Focus Shift

[Samotano]

Reading Michael's mirrorless camera and focus shift article, and especially, Sen Reid's addendum on focus shift, I'm wondering if anyone is aware of a test to check a lens for focus shift.  I find myself using live view more and more just for the purpose of achieving more accurate focus but have to open my lens to its widest aperture in order to best assess focus.  Now I'm wondering if that may cause more harm when I step down to f/18 or so.

[ErikKaffehr]

Hi,

Focus shift occurs when you stop down from large aperture (like f/1.4) to moderate apertures (like f/4), from f/4 down there will be probably very little focus shift.

Focus shift doesn't really exist by the way. It's just that many large aperture lenses are badly corrected, and peripheral rays are not focused at the same plane as the central rays. So the lens has two planes of focus, one for the peripheral rays and one for the central ones. Stopping down removes the peripheral rays so only the central rays remain.

So focus shift is essentially a bad excuse for a bed lens design!

If you use live view, there is a great chance that the peripheral rays are already masked by the sensor itself, so the effect may be reduced. So first suggestion is don't care, second suggestion is really that a lens with focus shift is not really a good design, third suggestion is that it may be best to focus at medium aperture if possible.

Best regards
Erik

[Nacnud]

Try the DoF preview button; mine works in Live View.
I find it very useful with a severely tilted lens for checking if the focal plane is deep enough.

[Jim Pascoe]

So focus shift is essentially a bad excuse for a bed lens design!

If you use live view, there is a great chance that the peripheral rays are already masked by the sensor itself, so the effect may be reduced. So first suggestion is don't care, second suggestion is really that a lens with focus shift is not really a good design, third suggestion is that it may be best to focus at medium aperture if possible.

Best regards
Erik

Eric

Is this really true?  Is it not that it is hard if not impossible to correct really wide-aperture lenses for the effect of focus shift.  Some of the very best lenses suffer from this, including as far as I know the Zeiss 50mm 1.4 and the Canon 50mm 1.2L.  It is one reason why I bought a Zeiss 50mm f2.  I'm far from being a lens expert but I didn't think it was so much bad lens design as a compromise when designing lenses that perhaps have to be built with regard to cost amongst other considerations.

Regards

Jim

[ErikKaffehr]

Hi,

Yes, I actually mean bad design. With live view we can now focus stopped down, but how do you focus a lens shifting focus when using full aperture?! If you cannot design a lens with consistent focusing at f/1.4 while don't you make a lens at f/2 that is consistent? Most lenses will be not used at full aperture most of the time!

Best regards
Erik

Eric

Is this really true?  Is it not that it is hard if not impossible to correct really wide-aperture lenses for the effect of focus shift.  Some of the very best lenses suffer from this, including as far as I know the Zeiss 50mm 1.4 and the Canon 50mm 1.2L.  It is one reason why I bought a Zeiss 50mm f2.  I'm far from being a lens expert but I didn't think it was so much bad lens design as a compromise when designing lenses that perhaps have to be built with regard to cost amongst other considerations.

Regards

Jim

[Jim Pascoe]

Hi,

Yes, I actually mean bad design. With live view we can now focus stopped down, but how do you focus a lens shifting focus when using full aperture?! If you cannot design a lens with consistent focusing at f/1.4 while don't you make a lens at f/2 that is consistent? Most lenses will be not used at full aperture most of the time!

Best regards
Erik

Eric, perhaps you mis-understood my point (or possibly I misunderstand myself).  Maybe I should rephrase it and ask, is it possible to build a lens, at a sensible cost, that is say f1.4 or faster, and does not suffer from the effects of focus shift when stopped down?  Are there any such lenses?  Of course some photographers want a very fast lens because of shooting in low light or for shallow depth of field, and so for them perhaps f2 is not enough. The Zeiss 50mm 1.4 for example is not a cheap lens, but is a lot cheaper than the Canon 1.2.  If they both exhibit focus shift when stopped down, is that bad design or a design compromise?  I am assuming you mean 'bad' as in not enough attention to detail by the designers or that they could have done better if they though it was an important element. Or are you just saying that building lenses with a very wide aperture is just a 'bad' idea in general because of the impossibility of correcting such designs.

Regards

Jim

[ErikKaffehr]

Hi,

I had two Minolta 50/1.4 lenses, both were awful at full aperture, but they don't seem to have a lot of focus shift.

AF actually is normally optimized for f/2.8 so it is not really good for focusing at wider aperture, and my Sony Alpha 900 essentially refuses to use f/1.4, unless I force it to use full aperture.

If you check f/1.4 lenses on Photozone they are not very impressive. I have checked trough a lot of lenses on Photozone and it seems that the Zeiss 85/1.4 may be quite OK, with relatively little focus shift. Photozone has usually an image on page three, illustrating longitudional chromatic aberration, and that image also illustrates focus shift (mouse over for different apertures).

Best regards
Erik

Eric, perhaps you mis-understood my point (or possibly I misunderstand myself).  Maybe I should rephrase it and ask, is it possible to build a lens, at a sensible cost, that is say f1.4 or faster, and does not suffer from the effects of focus shift when stopped down?  Are there any such lenses?  Of course some photographers want a very fast lens because of shooting in low light or for shallow depth of field, and so for them perhaps f2 is not enough. The Zeiss 50mm 1.4 for example is not a cheap lens, but is a lot cheaper than the Canon 1.2.  If they both exhibit focus shift when stopped down, is that bad design or a design compromise?  I am assuming you mean 'bad' as in not enough attention to detail by the designers or that they could have done better if they though it was an important element. Or are you just saying that building lenses with a very wide aperture is just a 'bad' idea in general because of the impossibility of correcting such designs.

Regards

Jim

[rickk]

I'll agree with Erik's assessment of the Minolta 50/1.4 lenses being awful wide open (my samples were MC and MD manual lenses and a Maxxum AF lens). Of course, the equivalent products I've used over the years by Nikon, Olympus, and Canon weren't any better at f/1.4 either. Fortunately, all those 50s produced great images at moderate apertures. Now that we can use them all on the same Sony NEX (or m4/3) body, it will be interesting to compare with one less variable.

I was intrigued by Erik's comment regarding AF optimization for a particular aperture -- something I had not thought about or read about previously. Additional details and practical consequences of that optimization would be appreciated (especially for the A900).

Regards,

Rick

[ErikKaffehr]

Hi,

Check this: http://en.wikipedia.org/wiki/Autofocus#Phase_detection

What I know is that most cameras have different sensors, some work at medium apertures and some work with large apertures. I have no proof that the large aperture sensors are optimized for f/2.8 , they work with f/2.8 or better.

If we assume that there is a significant focus shift between f/1.4 and f/2.8 it would probably be preferable to focus at f/2.8, and my impression is that AF is doing just that. But I do not know for sure.

The way phase sensing AF works is that it uses rays coming from opposite sides of the aperture. Focus shift is caused by the lens failing to focus the peripheral rays at the same plane as the central rays. So fully utilizing the aperture for AF would mean that central rays would always be defocused. Using f/2.8 is probably a decent compromise.

With present sensor designs, some of the peripheral rays are lost, see this article: http://www.luminous-landscape.com/essays/an_open_letter_to_the_major_camera_manufacturers.shtml. These effect may reduce the focus shift somewhat.


Best regards
Erik

I'll agree with Erik's assessment of the Minolta 50/1.4 lenses being awful wide open (my samples were MC and MD manual lenses and a Maxxum AF lens). Of course, the equivalent products I've used over the years by Nikon, Olympus, and Canon weren't any better at f/1.4 either. Fortunately, all those 50s produced great images at moderate apertures. Now that we can use them all on the same Sony NEX (or m4/3) body, it will be interesting to compare with one less variable.

I was intrigued by Erik's comment regarding AF optimization for a particular aperture -- something I had not thought about or read about previously. Additional details and practical consequences of that optimization would be appreciated (especially for the A900).

Regards,

Rick

[bjanes]

Hi,

Yes, I actually mean bad design. With live view we can now focus stopped down, but how do you focus a lens shifting focus when using full aperture?! If you cannot design a lens with consistent focusing at f/1.4 while don't you make a lens at f/2 that is consistent? Most lenses will be not used at full aperture most of the time!

Best regards
Erik

Erik,

I don't understand the purpose of the resolution figures you posted for the two lenses. The fast lens exhibits curvature of field, but AFAIK this has little to do with the spherical aberration which causes focus shift.

Regards,

Bill

[ErikKaffehr]

Hi,

I guess that the idea was to show that lens sharpness suffers at large apertures. I can agree it's not really relevant in the context, sorry for that!

Best regards
Erik

Erik,

I don't understand the purpose of the resolution figures you posted for the two lenses. The fast lens exhibits curvature of field, but AFAIK this has little to do with the spherical aberration which causes focus shift.

Regards,

Bill