Of Lenses and Cameras - Warning, Long...

[Jager]

Advertisement (gone after registration)

This is a copy of an article on my website...

Of Lenses and Cameras

 

 

Photography has always appealed to people with an outsized appreciation for gear. The finely wrought lenses and camera bodies which enable this passionate endeavor of ours present an evocative counterpoint to photography itself - where the rendering of an image is the thing. Many photo hobbyists become far more focused on the equipment side of the practice than to the pursuit of emotive images. Some fall so deeply into that rabbit hole that that is all they see. They live in a world of arcane measurement. They obsess over the most minute performance faults. They hold manufacturers at fault for failing to achieve industrial perfection.

 

Nothing wrong with that, I suppose. At least as long as one can continue to make the distinction between photography - the capturing of images - and the gear which makes it possible.

 

[Removed text to get under 10K char limit]

 

That's okay. Perfection is a state of mind, not something you'll find on the factory floor.

 

All of which is to say, I've always looked on with bemusement at those who like to photograph brick walls, trying to ascertain resolution deficiencies in their lenses; or finely-graduated yardsticks, figuring to find fault in a camera's focusing mechanism. To each their own. But that's not my idea of photography.

 

Me, I just shoot a bunch of pictures, normal pictures of people and such, and if most of them seem okay, that's good enough.

 

[Removed text to get under 10K char limit]

 

Which brings me to a confession: Over the last couple of days I've done a bunch of those very same gear-obsessed tests that I have long eschewed. The obsessive-compulsive, anal-rententive gear whores out there would be proud.

 

There was, I would protest though, a purpose.

 

As a bit of backdrop, I would note that I bought an 85/1.4 Nikkor some years back. As many of you know, this is a renowned lens, one intended for use wide-open, with glorious bokeh. Over on Nikon Cafe they call it "The Cream Machine." It's one of those rare non-Leica lenses which follows the Leica gestalt. You can happily use all the apertures you paid for, not just those you stopped down a few clicks.

 

Alas, mine seemed a bit soft.

 

Nearly a year ago I came across a new product called the "Lens Align Pro," a device for verifying focus accuracy in DSLR's. After noting several very positive reviews by photography people I respect, I went ahead and ordered it. It's one of those things that is simple in concept, but brilliant in execution. It's way more expensive than it ought to be - just because it IS simple. What you're paying for is that slap-yourself-in-the-forehead, why-didn't-I-think-of-that bit of epiphany. I rationalized the cost by simply contrasting it against the glass that I've invested in over the years. Compared to that this little device was a tifle.

 

More importantly, it worked.

 

Within ten minutes I was able to confirm that that 85/1.4 Nikkor of mine was front-focusing a significant amount. A quick trip to the menus on my D3 and a tweak to the AF Fine-Tune setting for that lens... problem solved.

 

Fast forward ten months and back to my Leicas, the real story here.

 

When my M9 arrived I mounted my Summicron 50 on it. This was the lens I had purchased along with my M6, my first Leica years ago, and that 50 had been my happy companion over many miles. But like the wife cast aside for the exciting new mistress, when the M8 came along that Cron had been superceded by a newer and prettier 50, a chrome Lux ASPH. For three years that new Lux shared all the honors with an also-new 28 Cron. The Summicron 50 sat on the shelf.

 

Even mistresses eventually grow familiar, though. And so with the return of full-frame I figured it was right to go back to the lens that had started it all. Even if just for a little while.

 

I fell in love again. I had forgotten how stirringly beautiful that 50 Cron rendered images. Putting it on the M9 had reminded me.

 

Days passed. And I needed to shoot an event at work. The conference room had crappy fluorescent lighting. And not nearly enough of it. I needed something fast. So back to the mistress I went.

 

Hmm. You shoot 150 images and you usually get an idea of things. My idea was that these weren't what I expected. Despite the slow shutter speeds and the capture-the-moment hair trigger requirement, too many seemed just a little soft. Something that had never happened with that 50 Lux before.

 

Thought turned to worry. And that to wonder. And then I fell down the rabbit hole.

 

On Saturday I pulled out my M9, that once-used Lens Align Pro, two tripods, and a handful of M lenses. I set the tripods three feet apart, a distance just beyond their minimum focusing requirement. I carefully set up the Lens Align Pro. I screwed a 1.25 magnifier into the viewfinder window. I mounted the camera and took a baseline shot to ensure everything was in alignment. And then, for each lens, I took six carefully-focused shots at its maximum aperture and six more at its next-to-maximum aperture. I deliberately re-focused between each shot, moving the rangefinder back and forth, seeking perfect focus, just like I do when taking real pictures. And I placed a Post-It note on the Lens Align Pro device so I'd have a permanent record of the shot details.

 

 

 

 

Making sure everything is aligned.

 

 

 

 

 

The test setup

 

 

 

 

 

Reading the results

 

 

 

I did this for my 35 Lux ASPH, my 50 Lux ASPH, and my 50 Cron. Then I moved the tripods another foot apart, to accomodate my last two lenses - the Noct, and my 90 Elmarit - and took the frames for those. Then I loaded the images onto my computer to take a look.

 

I was aghast. My M9 was front-focusing on everything save the Noct, which was seriously back-focusing. I went to bed thinking I'd be sending the camera to New Jersey.

 

On Sunday I decided to do some more tests. It would be another day before I could ship it, after all.

 

Once again I pulled out the tripods, the Lens Align Pro, my lenses, and both my M8 and M9. In the three years I had shot with it I never had the sense that my M8 had ever been off in any way. And so I was most curious to see how it compared to what I had seen on my M9.

 

I repeated all the shots, with all the same lenses, from the day before. First with the M8. Then again with the M9.

 

I did one more thing. Since I seemed to have a problem with front-focusing on my M9, I wondered how tweaking the focus ever so slightly would affect what I was seeing. So when that 50 Lux was mounted, after taking the normal, perfectly-focused shots, I also took a dozen shots with the rangefinder turned just a smidge towards the rear, towards infinity. Just a hair, just enough to come off perfect-focus.

 

The results were enlightening.

 

Having loaded everything into a spreadsheet...

 

 

 

 

 

 

I was able to draw some conclusions.

 

My M8 also shows consistent, albeit very modest, front-focus, despite never having empirically raised any concerns in my mind.

 

The two separate M9 tests, a day apart, showed a statistically meaningful divergence. In both instances the camera demonstrated front-focusing, but Sunday's tests were of a more modest degree - approximating the same very slight amount as in my M8.

 

Most importantly, those twelve shots I took with the 50 Lux, where I deliberately tried to counteract the front-focus by moving the rangefinder a tiny smidge to the rear, all showed very significant back focus. Said differently, an almost infinitesimal nudge of the rangefinder dramatically changed the apparent focus. We're talking the very slightest of nuance here, folks.

 

 

 

That last raised what seems to me to be the most important question here: At what point does any of this become meaningful? The great benefit of the Lens Align Tool is that it allows the most subtle gradations of measurement. But since manually focusing a lens involves a mechanical linkage, not to mention my own imperfect 56-year-old eyes, when do you decide that a problem truly exists?

 

The conclusion I've drawn is that my M9 is fine, as was my M8 before it. Leica may or may not deliberately inject a tinge of front-focus into their rangefinder calibration. I dunno. What I do know is that the shot-to-shot variation I see when trying to measure it is larger than any perceived level of error. If anything, I have a renewed respect for the inherent accuracy of Leica's rangefinder design. It's really quite amazing that these things work as well as they do.

 

And so out of the rabbit hole I climb.

 

I do take away a few other things, though. My Noct most definitely back-focuses to an unacceptable degree. That explains the too-many-soft-shots I always got with that lens, something I just attributed to its character. And my 90 Elmarit front-focuses way too much. both those lenses will be going back to New Jersey for re-calibration.

 

And my 35 Lux ASPH clearly exhibits focus shift, something a lot of folks deny, but something that is inherently part of this lens' design. In my case it starts out as very modest front focus at f1.4, moving to very slight back-focus at f2. That's optimized behavior for this lens, in my opinion. And it's something I had long inferred from my sample. Now I can actually see it.

 

And now I shall be glad to get back to taking real pictures, of real people and real things.

[Advertisement]

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

[wlaidlaw]

Jeff,

 

The very first thing I do after I have bought a new or second hand lens is run a focus test using the Nikon D700 chart that many of us on this forum use. I print it to A3+ to make it easier to read (the focus distances are nearly X3). If it is not focusing properly, I either return it or have it adjusted by agreement at the expense of the seller. I also check rangefinder convergence by using either a planet or the edge of the moon in the UK and some street lights 1000 feet below and 7 miles away in France.

 

My record to date:

Voigtlander - useless - miles off and often difficult to adjust. (3 lenses 28/1.9, 35/1.2 and 35/2.5)

Zeiss - all three spot on (21/2.8 35/2 and 50/2)

Leica - too many to mention but a bit of a mixed bunch. Usually not too difficult to get right.

 

Wilson

[cgthanos]

Excellent summary... I did nearly the same thing with my m9 several weeks ago, with very similar results (although more exaggerated front-focusing). The results on the 75 cron were even more disappointing, and I couldn't envision ever being able to compensate for this aberration in the field. I decided on a rangefinder calibration, which brought all of the lenses (35lux, 50lux, 50cron, 75cron) into perfect focus. I've read other similar reports on this site.

 

I realize that your last test gives you the impression that your focusing issues are within the range of human error, e.g. the slight intentional twist of the focus ring that resulted in over compensation. But I encourage you to have the unit calibrated -- for me, the improvement in focus and sharpness was significant, and my real world focusing (eyes, etc) is now spot-on, something I can't say ever was the case with my 1ds3.

[Guest mc_k]

have read a couple of other threads recently where a second go at the focus test was required. May be a good idea to use 2 of the 3: newspaper at 45 degrees, ruler, test chart. If your lens/camera is going back for a couple of months, make sure you have the guilty party. I think you should also be testing at infinity, maybe other distances.

[wlaidlaw]

have read a couple of other threads recently where a second go at the focus test was required. May be a good idea to use 2 of the 3: newspaper at 45 degrees, ruler, test chart. If your lens/camera is going back for a couple of months, make sure you have the guilty party. I think you should also be testing at infinity, maybe other distances.

 

To test at infinity, you really need a lens collimator and/or an optical test bench. I am exceeding lucky that just 6 miles away from me I have Kelvin of Protech, who has these things and will check a lens for me while I wait.

 

Wilson

[johnbuckley]

Jeff - I admire obsessives. People who fall headfirst down the rabbit hole are okay in my book. And it should be noted that there is a special kind of headache, not replicable doing anything else, that settles upon the skull when measuring lenses as precisely as you have. But I have a confession to make.

 

I don't think my 35 Summilux Asph has any degree of focus shift. I'm sure if you tested it like you have, it would certainly be revealed. The reason I don't think it has focus shift issues is because I've learned to let go, sort of. I figure I shouldn't fret over that the 90 percent of the time that I'm using it where the focus shift would be immaterial. The 10 percent of the time when it's really, really important, I try taking extra care -- chimping with my reading glasses on -- til I get what I'm in need of. You don't always have that luxury, I know, and admittedly, I'm not a pro.

 

But as obsessive as I am, I've learned to accept a half inch front focus here and there.

 

Because to not learn to live with it would lead to a) being twisted like a pretzel all night, not sleeping, c) being on a first name basis with the Fed Ex guy who ships stuff to New Jersey or Solms.

 

As between perfection and Buddhist detachment, I'll take the softening of a prime bokeh.

 

Of course, I did sell my original Noctilux, because it was driving me a little crazy...

[Scott Root]

Advertisement (gone after registration)

Excellent summary... I did nearly the same thing with my m9 several weeks ago, with very similar results (although more exaggerated front-focusing). The results on the 75 cron were even more disappointing, and I couldn't envision ever being able to compensate for this aberration in the field. I decided on a rangefinder calibration, which brought all of the lenses (35lux, 50lux, 50cron, 75cron) into perfect focus. I've read other similar reports on this site.

 

I realize that your last test gives you the impression that your focusing issues are within the range of human error, e.g. the slight intentional twist of the focus ring that resulted in over compensation. But I encourage you to have the unit calibrated -- for me, the improvement in focus and sharpness was significant, and my real world focusing (eyes, etc) is now spot-on, something I can't say ever was the case with my 1ds3.

 

I thought the 75 cron asph had floating elments and no focus shift, but please correct me otherwise?

[cgthanos]

Scott

 

You are right... I was just pointing out that the front focusing behavior, shot wide open, was most pronounced with the 75 f/2, and it wasn't until I examined that lens in detail that I realized that all of the lenses were front focusing. That lead to the conclusion that it was probably a problem with the rangefinder, and after calibration, all of the focus issues disappeared.

 

Chris

[250swb]

Isn't there is always going to be a tendency towards front focus shift until the aperture is stopped down? I find focus shift stops on Leica lenses around F4-f5.6. And it always seems to be a similar case, such that front focus is reported but aperture related focus shift is not factored in. I'd be interested if anybody knew if Leica calibrate the lens for its widest aperture, and risk focus being out at f5.6, or do they cabilbrate it for being correct at f5.6, but therefore out at its widest? I would assume Leica calibrate it at the point that aperture related focus shift stops/is reduced otherwise its a moving playing field. Just a thought.

 

Steve

[lars_bergquist]

If you take the lens as such, you can only set 'focus zero' at the widest aperture, and compare against that. In that sense, no lens does ever exhibit front or back focus at that aperture -- by definition! Focus shift, which is what we are talking about, happens as we stop down.

 

So if a lens shows front or back focus wide open, this is a matter not of the lens design, but of the setting of the lens in manufacture, and of the adjustment of the rangefinder. Leica do adjust to focus at maximum aperture, BUT -- you must also specify the distance! Focus is not exactly the same across the entire focusing range. I do not know at what distance Leica do adjust their lenses.

 

There is a rumour that Zeiss calibrate their lenses at one stop down, not at full aperture.

 

Now the Summilux-M 1:1.4/50mm ASPH and the Summicron-M APO 1:2/75mm have floating elements, which have the double job of maintaining definition at close focusing ranges (a notorious problem with very fast lenses) and neutralizing residual spherical aberration that would cause focus shift. So both of these lenses should not only be spot on wide open, but remain so as you stop them down. And that is exactly what my 50mm Summilux does. Any discrepancy is due to an error in the adjustment of the lens or the rangefinder, and you will have to send both the lens AND the camera to a highly qualified shop, with a collimator and somebody who knows how to use it. And you may find to your horror that you will have to do the same with all your other lenses ...

 

Have fun.

 

The old man from the Easy Age of Scale Focusing

[Guest mc_k]

originally the 50 C-Sonnar was calibrated a couple of stops down, to compensate for shift...(not just a rumor)

[tobey bilek]

I just used a yard stick at 45 deg. Brilliant.

 

I found the 50 1.8 Nikkor does a few interesting things.

 

First, set it 1.8 and 5.6 is off. Set it to 5.6 and 1.4 is off. Must be focus shift.

 

Second. Compromised on getting 5..6 correct. Adjusted the camera compensation figuring all is now well.

 

NOT. The camera now thinks 20` is infinity. Close and far corrections are different.

 

More brilliant. Reset the camera to oriiginal and put the yardstick back in the closet until someday when I need it to measure distance, not calibrate lenses.

 

Smile because I learned a few lessions and saved the cost of the machine and stopped driving my self crazy over a non existent problem.

 

I decided Michael`s machine is great for calibrating macro lenses, not general purpose photo lenses. I learned to use macro lenses for close work. I learned auto focus is not for macro work. I learned to buy a Katz Eye screen for my D700. Every shot is perfectly sharp every time just as when I used my R cameras.

 

I have always known plane of focus setting is different for film and digital. Film is not flat so there is a compromise. The film channel has debth. The digi sensor is perfect distance and flat in theory. So the setting used to reach a compromise with film will be off with digital.

 

So your precious film/digi lenses are not interchangeable between cameras. Quit obcessing over it and work on taking good pics. If the focus is off 2" at 20 feet nobody will notice. Put your viso on for close work and do not depend on a RF. Oh wait, is the viso screen set for film or digi? Now where did I put that stick?

[250swb]

I just used a yard stick at 45 deg. Brilliant.

 

I found the 50 1.8 Nikkor does a few interesting things.

 

First, set it 1.8 and 5.6 is off. Set it to 5.6 and 1.4 is off. Must be focus shift.

 

Second. Compromised on getting 5..6 correct. Adjusted the camera compensation figuring all is now well.

 

NOT. The camera now thinks 20` is infinity. Close and far corrections are different.

 

More brilliant. Reset the camera to oriiginal and put the yardstick back in the closet until someday when I need it to measure distance, not calibrate lenses.

 

Smile because I learned a few lessions and saved the cost of the machine and stopped driving my self crazy over a non existent problem.

 

I decided Michael`s machine is great for calibrating macro lenses, not general purpose photo lenses. I learned to use macro lenses for close work. I learned auto focus is not for macro work. I learned to buy a Katz Eye screen for my D700. Every shot is perfectly sharp every time just as when I used my R cameras.

 

I have always known plane of focus setting is different for film and digital. Film is not flat so there is a compromise. The film channel has debth. The digi sensor is perfect distance and flat in theory. So the setting used to reach a compromise with film will be off with digital.

 

So your precious film/digi lenses are not interchangeable between cameras. Quit obcessing over it and work on taking good pics. If the focus is off 2" at 20 feet nobody will notice. Put your viso on for close work and do not depend on a RF. Oh wait, is the viso screen set for film or digi? Now where did I put that stick?

 

Thats kind of what I was thinking Tobey, I don't understand how it is possible to set up the camera/lenses to cover the full range accurately over all the different aperture/distance variations.

 

Given there are no 'digi' lenses, if Leica do anything it is to optimise closest focus at say f5.6 or thereabouts and the rest falls into place with increased distance from the subject, in other words the compromise is around a given f stop and closest focus point, but not the widest f stop and closest focus point. So you wouldn't base the necessity for an average calibration (to cover the full distance/aperture range) around an extreme aspect of that range (wide open and closest focus), would you?

 

I've just done my own crude tests. At the closet focus distance all my lenses front focus wide open. If I stop down a modest amount (f4) they are all spot on and super accurate (the focus point moves back to what the rangfinder says is in focus). If I gradually move my tripod further away from my target (a slanting ruler) the focus gradually becomes more accurate wide open, and is as accurate as ever stopped down. Beyond a couple of yards there is no important discernable difference between wide open and stopped down. This seems a sensible compromise to me given my M9 is not a macro camera? Am I wrong?

 

Steve

[Guest mc_k]

Thats kind of what I was thinking Tobey, I don't understand how it is possible to set up the camera/lenses to cover the full range accurately over all the different aperture/distance variations.

 

Given there are no 'digi' lenses, if Leica do anything it is to optimise closest focus at say f5.6 or thereabouts and the rest falls into place with increased distance from the subject...

 

have not heard anything like this before...I have always assumed the lens is calibrated for accuracy wide open, throughout the range.

[Guest mc_k]

...So the setting used to reach a compromise with film will be off with digital...

 

here is my understanding...there is only one standard because a calibrated lens will focus the image at a set distance behind it. The film cameras are apparently more forgiving because the film is not quite flat and the emulsion has depth, so the plane of critical focus is not quite a plane. But the lenses are not "off" with digital cameras.

[250swb]

have not heard anything like this before...I have always assumed the lens is calibrated for accuracy wide open, throughout the range.

 

I don't know, its conjecture as you will have gleaned from my comments.

 

This isn't really important in the wider sense except that amateur 'tests' are done at close focus and generally wide open, perhaps leading to the sense that front focus is a show stopper, when in fact it is only troublesome if you shoot very close up and wide open most of the time. Front focus is effectively gone by the time the camera is either stopped down a bit, or moved a foot or two further from the subject. In other words its the test that is exaggerating the optical quirk of aperture related focus shift, not real world use of the camera. In effect we are worrying pointlessly about something that appears normal for a Leica lens, despite the fact that it can be dialled out in re-calibration.

 

And given aperture related focus shift seems common, does calibrating the lens to be spot on wide open at closest distance just mean the problem is moved somewhere else in the range, like perhaps manifesting itself as back focus as the lens is stopped down?

 

Steve

[Jager]

Some interesting thoughts and considerations here. Thanks for the input.

 

I did my tests at near minimum focusing distance and at maximum aperture because I assumed that any errors would be most pronounced there - and would then remain consistent throughout the focusing range. I might do some further tests at greater distances to see what that does. Although the increasing depth of field that comes with distance will inevitably mitigate modest front or back focus, I've always assumed that a front-focusing lens will continue to front focus at any distance (focus shift at the 2-3 stops just down from maximum aperture aside).

 

I guess this is a slippery slope I've stumbled upon.

 

It does make you wonder if folks' widely varying views of particular lenses aren't to some degree affected by how well their particular camera bodies interact with them. Could be that a lot of the "I never really cared for such and such lens" might be about sub-optimized lens/body combinations.

 

It would also be interesting to know how exactly Leica calibrates their lenses. If not at minimum focus distance and maximum aperture, then what?

 

Mostly, like John, I'm striving to let all this go. I had to smile at his comment about getting on a first name basis with one's FedEx delivery person. I wish the FedEx folks all the best... but I really don't want to get in the mode of seeing 'em all the time...

[Guest mc_k]

I agree you should put more weight on how your real pictures turn out than the kitchen table focus tests.

 

For a couple of lenses though the focus shift is real enough. The two copies of the 1.0 that I checked shifted a couple of inches by f/2.8 (at 3-4 ft.) and I think would be hard to use there; I haven't even tried. Apparently Leica has in the past set a lens like the 1.0 or 35/1.4 forward at customer request, to compensate for shift.

[Advertisement]

Hallo,

Look here: Leica M