Why is AF Fine Tune necessary?

[peter_nelson1]

The D300 has an AF fine tune feature where it can remember the correction needed

to make AF work properly on up to 12 lenses. I've already had to use this on several

lenses.<br><br>

 

But my question is <b>WHY?</b> What is it about their AF system, or the lenses,

that should require DIFFERENT adjustments for different lenses? The film plane is

the film plane, and when something is in focus it's in focus, and the AF system

adjusts the focus until it is. So what it there to correct?<br><br>

I don't know how the Nikon AF system works, but even if it uses separate sensors,

i.e., not ones co-located at the film plane, and these sensors were NOT at the exact,

precise optical distance as the film plane, they would still be off by some fixed

amount, so why would different lenses need different corrections?<br><br>

 

Can anyone with some technical understanding of the AF system explain this?

<br><br>

 

 

(also, while we're on the topic, how does the AF-fine tune system identify the lens so

when I attach it, it knows what stored correction to us, and how do I delete stored

corrections for lenses I no longer use?)<br><br>

 

Thanks in advance!!!

[pete_s.]

http://www.photo.net/bboard/q-and-a-fetch-msg?msg_id=00Nd7l

[ellis_vener_photography]

There has been and always will be mechanical differences in fit between various lenses and various bodies. The film industry has known this and done this kind of fine tune mlens to body manually for decades and high end still photographers have done it as well with their lenses as well. Most people just didn't have either the knowledge of the necessity for it or the facilities to do it and now Nikon has built it in. By the way the Canon EOS 1D Mark 3 has this feature in a limited way as well.

 

"how does the AF-fine tune system identify the lens so when I attach it, it knows what stored correction to us, and how do I delete stored corrections for lenses I no longer use?"

 

Nikon AF lenses have a microchip in them that connects to the camera's computers relaying all sorts of data to the camera. Nikon is just exploiting this capability.

[ilkka_nissila]

The AF sensors do not measure the light that falls on the film/sensor plane. They work based on light which is diverted through mirrors from the light which would otherwise go to the viewfinder. Depending on the optical design of the lens, the AF sensor may or may not work accurately.

 

The only lens which I have for which I find the autofocus sensor doesn't work well is the 25mm Zeiss, so I may try the focus fine tune feature on that lens if I need to use the focus dot. I just haven't bothered with it yet - all my other lenses seem to focus fine on the D3.

 

I would just adjust the system individually for those lenses which seem to require it - when you have real-world problems, not otherwise. Notice that this correction might be dependent on focused distance and zoom setting. It might even depend on the aperture.

 

Some people have issues with the 105mm DC and digital SLR autofocus systems. I have issues with the focusing accuracy of the D200 and 105mm DC combination but not really with the D3. I would just assume the source of these problems is that many of the optical designs haven't been made to current (DX sensor pixel-peeper) requirements in AF precision.

[radfordneal]

There seems to be considerable confusion about this!<P>

 

I'm not an expert, but I had thought the reason was that most lenses have some degree of spherical aberration, which means that the "most in focus" point varies with the f-stop. The AF system effectively uses something like f/5.6, and hence may not be doing the optimal thing if you're taking the picture at f/1.4. Since

focus is more crucial at wide f-stops, you might want to adjust so that the AF system is optimal for the wider stops rather than f/5.6. Since the amount of

spherical aberration will vary from lens to lens, this would have to be done

for each lens separately.<P>

 

Can anyone out there <I>who actually knows</I> comment on whether my understanding is correct?

[peter_nelson1]

"There has been and always will be mechanical differences in fit between various lenses and various bodies."

 

Obviously, but what does that have to do with it? So the motor has to turn the helical an extra few degrees for this lens -vs- that lens - it still knows when it's achieved focus by its AF sensors.

 

"The AF sensors do not measure the light that falls on the film/sensor plane. They work based on light which is diverted through mirrors from the light which would otherwise go to the viewfinder. Depending on the optical design of the lens, the AF sensor may or may not work accurately. "

 

But that doesn't answer the question either - as I said in my OP, if the AF sensors are out of alignment with the film plane that would produce a _uniform_ error, not one that would vary with different lenses.

 

"http://www.photo.net/bboard/q-and-a-fetch-msg?msg_id=00Nd7l"

 

I looked at this and the consensus view seemed to be that if a lens had a tilted element it might not hit AF sensors on opposite sides evenly. I don't think this is the answer, either, because . . .

 

1. I almost always use just a single AF point in the middle to focus with but I still find I need to AF fine-tune some lenses.

 

2. If a lens has a tilted element bad enough to throw off the AF sensor to a dgree that required fine-tuning this would also produce visible focus errors in the image. One incoming qual test I always do with a new lens is to shoot newspapers on a wall from a tripod aligned square to the wall to test resolution and uniformity of focus. (in fact last night I rejected a Tokina 11-16 because it failed this test, and it's on its way back to the dealer as I write this!) But my Nikkor 24-70 passed that test with flying colors, but it still needed AF fine-tune.

 

"Nikon AF lenses have a microchip in them that connects to the camera's computers relaying all sorts of data to the camera. Nikon is just exploiting this capability"

 

So AF fine tune doesn't work with non-Nikkors? (having rejected the above Tokina, I'm all-Nikkor today)

[peter_nelson1]

<i>The AF system effectively uses something like f/5.6</i><br><br>

 

Please explain what this could possibly mean. My understanding is that all AF systems do their AF wide-open to maximize light and to get the narrowest DOF.

[dmitriyk]

<i>The AF system effectively uses something like f/5.6, and hence may not be doing the optimal thing if you're taking the picture at f/1.4. Since focus is more crucial at wide f-stops, you might want to adjust so that the AF system is optimal for the wider stops rather than f/5.6.</i><br>

<br>

Say what?

[michael_darnton1]

My understanding, which may not be authoritative, is that the autofocus system works like the split-image screen in a manual-focus camera.

 

The split-image effectively changes the view at the center of the screen via prisms pointed to either side of center of the back of the lens' aperture. Focusing uses triangulation similar to how the two windows of RF cameras determine a distance.

 

This has to happen at some aperture less than full, because not all lenses are the same speed/size-of-hole, and the system has to focus all of them. So the manufacturer picks a width--say f5.6, and all lenses focus with that effective aperture, not wide open--glass outside the borders of where the RF system is looking is ignored. When a smaller aperture is used, half of the split-image blacks out because you can slide your eye left or right to include one or the other side of the slightly-blocked aperture hole, but not both at once. That's why the effective focusing aperture is less than that of the lens. To make up for the resulting zone-of-no-feedback in the center of the range on larger apertures we compensate for by racking focus back and forth to center between equally out-of-focus areas and hopefully find the focus point within the dead zone.

 

People who've done a lot with split-image focus vs groundglass, doing critical work, know that because the center of the lens isn't being used for the distance calculation, aberrations in the lens can throw the focus off a bit, and all lenses are different in this regard--for critical work it's better to use the groundglass ring. Since each lens is slightly different, this is where the need to adjust the system so that it works properly.

 

As already pointed out, calibration of the flange distance has nothing at all to do with this, since on both manual and autofocus cameras the focus is determined based on perceived sharpness, not flange distance, which is completely irrelevant to the process.

 

In the D300, this system isn't used for live view focusing in tripod mode--that's done on contrast which is similar to the way we focus on groundglass, not as when using split-image or normal autofocus.

[michael_darnton1]

Whoops--I cut a sentence:

"... this is where the need to adjust the system so that it works properly" is an advantage in the D300 additional to the possibilities we had before, not a "fix" for a defect in the camera, I suspect.

 

This may not matter at all to most people because, frankly, I don't see a whole lot of well-focused pictures out there. . . most people don't really care, if you look at their photos, not their words. I think, then, that the inclusion of this on the D300 mostly stirred up a hornet's nest, making people think there's a new problem rather than an old one finally being dealt with.

[peter_nelson1]

"This has to happen at some aperture less than full, because not all lenses are the same speed/size-of-hole, and the system has to focus all of them. So the manufacturer picks a width--say f5.6, and all lenses focus with that effective aperture, not wide open--glass outside the borders of where the RF system is looking is ignored. When a smaller aperture is used, half of the split-image blacks out because you can slide your eye left or right to include one or the other side of the slightly-blocked aperture hole, but not both at once. That's why the effective focusing aperture is less than that of the lens. To make up for the resulting zone-of-no-feedback in the center of the range on larger apertures we compensate for by racking focus back and forth to center between equally out-of-focus areas and hopefully find the focus point within the dead zone. "

 

I don't see how this would work with a modern multipont focussing system where the user can arbitrarily select his specific focus point. As I said, I use single-point, and over the weekend I was shooting an oriole in an apple tree with my 300 f/4 where there were lots of branches at various distances all around and in front of and back of the bird. I had no trouble putting my focus point exactly on the bird and focussing just on him.

 

Can you please point us to a source of information on your thesis as it applies to a modern DSLR? Thanks in advance.

[john_beaver1]

I addressed this on a previous thread.

I guess the real question is are your pictures out of focus ?????

 

Nikon (d300) uses phase-detection auto-focus on a special focusing-sensor.

In live mode on a tripod it uses contrast detection adjusting the focus based on the

actual image on the sensor.

 

Just to be sure, I put a bunch of lenses on an Arriflex motion picture focusing

chart, and regardless of where the fine-tuning was set, it always focused properly.

On one zoom there was some discrepancy (when I zoomed out the focus shifted

slightly) but the camera corrected for it as long as I focused at the correct mm.

With fine-tuning I was able to correct this back-focus error so now it stays sharp if I

change the focal length without re-focusing.

 

If the elements of a lens are not exactly on plane with the sensor, then an

adjustment could help correct some of the error.

 

On a nikon camera, the lens is always wide-open until the shutter is released, even

in full manual mode. So it is always focusing at it's widest aperture. That's why it

has a depth of field preview button.

[blumesan]

Lots of guessing going on here. And, as Peter has pointed out, most of the guesses don't withstand the test of simple logic. Has anyone got a link to the horse's (Nikon's) mouth?

[michael_darnton1]

For a start, read the definition of phase detection here:

http://en.wikipedia.org/wiki/Autofocus

which is basically the definition I gave.

[ilkka_nissila]

I wasn't talking about alignment of the optical components but simply that the camera sensor and the AF sensor sample different subsets of the light transmitted by the lens. This can lead to discrepancies. Obviously if you use live view autofocus then it's based on the sensor data - but not very practical IMO.

[michael_darnton1]

Lacking a Nikon statement, which I don't think we'll get, here's another analysis supporting my position:

http://www.markerink.org/WJM/HTML/af_expla.htm

Don't knock it if you don't understand it. :-)

[dmitriyk]

Right, but where are you pulling the business about f/5.6 out of?

[michael_darnton1]

The last big paragraph in the link above-- http://www.markerink.org/WJM/HTML/af_expla.htm -- deals with that. The number 5.6 was suggested by someone else, and I went with it. The salient point, though is that the functional f opening must be at least (small) high enough to allow *every* lens that will be focused to function, and given that many commonly-used zooms shift to 5.6 at their long ends, that seems to dictate 5.6 if not something even smaller. Consequently, no matter how fast of a lens you hang on your camera, the autofocus isn't going to be focusing at anything close to the max aperture--it will simply ignore the extra glass hanging out past the f5.6 limit that's predetermined by the system.

 

This is similar to the reason point-source enlargers work so well: it doesn't matter how fast of a lens you're using: the effective aperture is the size of the light cone at the aperture plane, which, ideally, in a perfect system, is less than the full opening. With that situation, you use the sharper central portion of the lens (effectively, stopping down for sharpness) without a bit of diffraction loss, since there's no aperture for the light to pass through and diffract and degrade.

 

I sense that some of the critics of these concepts are protesting not because they understand the concepts and disagree, but because they don't understand, so maybe some deeper thinking is required, what?

[raffal]

d300 manual, page 327 : "AF tuning is NOT recommended in most situations: use only when required."

page 328 : "The camera maybe UNABLE to focus at minimum range or at infinity when AF tuning is applied".

Case closed ! :) raf

[amitava_bhowmik]

I swear.....this feature on the D 300 is really a miracle. Something i always wanted for my faithful D 70. I tried the firmware upgrade provided by Nikon later to correct backfocus. It was better,..but not as good as i had hoped it to be. Still pics are not tack sharp. Why does not Nikon release this "auto focus fine tuning" feature for various lenses as a firmware upgrade? We would all be hugely benefitted. Are you listening Nikon????? Possible????????????????

[matt_smith24]

<p>d300 manual, page 327 : "AF tuning is NOT recommended in most situations: use only when required." page 328 : "The camera maybe UNABLE to focus at minimum range or at infinity when AF tuning is applied". Case closed ! :) raf<br>

No, not case closed. The word you should've put in caps is "MAYBE" and "maybe" does not mean "will be". Obviously it is only recommended if it is needed. It's not a feature to play around with if you don't know what you're doing.<br>

My D3 camera body front focuses and I have the default AF Fine Tune set to "+15" to compensate. It really needs to be sent to Nikon to be recalibrated but meanwhile the AF Fine Tune works great and all my lenses now focus correctly. It would do exactly the same if I had the AF Fine Tune default set to "0" and each of my Nikon lenses set "+15". All my lenses still focus at minimum range or at infinity.</p>

<p> </p>

[matt_smith24]

<p>d300 manual, page 327 : "AF tuning is NOT recommended in most situations: use only when required." page 328 : "The camera maybe UNABLE to focus at minimum range or at infinity when AF tuning is applied". Case closed ! :) raf<br>

No, not case closed. The word you should've put in caps is "MAYBE" and "maybe" does not mean "will be". Obviously it is only recommended if it is needed. It's not a feature to play around with if you don't know what you're doing.<br>

My D3 camera body front focuses and I have the default AF Fine Tune set to "+15" to compensate. It really needs to be sent to Nikon to be recalibrated but meanwhile the AF Fine Tune works great and all my lenses now focus correctly. It would do exactly the same if I had the AF Fine Tune default set to "0" and each of my Nikon lenses set "+15". All my lenses still focus at minimum range or at infinity.</p>