is unheard of that any lens exhibit 'absolutely no' focus shift, particularly any fast-ish zoom lens; I couldn't see any without inspecting at 4:1 pixel magnification. That's close enough to "none" to satisfy me.  :rolleyes:

 

Whether some other examples do show significant focus shift, well, unit to unit variability applies to the output of any production line. The Vario-Elmarit-SL 24-90mm is no exception. 

I do believe that both Loyd and Ming had pre-production samples.

Link to post
Share on other sites

I'm not aware of a FAQ, but here are the basics...

 

Front/Back focus is where the camera consistently sets focus slightly in front of or slightly behind the actual point of focus.  It would be visible at all apertures in the same amount.  It is most commonly an issue for cameras that use phase detect autofocus systems such as DSLR's since the sensor doing the focusing is not the same physical sensor as the one taking the images.  You can have a mechanical mis-match.  Some newer cameras allow user adjustment to address this issue.  It's not unlike rangefinder calibration that can be an issue with 'M' cameras and, particularly, longer focal length lenses.

 

Focus shift is an attribute of the lens, not an attribute of the camera.  It's where the point of best focus varies with f-stop.  It is generally caused by residual spherical aberration in the lens design.  If the outer portion of the lens elements themselves have a slightly different point of best focus than the central portion of the lens (spherical aberration), you will see the focus shift slightly as you stop down the lens.  Since most cameras--including the Leica SL--always do their auto focusing with the lens wide open, you can see the point of best focus shifting as you stop down.  Most lens designers try to ensure focus shift is small enough that the increasing depth of field as you stop down keeps the focus shift invisible.  I would say that is the case with the 24-90 based on my initial, quick test.  The numeral "7" is clearly the point of best focus at f/4, but by the time you stop down to f/11 the focus point appears to have shifted slightly closer to the camera.  The right-hand side of the image is in better focus overall than the left-hand side in the f/11 sample above, but the "7" is still well within the range of critical sharpness.

 

Cameras using contrast autofocus systems such as the Leica SL should not have an issue with front or back focusing since the main imaging sensor is used to detect focus and to capture the image.  There are no mechanical tolerances involved (as long as the imaging chip is square to the optical axis, but that's a different can of worms).  However, both phase detect and contrast autofocus systems can be susceptible to focus shift since focusing is performed with the lens wide open.  Lenses for range finder cameras are often calibrated at a particular f-stop that represents a good compromise.  I believe Zeiss, for example, calibrates their lenses at f/2.8 whereas Leica calibrates them wide open.  Some owners prefer the Zeiss system, others the Leica (depending on what percentage of their use is with the lens at maximum aperture).

 

Does that explain the difference?

 

- Jared

 

Lenses can have focus issues beyond focus shift, but typically not nearly to the degree people think.  The issue is more often camera and lens tolerance interactions... http://www.lensrentals.com/blog/2008/12/this-lens-is-soft-and-other-myths

 

Jeff

Link to post
Share on other sites

I do believe that both Loyd and Ming had pre-production samples.

 

I've spoken with Ming, and he (and I think Lloyd as well) have been shooting with current firmware.

I spent a couple of hours today trying to duplicate the problem and failed pretty miserably - I guess it was more prone to front focus than back focus (good as far as I'm concerned). Shooting at all focal lengths and all apertures hasn't brought up any nasties at all. 

And if the wonderful guys here aren't finding it either . . . . 

  • Like 5
Link to post
Share on other sites

Well, my testing was extremely limited--one local length and one focus distance.  No problems with that one test, though (as my samples demonstrated).  Focus was accurate wide open, and point of best focus stayed well with in the depth of field as I stopped down.  Nothing to complain about at this point.

 

- Jared

Link to post
Share on other sites

I too spent a dull hour trying to run some tests ...... using 1 point focussing

 

No difference between f4 and f11.........

 

However the focus point was close to the rear end of DOF limits ..... with a large preponderance of the 'in focus' part of the image in front. 

 

This seems to be contrary to the normal situation with mechanical optical focussing......  and I would have expected the optimum focus point to be bang in the middle of the DOF with an AF system (or am I wrong?)

 

Of course this depends on the translation of the optimum contrast point on the sensor being mapped to exactly the same point on the EVF/LCD ........ but doing the same tests with the camera upside down yielded the same result as the right way up, so there isn't a mismatch. 

 

I assume it is a characteristic of the lens .... or how Leica have fixed the lens firmware ...... although the reasons why elude my simple brain.

 

If my combo is typical then this probably explains some observations of 'front focussing' or apparent focus shift .... as I suspect fairly small user or AF errors might generate this appearance.

 

It certainly doesn't seem to be an issue in normal photography ...... I had a day of portraiture where eye focus is critical and my main reason for poor images where using the default auto iso shutter speed (1/80) at 90mm/f4 with a non static target ..... not focussing ....

 

Of course if you specialise in inclined plane document photography you may have problems ..... :unsure:

Edited by thighslapper
  • Like 1
Link to post
Share on other sites

I too spent a dull hour trying to run some tests ...... using 1 point focussing

 

No difference between f4 and f11.........

 

However the focus point was close to the rear end of DOF limits ..... with a large preponderance of the 'in focus' part of the image in front. 

 

This seems to be contrary to the normal situation with mechanical optical focussing......  and I would have expected the optimum focus point to be bang in the middle of the DOF with an AF system (or am I wrong?)

 

Of course this depends on the translation of the optimum contrast point on the sensor being mapped to exactly the same point on the EVF/LCD ........ but doing the same tests with the camera upside down yielded the same result as the right way up, so there isn't a mismatch. 

 

I assume it is a characteristic of the lens .... or how Leica have fixed the lens firmware ...... although the reasons why elude my simple brain.

 

If my combo is typical then this probably explains some observations of 'front focussing' or apparent focus shift .... as I suspect fairly small user or AF errors might generate this appearance.

 

It certainly doesn't seem to be an issue in normal photography ...... I had a day of portraiture where eye focus is critical and my main reason for poor images where using the default auto iso shutter speed (1/80) at 90mm/f4 with a non static target ..... not focussing ....

 

Of course if you specialise in inclined plane document photography you may have problems ..... :unsure:

 

 

Can you describe your test? 

 

I used manual focus when I tested for focus shift. I didn't see any. Far as I could tell, DoF grows in the normal way as you stop down, with the critical focus plane being at about "1/3 into" the acceptably sharp zone.

 

Never even occurred to me to use the AF system for testing focus shift...! I don't really trust them for critical focus work.

Edited by ramarren
Link to post
Share on other sites

I too spent a dull hour trying to run some tests ...... using 1 point focussing

 

No difference between f4 and f11.........

 

However the focus point was close to the rear end of DOF limits ..... with a large preponderance of the 'in focus' part of the image in front. 

 

This seems to be contrary to the normal situation with mechanical optical focussing......  and I would have expected the optimum focus point to be bang in the middle of the DOF with an AF system (or am I wrong?)

 

Of course this depends on the translation of the optimum contrast point on the sensor being mapped to exactly the same point on the EVF/LCD ........ but doing the same tests with the camera upside down yielded the same result as the right way up, so there isn't a mismatch. 

 

I assume it is a characteristic of the lens .... or how Leica have fixed the lens firmware ...... although the reasons why elude my simple brain.

 

If my combo is typical then this probably explains some observations of 'front focussing' or apparent focus shift .... as I suspect fairly small user or AF errors might generate this appearance.

 

It certainly doesn't seem to be an issue in normal photography ...... I had a day of portraiture where eye focus is critical and my main reason for poor images where using the default auto iso shutter speed (1/80) at 90mm/f4 with a non static target ..... not focussing ....

 

Of course if you specialise in inclined plane document photography you may have problems ..... :unsure:

 

 

Many thanks for your interesting observations and explanations!  Much appreciated.

Maybe Lloyd should be informed about your findings.

 

Edited by k-hawinkler
Link to post
Share on other sites

Can you describe your test? 

 

I used manual focus when I tested for focus shift. I didn't see any. Far as I could tell, DoF grows in the normal way as you stop down, with the critical focus plane being at about "1/3 into" the acceptably sharp zone.

 

Never even occurred to me to use the AF system for testing focus shift...! I don't really trust them for critical focus work.

well, as the whole point of getting the 24-90 zoom is AF, that's what I did.

 

I suppose I could repeat it focussing manually and see what happens ...... but as you have done it and found nothing exciting it seems pointless. 

 

like a lot of critical lens/camera tests they have little relevance to normal everyday use ......

  • Like 3
Link to post
Share on other sites

(I don't own SL but curious)

Doesn't SL body has AF fine tune capability?

 

 

Yes, it's called "manual focus" ...  :rolleyes: Sorry, couldn't help myself. 

 

The SL is a CDAF-only focusing machine. AF fine tuning is for PDAF systems that rely upon precise calibration of expected and actual focus positions in the lens mount; CDAF doesn't work that way at all, there's nothing to adjust since it works by quickly evaluating the point of best contrast to arrive at the correct focus setting. So, in truth, my tongue-in-cheek statement is the correct answer: After the focusing system has locked in the evaluated focus setting, you tweak it for fine tuning (while holding the shutter release at the half press point) by turning the manual focus ring. 

 

This is why when I use the 24-90, my favorite working methodology is to use MF with the toggle-button set to AF Only. That way I press the button, the camera sets its best guess at focus, and I just tweak it to the finish setting if it seems to need an adjustment. 

  • Like 1
Link to post
Share on other sites

Based on Lloyd's most recent post, the issue appears to be that the location of the actual autofocus point doesn't match the location selected in the viewfinder.  The single point crosshairs are wrong.  That isn't something that would get picked up in the sample test results I posted earlier since I was using a flat target that didn't change distance as you move vertically.  I'm a little confused by Lloyd's most recent comments, though... He initially described the problem as one of "focus shift" not inaccurate focus.  I'm not a subscriber to his website, though, so I can't read the full, initial article.  I wouldn't describe the bug he mentioned in his most recent post as "focus shift", though.  More like "inaccurate autofocus point".  Still, it could be an issue in some real world picture taking scenarios.  I'll see if I can confirm it with my own camera.  

 

- Jared

  • Like 2
Link to post
Share on other sites

Lloyd is quite correct.  The focus point shown at normal viewfinder magnification is slightly different from the focus point shown at 6x or 10x magnification.  The actual focus point used is the one shown at 6x or 10x and is slightly lower (landscape view) than the point normally displayed.  I will reiterate:  it probably would be better if this issue isn't referred to as "focus shift" since that has a different, already well understood meaning.  This has nothing to do with the point of best focus changing as the aperture changes.  Hopefully, Leica will fix this quickly in firmware before people start complaining about the poor quality of images they are getting from their 24-90/SL combination.  In the meanwhile, it should be pretty easy for us to adapt our shooting slightly.  Most commonly this would cause a problem when shooting portraits, I suppose, where the actual focus point would be slightly to the right of where you think it is.  

 

- Jared

  • Like 1
Link to post
Share on other sites

Nice to know I wasn't imagining things. 

 

Still can't work out how this happens with AF though ....

 

Easy enough for it to happen with AF.  The camera simply isn't focusing where you told it to.  It's focusing slightly "below" (in the viewfinder).

 

- Jared

Link to post
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

×
×
  • Create New...