Nikon AF help/info

[NorthernFocus]

In spite of all of the money I have invested in expensive Nikon gear, I still shoot exclusively with a single focus point and continuous focus mode. In Nikon terms I shoot AF-C mode and S point selection. I do so because historically, though it has its downsides, in my experience it is the only truly reliable mode of knowing what the camera will focus on in a given situation. But it seems a waste not to use the technology we're paying for and my first inclination is always that the equipment is fine and most problems are caused by operator error. So every few months I go through the exercise of trying the various "dynamic" area modes in the Nikon AF system. Most recently I did so when shooting BIF as posted in this thread

Well my recent results were pretty much the same as with previous tests. When shooting in dynamic 9 or 21 point AF, I seem to get more soft shots. Not really OOF but just not right. What's even more confounding is trying to figure out why the camera chose the point that it chose on a given shot as indicated by the focus point displayed when reviewing the RAW files in ViewNX2. And most confounding of all is that the focus point indicated by ViewNX2 will sometimes be in clear blue sky yet the bird will be in sharp focus.

To-date, I've never found anything other than general explanations of which focus modes work best in different situations. Never any real description/explanation as to how/why. So if anyone can shed any light on how the Nikon AF works, has any links to details/substantive explanations, etc, please share.

[Shadowman]

I agree with your assessment, the 3-D tracking mode can hit multiple focal points and still have nothing in focus, or else the camera focuses on the closest object and this can be troublesome with moving subjects.

[pnodrog]

I suspect in the end you will wish you had never asked......

................ I'd have preferred to remain blissfully ignorant....

This thread will be interesting. I tend to take photographs of static scenes so the multiplicity of AF focus points seems to offer very little advantage.

[Manfred M]

Dan you seem to use AF very much the same way I do; AF-C mode and single point selection. I've switched to back-button focusing as well, so that my focus point does not accidently move when I recomposed my shot My testing gives the same results as you have, especially when using long lenses at large apertures.

I use the same approach when I do portraits and street photography; focus on the eye and recompose and then press the shutter release. This method is not 100% fool proof if the subject is close to me and moving towards me.

I've never quite figured out the matrix metering modes; as we know there is a single focus plane, so as far as I can tell, the matrix modes analyse the subject and then select a compromise focus plane, where all of the active focus points are somehow used to determine this hypothetical focus plane.

I would like a better explanation, but the manufacturer's manual seem to be written by the marketing team, and information found on the internet tend to be from people who do not seem to be particularly strong technically (even though they think they are).

[lukaswerth]

A brief general observation:

today's DSLRs come with many, many bells and whistles. Some of these are often very useful, but they are also part of the company's marketing strategies. They produce their own secondary industries, books written, internet "gurus". But an intimate technical knowledge does not make a person a better photographer, as photography is about making pictures, seeing, composing, finding one's own style. Technology is a means to an end. So if you get what you want with one way of auto focus, there is no need of trying the others.

(Even though discussions about different autofocus modes and such like are always entertaining.)

Lukas

[NorthernFocus]

Thanks for the comments thus far, folks. Nikon is frustratingly tight with technical information. I suppose they view it a maintaining a competitive advantage over others. As Lukas points out that mindset does benefit a cottage industry of folks like Hogan, Rockwell, etc. whom we all look to in an attempt to fill in the blanks.

...Technology is a means to an end. So if you get what you want with one way of auto focus, there is no need of trying the others.

Thanks, Lukas. Much of what I shoot could be generically termed "action photography". What I want is 100 percent perfectly focused shots which I don't currently achieve

[Manfred M]

Thanks for the comments thus far, folks. Nikon is frustratingly tight with technical information. I suppose they view it a maintaining a competitive advantage over others. As Lukas points out that mindset does benefit a cottage industry of folks like Hogan, Rockwell, etc. whom we all look to in an attempt to fill in the blanks.

You forgot to add that these authors are taking educated guesses at how these things work and are wrong just about as often as they are right.

Rockwell, who likes to present his conjectures as "facts", is often "right" about what he writes. Unfortunately he forgets to include his assumptions and other parameters that could impact what he is saying, so people assume these are "universal truths", as opposed to ones that are correct only in very specific circumstances (which he forgot to mention).

Hogan is also quite knowledgable, but where he falls down is in his lack of a formal technical background. He's been around photography long enough that his guesses, based on his experience, are often quite good. Unfortunately, he is so sure of himself, that when he does guess wrong he will go off on a completely incorrect tangent, not realizing what he is doing.

Thanks, Lukas. Much of what I shoot could be generically termed "action photography". What I want is 100 percent perfectly focused shots which I don't currently achieve

While I am not an "action" photographer, I find the same thing in my shooting. That being said, my success rate is much higher than it ever was using manual focus equipment.

[NorthernFocus]

I agree with what you said in your last post, Manfred. The key phrase in my prior comments is that we(collectively) read those folks in an attempt to fill in the blanks. Unfortunately the "experts" suffer the same human failing often seen in forum posts i.e. an inability/unwillingness to see things from a perspective other than their own. IMO that is KR's primary failing.

Also as you stated, the "experts" are largely guessing(albeit educated guessing) at how functions are technically achieved. Unfortunately they typically express their conclusions/opinions implicitly as facts. IMO this is Hogan's primary failing. While he seems to be more open minded than KR, he still suffers from a significant dose of ego.

[Manfred M]

I think we read these authors in the same way. Both suffer from the same issue; overinflated egos.

I wonder if this is part of their marketing strategy; after all they are making money with their "sage" advice and the noise attracts people to visit their websites.

[PhotomanJohn]

I'm a little late in getting into this discussion but had the same initial experiences and quickly went to only using single point, AF-C, back button focus for all my shots. Dan is correct in that we really should better understand the technology we have in our hands. So being one who thrives on the technical side of photography, I am going to put some serious effort into understanding the modes that I don't use. Someone put a significant amount of effort into creating and refining these features. One would think they would be good for something.

Dan - Thanks for the question and I will let you know if I make any break-throughs.

John

[Shadowman]

Thanks for the comments thus far, folks. Nikon is frustratingly tight with technical information. I suppose they view it a maintaining a competitive advantage over others. As Lukas points out that mindset does benefit a cottage industry of folks like Hogan, Rockwell, etc. whom we all look to in an attempt to fill in the blanks.


Thanks, Lukas. Much of what I shoot could be generically termed "action photography". What I want is 100 percent perfectly focused shots which I don't currently achieve

I picked up one of the Magic Lantern Genie Guides for one of my Nikons and the book spent a massive six pages on the auto-focus modes, however the guides do provide you with a few good tips not shown in the manuals and would probably overlooked with skimming through the menus.

[PhotomanJohn]

OK, the following is a summary of my initial testing of how the main AF modes work on a D800 which I believe is the same as the D7000. I also believe that there were some changes to how the autofocus algorithms work about the time the D7000 came out which probably contributed to the complaints with the D7000 when if first came out. So if you have a D300, D700 or the like, your results may vary from mine. I didn't fool with the 3-D and auto modes.

AF-S or AF-C, Single Point - Initial Focus Capture

The camera focuses on the feature at the selected focus point. If there is no feature at the selected point (blank sky or the like) it will focus on a feature, if there is one, at one of the focus points immediately surrounding the selected one. If there are no features close to the selected point it will not focus.

AF-C, Single Point – Tracking Subject or Camera Movement

The camera refocuses on whatever feature is within the selected focus point unless the point becomes over a featureless area then it holds the focus until a feature returns to within the selected focus point (not just close to the selected focus point as in initial capture) and then refocuses.


AF-C, Dynamic – Capture and Tracking

Initial focus capture is the same as described above except that when there is no feature at the selected focus point, it will focus on features within a couple of rows of focus points (not just one as noted above) around the selected point. The focus then remains on the initial subject as the subject or the camera moves as long as it stays within the selected focus point area. The exception is that if the selected focus point moves onto a subject that is closer than the original focus plane, the focus with snap to that closer subject even if the original subject is still within the selected focus point area.

If the original subject moves outside of the selected focus point area, then the focus will snap to a subject as it does during the initial capture. If there are no features around the selected focus point, then the camera will hold the focus until there is something to focus on.

These are my observations and probably need some refinement but it should be a starting point and shows that there is more to how the AF works than many of us understood.

John

[NorthernFocus]

That's a good summary, John. There are additional nuances to how/when the system decides to switch to one of the other focus points in a selected group. Therein lies the real rub. I've done a bit of testing also with small targets and different distant backgrounds. In my tests the system responds differently with different BGs. That is the main reason I abandoned multi-point was due to the unpredictability. I wish I had kept my test images to show the setups I used. And it would take me a bit of text to try to describe verbally which I don't have time to do at the moment.

One other thing that you did not mention is the focus tracking delay setting. Depending on how you have this set, there is a delay in the system from the time it loses target completely until it decides to refocus on the originally selected focus point. One catch to this is that the delay works even if a single focus point is selected. I learned this the hard way. If you focus on a spot anticipating a fast moving object passing between that spot and the camera, when the object arrives the camera will not refocus on it until the delay expires. I figured that one out by blowing several hundred shots at a dog agility trial before I noticed what was happening

So for those who do use single focus point, beware if you haven't already figured that one out. There are times like shooting BIF when the delay is desirable. There are other times when it is decidedly not so...

[PhotomanJohn]

Dan - I'm sure the focus algorithms are more complicated than my simple tests revealed. These cameras are very complicated which caused me a more painful transition from medium and large format manual cameras to modern digitals than I expected. Being complicated is one thing, but when camera automation is not defined as is the case we are talking about along with scene modes on the consumer cameras and the like, IMO this makes these function nearly unusable to the serious photographer. You can't depend on a function that you don't know how it works. This is a mistake on the manufacturer's part especially in the case of pro cameras.

Yes that delay business is an issue. My suggestion, especially to individuals using back button focus and focus and recompose, is to disable the tracking delay as you mentioned (set to Off) along with AF-C Priority and AF-S Priority (set both to Release) so the camera will always take the shot when you push the button whether it thinks the image is in focus or not. These three options are a1, a2 and a3 on the D800.

John

[chauncey]

I have shot a fair number of BIF series using my Canon gear...it is not that unusual.

There are a fair number of "White Paper" instructional files from the Pro's on how they "set-up" their
Custom Functions for different shooting scenarios. Are they limited to the Canon line?

[NorthernFocus]

..There are a fair number of "White Paper" instructional files from the Pro's on how they "set-up" their
Custom Functions for different shooting scenarios. Are they limited to the Canon line?

"for different shooting scenarios" is the operative part of that statement. Every time I've quizzed the author of such papers the answers result in multiple qualifying assumptions. That is kind of the point of this thread. I'd like to understand the technology rather than follow others instructions by rote or come up with my own dozen or so different sets of settings by trial and error. Understanding why the camera behaves as it does allows one to foresee problems and adapt to different conditions.

Then again, if the pros would honestly share information such as their hit ratio when shooting BIF, I might save a lot of time and worry. It is possible that I'm tilting at windmills

[george013]

I don't know if you have seen this site but it seems to me that it answers some questions.
http://www.nikon.com/about/technolog.../caf/index.htm
George

[NorthernFocus]

Thanks for the link, George. But as with all other Nikon literature I've seen, the text lacks the detail that we are talking about.

...simple predictive tracking AF does not provide maximum focusing performance for a subject that abruptly changes direction moving at high speed or a subject with low contrast moving randomly. To focus accurately on such a subject, the AF system must accumulate data of the locus of the subject’s movement using multiple focus areas so that appropriate judgment can be made.

Which multiple focus areas? Are focus points outside the selected focus group used? What data? Is it based solely on phase detection or does color play a part?(as mentioned in other Nikon literature)

...This system boasts an advanced AF function. For example, in one setting it selects the most appropriate AF area by taking the photographer’s intention into consideration.

And how does it determine "the photographer's intention"?

There is also a mode which combines adjacent focus areas into groups and automatically selects the focus area from the focus area group in which the subject is positioned...

And how does the AI determine what is the intended subject?

This lack of available detailed explanation is the purpose for this thread. To-date I've not find more detailed descriptions of what information the AF uses to track moving objects. I've seen multiple references that color is used in addition to phase/contrast. Nikon literature suggests that color is only used in 3D tracking mode. However some testing I've done myself suggests to me that color plays a part in dynamic area mode as well. That's the sort of information that would be nice to have. Then again, if I could confirm that color is a factor, all that would do would be to support the method that I currently employ with Nikon AF. But at least I'd know and could stop doubting myself

[george013]

Well, I only use AF-s for my shots. But thinking loud, with 1 focuspoint and AF-c continuous pointed on the subject the camera can only detect a 2-d movement, coming to you or going from you. If you want to follow a 3d-movement you will have to use more focuspoints. There you have several options. For the difference between them you will have to look in the manual.

I thought you wanted to know how things are working, not how to use them. That's all in the manual.

I just tried with my camera. Select the setting like 3d, af-c, then select a focuspoint, point it to your subject and start the proces.

Be aware that in basic settings with AF-c the shutter release works also when the subject is out off focus.

I don't know what gear you have. I thought I read somewhere a D800 but I can't find it back. And your photos are without exif.

George

[NorthernFocus]

...with 1 focuspoint and AF-c continuous pointed on the subject the camera can only detect a 2-d movement, coming to you or going from you. If you want to follow a 3d-movement you will have to use more focuspoints...

In the context of focus, only motion towards/away from the camera is relevant. Lateral motion only alters focal plane in as much as it isn't purely lateral (i.e.on a tangent to the radius around the camera's position). The multiple points are all detecting position of focal plane. The various available groupings simply accommodate the photographer's inability to maintain a single selected point on the target/subject. The primary question at hand is how the algorithms function which determine which point to use for obtaining focus. Even when shooting static subjects in multipoint mode, the camera will often decide to switch from the primary point to another point in the group. This can ruin a shot when critical focus is needed with long lenses and wider apertures.

...I thought you wanted to know how things are working, not how to use them...

Just so.

FYI I currently shoot D4, D810, and D7100. The only notable differences in focus functions that I've noted are speed and ability to handle low light.