How is backfocusing the fault of the lens?

Garbz · Nov 4, 2007

Ok can someone explain this to me in simple terms so my engineering mind can comprehend it?

I see this all the time, most recently about 30 seconds ago in a thread about equipment containing a 18-50mm Sigma lens where someone mentioned the downsides were that there are dodgy units out there with backfocusing issues.

Now how can back focusing be a lens issue? Either the picture is in focus or it isn't. If the distance between the lens element and the 90 degree reflection of the rapid return mirror and finally the focusing screen is the same as the distance between the lens element and the focusing plane, then any shots which are remotely out of focus must fall on to the onus of either the user, or the camera. The former being not paying attention when focusing and the camera has in its opinion correctly focused on something behind the subject, and the latter being that the mirror (not the lens mounted if the laws of physics are to be obeyed) is positioned incorrectly so that the focus does not appear to be the same on the ground glass as the focal plane.

In the end it's the cameras job to look at the incoming light, detect the phase from the split beams in the autofocus module and move the lens focusing motor until it is good. If the picture is then out of focus then the camera made the mistake (or the user did).

The only thing the lens can be really at fault for is being too slow and not letting enough light through for AF to work (like f/6.3 and above lenses may) or not being able to get any focus at all as a result of a misaligned element.

I don't get it. Were the people who reported their back-focusing problems were fixed when they replaced a lens inhaling something at the time?

Sw1tchFX · Nov 4, 2007

the only problems i've had with inaccurate focusing is when i busted the AF motor on my 17-55. It wouldn't' focus past 10 feet.

Iron Flatline · Nov 4, 2007

A lens has several elements, and they need to be built to exact tolerances. A senor also needs to be locked into place. If any of these parameters are off by any amount, you have focusing inconsistencies. Think about it - there's only a few milimeters between the back of the lens and the sensor. If it is off by just a fraction of said milimeter, you start getting into real issues. And that's just with primes. Now imagine a shoddily-built zoon lens, and suddenly you're really compounding the situation.

This is especially true in digital photography. Film itself had a certain depth, esp. color film. The different emulsions carried the three light colors, and your greens were probably razor-sharp but your blues and reds were slightly softer. And of course the actual film-strip had a little play in it. A digital sensor however carries everything on one exct plane... so if the lens is not perfect, you will see it very quickly.

This becomes an issue through the focal range. When you focus on something near or far, you are moving the elements in your lens. It's even more extreme with a zoom lens. Well, at some point the manufacturer needs to calibrate the lens. Typically that's at max aperture, and a certain pre-determined focal distance. That starts changing rapidly as you focus on something 25 yards away. Suddenly you're off by a half yard.

If you really start testing all your lenses in a controlled environment, I bet you'll find that they all back-focus or front-focus a little. Find a billboard, and shoot it at various apertures using a tripod. Also, do it from various distances. And once you've focused, turn off the AF so that it stays constant.

Iron Flatline · Nov 4, 2007

Oh, and on modern SLRs the lens is always wide-open when the camera meters. It sets the aperture as you fire. The DOF Preview button closes down the aperture... may as well rename it the Aperture button.

Garbz · Nov 5, 2007

The only thing that made any sense in that post is that stopping down the aperture can affect the focus point, in which case the focus the camera read would be different to the focus the image takes. In every other case any lens fault would be visible in the viewfinder before pulling the trigger, and on some cameras like the way I have the D200 set atm you wouldn't be able to trigger at all since it's out of focus.

But even that makes little sense. None of the elements move when an aperture is closed (or at least they shouldn't maybe this is why backfocusing is a fault). So if none of the elements move yes the DOF could be biased one side or the other which is visible only now, but the subject that was in focus should is still in focus. Plus everytime I've seen someone complain about backfocusing the image had a small DOF which means that often the aperture changes little if at all.

Everything else like the back element of the lens being closer to the sensor because of the mirror would also make it closer to the ground glass and thus the AF system this problems like this are essentially non existent providing it doesn't prevent infinity focus.

Iron Flatline · Nov 5, 2007

No, I happen to know this topic well because Rangefinders have no mirrors, are closer from lens to sensor (thus have even smaller tolerances) and are notorious back-focusers. Typically it's a little bit the lens, and a little the camera - so once your kit is set, you send the whole thing in (lenses and cameras) and have the whole shebang calibrated to one another. I just did that, so now all my stuff's tight... provided I don't bang it against the wall too many times.

Yes, when you're shooting at f/11 you have a huge DOF and you'll never notice that the focus is slightly off. But if I use my f/1.0 or f/1.4 lenses on my Leica, even the slightest front- or back-focusing issues are clearly visible. Most SLR users will never know, because their fastest lenses are f/2.8, which is enough DOF to compensate for minor issues.

I think you misunderstood me, I don't think that stopping down can affect the focus. But on a prime lens, the elements HAVE to move when you're focusing. If you're shooting with a 50mm f/1.4 something HAS to move when you're focusing. In case of Leica, they calibrate their lenses to the widest aperture at the shortest focal distance, usually around 0.7 meters. I find that actually to be a poor choice, they used to calibrate them to 1.5 meters, a much more useful portrait distance. At that range you would shoot wide open, and your DOF is razor thin - if you've got focusing issues you'll know right away, and they can't be compensated for easily due to the DOF.

usayit · Nov 5, 2007

I think I understand Garbz's question... I'll restate a little differently.

SLRs determine focus via examination of intensity differences... high contrast.. hense why it is difficult to obtain focus on a flat wall and easier on a high contrast vertical edge. This is what I remember to be passive auto focus. The camera examines the information collected through the lens and sends small adjustments to the internal AF motor of the lens until maximum contrast/intensity difference is established.

So if the camera is determining focus through "relative" increments of focusing out or in and simply informs the lens, then why is it the lens that needs adjustment for back focus?

Unfortunately.. I don't have an answer...

I do see what Iron is coming from though... (I shoot with rangefinders as well). Rangefinders are a bit different... they are "Active" focusing creatures. A rangefinder (duh) with a completely different set of optics is used to determine distance to subject. Then that information is mechanically set on the lens itself. In other words, the rangefinder says 10meters. We then set the lens to 10meters which assume is accurately calibrated. We never examine the image coming through the lens itself for focus accuracy. It is because we never examine the TTL image that a slightly miscalibrated lens in an Active focusing mechanism can easily present focus errors. ie... if the lens focus is actually 10.1 meters when we set it to 10 meters.

I also believe this is why the manufacturing tolarances of a rangefinder has been historically tighter than typcial cameras... This is also why many serious Leica shooters will send a lens+rangefinder body for adjustment together and keep that lens dedicated to that camera body... that particular lens and camera are considered "married"....

So I hope there's some additional info but sorry.. I don't have the final answer.

Anyone remember the polariods with sonic focusing??? Another example of Active focusing... wonderful engineering. Sound determines distance... sounds cool...

btw. Iron.. any back focusing issues with the nocti+M8?

Iron Flatline · Nov 5, 2007

None now that I sent everything to Solms to get married. Whatever focus issues there were prior to that I put down to shooting in low light at f/1.0 and too much wine. But everything is super-aligned right now, and the sharpness out of a perfectly calibrated body and 35mm Sumilux ASPH is breath-taking. The Leica sensor has these really fat pixels that just hoover up detail. It's quite something,...

Garbz · Nov 5, 2007

Yes I too understand how this can happen with a Range finder as the focus isn't done through the lens and thus what you see is not necessarily what you get. But in an SLR the lens doesn't have to be calibrated to a body. I've shot with f/1.4 50mm which also gives you focus issues if you so much as take a breath between focusing and pushing the button, but I still think my point is being missed.

Take for instance the picture below which shows an internal SLR layout. The blue lines is the line from the lens, the red line is the line from the mirror. So now if the distance between mirror and sensor is identical to the distance between the mirror and the ground glass, it doesn't matter how far forward or backwards you move the lens element, or how the focus element inside the lens is positioned. What you see on the ground glass is what is projected onto the sensor. AND what is seen on the ground glass is what the camera has told the lens to focus on.

The only way back focusing can be an issue is if the two groups lines in red are not the same distance from ground glass as from sensor, therefore a camera issue. I don't see how it could be a lens problem which many people blame it on.

JerryPH · Nov 5, 2007

Garbz said:
The only way back focusing can be an issue is if the two groups lines in red are not the same distance from ground glass as from sensor...

... or if the lens firmware incorrectly interprets and sends the wrong information to the camera. Many Sigma lenses that have been sent in and corrected come with a paper referring to a firmware adjustment.

usayit · Nov 6, 2007

but in that case, the lens just chases and the camera never confirms focus.

Helen B · Nov 6, 2007

So why do some Sigma lenses, including my 30 mm f/1.4, need to be modified by Sigma in order to autofocus correctly?

I can't speak for other cameras, but the Nikons I've seen have a different AF arrangement to that shown by Garbz. There are partially silvered areas on the main mirror, allowing some light to pass through to a second mirror that reflects the light down to the AF sensor. Therefore the manual focus and AF have two different mirror stops that can be individually calibrated.

This also means that the light travelling to the AF sensor has to pass through the back of the mirror. This affects the focus position, and the effect varies slightly with the angle the rays make to the main mirror. This is similar to the use of a filter behind the lens.

All very strange.

Best,
Helen