Dynamic-Area vs 3d Tracking Autofocus – Haven’t Found a Good Answer?

elbow770

Active member
Messages
70
Reaction score
2
I am writing about a question that I have seen asked before, but have not yet found a clear answer. Could someone explain to me the tradeoffs between dynamic-area vs 3D tracking? In other words, I think I understand what each of them does, but I am not clear on why I would use one over the other. In particular, if I am using dynamic-area tracking, I have a number selections possible for the number of focus points: 9 points, 21, 51. My first question is why not just always select 51 points? What is the advantage of selecting fewer? Second, why not then always use 3D tracking? What do I get from one that I cannot gain from the other? What are the tradeoffs? I keep hearing that 3D tracking uses color? So what? Do the others not? And if not, does it matter? I also understand 3D theoretically allows me to recompose. But what if I choose not to recompose? Why not just use it anyway? What would I gain by just selecting the 51 point autofocus, and what would I lose? Lastly, why would I ever choose single area focus mode? Why not just choose continuous/AI while my subjects are standing still? I would appreciate if people would help me to identify the concrete tradeoffs.
 
So many questions would take forever to answer them all. You may want to look at back button focusing and always leaving it in C mode. It gives you all the benefits of both S mode and C mode and more manual control over focus. Drawbacks are that it doesn't give focus confirmation beep and some things such as the "red light" AF assist that comes from using an external Nikon flash on Camera (this "red light " AF assist that comes from the flash is much, much more accurate than using the AF-light from the camera).
 
Last edited:
Really, I'm not seeking the answer to many disparate questions. I'm looking for an overall principle, theme, or central logic that can help me make sense of the different options.

Thanks
 
Oh I see. The principle is how much control you give the camera for focus vs. yourself. F.e. 9 point lets the camera "attempt to track" the subject within a small 9 point box. Vs 51 point is a huge box for the camera to "choose from" (i.e. you are probably not going to get the focus you want). Af-A I personally would avoid entirely. 3d is sort of a compromise in a sense. You tell it exactly where to focus (i.e. the subject's face) and the camera tries to stay on that face as it moves about (or as you move the camera about). 3d does work well IMO. Back button focus IMO gives you the most control. you can use it with 9 point, group etc. but you are telling the camera when to focus (i.e. you press the focus button for a specific target). For hard to focus things such as BIF this works amazingly well. I assigned the Fn button on the front to "AF-ON" so it is the "back button focus" button. I don't like using the rear button for BBF because you have to stick your finger in your face and you are introducing some forward tilt to the camera as you shoot vs. Fn button you pull back like on the trigger of a gun.
 
But on 51 point focus, I can choose the initial focus point? Why wouldn't I then get the focus I wanted? Wouldn't the camera begin with my selection point and then track across the frame? How would 9 point focus help? In other words why not always choose 51 focus points and then select the initial one and let the camera track? Otherwise what is the point of having so many focus points?

Thanks again.
 
I am writing about a question that I have seen asked before, but have not yet found a clear answer. Could someone explain to me the tradeoffs between dynamic-area vs 3D tracking? In other words, I think I understand what each of them does, but I am not clear on why I would use one over the other. In particular, if I am using dynamic-area tracking, I have a number selections possible for the number of focus points: 9 points, 21, 51.
With the Dynamic AF-Area modes (9, 21, 51, etc.) The primary focus point you select and it is used to intially lock onto a target, After it locks on (the single primary focus box) the surrounding sensors are used to detect small movement of the subject within the range of the primary AF sensor. That helsp keep lock onto the target if you fail to keep it in the origianl focus box ....temporarily. If you don't move the camera to keep that target in the origianl focus box... focus will shift to another target in the primary focus box. The Dynamic AF-Area modes help you keep lock while panning (helps you stay locked on while your move the camera to keep the target in the original box) but will not actually track the subject in the same way AF-Area mode "3-D" does. The Dynamic AF-Area modes momentarily hold lock so you can catch up and move the camera back to the target.

With AF-Area mode "3-D" you select the target with the focus box you choose. Once the camera locks onto that target the camera uses the remaining focus boxes to keep lock onto that target as it moves across the viewfinder/scene. You don't have to follow the target and keep it in the view of the original box like you do with the Dynamic AF-Area modes. The camera follows it and doesn't care if you move the camera to keep the target in the original focus box. AF-Area mode 3-D also uses the RGB meter sensor to help follow the target.
My first question is why not just always select 51 points? What is the advantage of selecting fewer?
More points give the camera more leeway to work with but also gets confused easier (more sensitive). If the background is less crowded (like a big bird flying across blue sky)... AF-Area mode "51" works great and helps keep lock over a wide area. But as the background behind the target gets crowded (bird flying in front of trees) ...it has a harder time keeping track of target vs background in that large an area. 21pt or 9pt will do better because it has less area to consider. And when the area is really hard, like a bird sitting in the middle of a bush, you may have to switch to "single"
....Second, why not then always use 3D tracking? What do I get from one that I cannot gain from the other? What are the tradeoffs?
Because 3-D tracking is for keeping one target locked as it moves across the screen.... and won't let you lock on to something else. Great for locking on to one player in a soccer match and changing framing as the player moves... but really crappy when you want to change to a different player or target.
...I keep hearing that 3D tracking uses color? So what?
Yes, AF-Area mode "3-D" uses the RGB meter to use color to help keep track as the target moves around the screen/Viewfinder. Helps it be more precise/useful.
Do the others not?
They do not
And if not, does it matter?
Yes because the added use of the RGB meter takes more computing power and can sslow the systems reaction time a little. That slowing can keep the Dynamic modes from being as responsive to change as they need/can to be.
I also understand 3D theoretically allows me to recompose.
Yes...as your recompose AF-Area mode "3-D" will keep lock on the target if it can.
But what if I choose not to recompose? Why not just use it anyway? What would I gain by just selecting the 51 point autofocus, and what would I lose?
"51Pt" assumes you will try to keep the target in the original box. If you don't (in a short amount of time) it changes lock to something else in the original box.
Lastly, why would I ever choose single area focus mode?
When you need more precision and the area around the subject is grounded (in terms of contrasts) and the camera starts locking onto things other than your intended target. Shooting that small bird in a tree...AF-Area mode "9pt" may start locking onto branches near the tint bird while AF-Area mode "single" won't get distracted by stuff around the tiny bird...because it can only see what is in the one focus box and not the 8 other focus boxes around it.
Why not just choose continuous/AI while my subjects are standing still?
I only use AF-C ...because unless the camera is on a unmovable stand and shooting an unmovable subject...subject to camera distance can change. Even if the subject is still...you might move. Only a player when the Depth of Filed (DoF) is critical. Also, I use "back button" focus so being in AF-C is critical.
 
But on 51 point focus, I can choose the initial focus point? Why wouldn't I then get the focus I wanted? Wouldn't the camera begin with my selection point and then track across the frame?
No...it only momentarily keeps lock as the subject moves out of the Filed of View of the initial focus point you used. If you don't move the camera to get the target back in the FoV of the point...the camera will lock on to something else using that initial focus point.
How would 9 point focus help? In other words why not always choose 51 focus points and then select the initial one and let the camera track?
Dynamic modes don't "track"...they help you keep the target in the filed of View of the initial focus box.
Otherwise what is the point of having so many focus points?
51pt is easily confused when the scene is complicated. "9pt" has less area to keep track of so gets less confused. "51pt" works great at an Airshow with planes flying in the sky... and works horrible at a rugby match :)
Thanks again.
--
My opinions are my own and not those of DPR or its administration. They carry no 'special' value (except to me and Lacie of course)
 
Last edited:
Thanks, this is very helpful. But if I use AF-Area mode how then can I tell what the camera has locked onto? What if the camera has locked onto something different than what I intended or thought I was focusing on? How can I know this and make adjustments, accordingly?

Best
 
I am writing about a question that I have seen asked before, but have not yet found a clear answer. Could someone explain to me the tradeoffs between dynamic-area vs 3D tracking? In other words, I think I understand what each of them does, but I am not clear on why I would use one over the other. In particular, if I am using dynamic-area tracking, I have a number selections possible for the number of focus points: 9 points, 21, 51. My first question is why not just always select 51 points? What is the advantage of selecting fewer? Second, why not then always use 3D tracking? What do I get from one that I cannot gain from the other? What are the tradeoffs? I keep hearing that 3D tracking uses color? So what? Do the others not? And if not, does it matter? I also understand 3D theoretically allows me to recompose. But what if I choose not to recompose? Why not just use it anyway? What would I gain by just selecting the 51 point autofocus, and what would I lose? Lastly, why would I ever choose single area focus mode? Why not just choose continuous/AI while my subjects are standing still? I would appreciate if people would help me to identify the concrete tradeoffs.
Hi.

Dynamic modes:

In the dynamic modes, the camera is continuously trying to focus on a subject as long as you try to keep the subject in the chosen focus square by following the subject with the camera. (Panning).

You get some help from the other focus squares - in relation to which of the modes, you have chosen. Many is not always better, I've been told.

If you loose the subject, another square will very shortly take over catching the focus - but you have to have the subject back to the initial square quickly, otherwise the camera will find a new subject in the initial square.

3D Dynamically Tracking:

The camera it self tries to find a moving subject in the chosen square. You are originally not supposed to move the camera (Panning) - but newer cameras have this mode improved (from what i hear) to be able to work with panning too.

When a moving subject is caught (focused), the focus system follows that subject through the frame (focus square shifts) - keeping it in focus in what ever square, the subject might enter. If focus is lost, the camera will (if I recall right) not return to the original square - but will try to find a moving subject in the square, where focus was lost.

The camera gets help from all the focus points, the RGB (metering) sensor and a database to do the focusing.

In my opinion - the 3D can not be effective together with panning. When you pan, everything in the scene - except the moving subject - will by the AF-system be seen as moving - but I have never tried. :-)

Regards.

BirgerH.

PS.: Using the term "focus squares" is an approximation, Actually the squares represent some detector arrays on the focus sensor - which are not sized precisely as the squares.
 
Thanks, this is very helpful. But if I use AF-Area mode how then can I tell what the camera has locked onto?
If you practice with the different AF-Area modes under different conditions, you will learn to "see" it locking onto different targets and such. With just a little practice, it becomes almost second nature when it comes to knowing what the camera is locking on to. The learning curve can be steep by fast. Depends on you.
What if the camera has locked onto something different than what I intended or thought I was focusing on?
Then you correct it. With practice it's a non-issue.
How can I know this and make adjustments, accordingly?
You learn. Only takes a little effort. Really only takes a few hours, a few beers, and a comfortable porch to practice on :)
Good Luck!
 
Thanks, what though specifically or concretely do I "see" in the viewfinder. Do I then see a bold red icon locked onto another image?

Best
 
Thanks, what though specifically or concretely do I "see" in the viewfinder. Do I then see a bold red icon locked onto another image?

Best
You see the focus indicator going from solid to flashing "front" and "back" indicators. You also learn to see the target itself snapping into sharp focus vs something near in front or behind it.
 
Thanks, what though specifically or concretely do I "see" in the viewfinder. Do I then see a bold red icon locked onto another image?

Best
It varies from camera to camera, but generally the selected focus box will illuminate red when focus is acquired and a focus indicator dot in the viewfinder info panel will glow. You can also control whether the camera beeps when focus is acquired, but this rapidly gets annoying.

Nikon AF systems and AF systems in general are very complex owing to the many situations they will be used in. A good part of learning to use the camera properly is in understanding how they work. The manual is concise, but somewhat opaque. I would suggest that you

1. visit the Nikon website and study their self-teaching course on basic AF

2. purchase a book on the subject. Just recently published is "The Nikon Autofocus System" by Mike Hagen, which goes into exhaustive detail with use cases.

My observation with Nikon's PDAF system is that its 3-D Tracking mode is what new users expect it to do, and if they are coming from a mirrorless system especially - because mirrorless systems' big strength is in their ability to precisely acquire and track subject features (like eyes) using the high resolution of the main sensor. DSLRs can't do this (the main imaging sensor is not exposed to light until the moment of exposure, AF is performed with a separate specially designed sensor array and augmented by the exposure sensor in the viewfinder). 3-D Tracking is in my estimation not really that useful a mode for moving subjects until you get to the higher end Nikon bodies like the D500, D750, D810, and D5. In these bodies the exposure sensor is of high enough resolution and the camera GPU high enough performance that it can track well once you acquire focus on something.

Another thing to think about is that exposure and AF are coupled - especially if you are using Matrix Metering. This makes focus and recompose with large movements of the subject in the frame a bit problematic because the camera determines exposure based on what's under the focus zone at the moment of capture. So, if you lock focus on a bright subject in front of a dark background and then move the camera to position the subject elsewhere in the frame - making the original focus zone much dimmer - the camera may overexpose.

Nikon's operating system is intended for you to slew the focus zone (using the joydisk) to where you want the subject to be placed. This keeps AF and exposure coordinated. If you must focus-and-recompose - and in cameras with limited, low resolution AF zone coverage some F&R is inevitable - 3-D tracking is Nikon's answer. Unfortunately, it is still limited to the focus zone coverage you see in the viewfinder.
 
Another thing to think about is that exposure and AF are coupled - especially if you are using Matrix Metering. This makes focus and recompose with large movements of the subject in the frame a bit problematic because the camera determines exposure based on what's under the focus zone at the moment of capture. So, if you lock focus on a bright subject in front of a dark background and then move the camera to position the subject elsewhere in the frame - making the original focus zone much dimmer - the camera may overexpose.
I would also add, don't forget that the AF point bias associated with exposure is generally negated by use of the Dynamic AF-Area modes.
 
Another thing to think about is that exposure and AF are coupled - especially if you are using Matrix Metering. This makes focus and recompose with large movements of the subject in the frame a bit problematic because the camera determines exposure based on what's under the focus zone at the moment of capture. So, if you lock focus on a bright subject in front of a dark background and then move the camera to position the subject elsewhere in the frame - making the original focus zone much dimmer - the camera may overexpose.
I would also add, don't forget that the AF point bias associated with exposure is generally negated by use of the Dynamic AF-Area modes.
 
AF on, AF-C 9 points, menu item a3 to off. Release priority. It works.
 
Thanks to all - lots of good information, clearly presented; thanks. Another resource that I have found helpful.

http://1000wordpics.blogspot.dk/2013/07/nikon-autofocus-guide-d600-d7100-and.html
That article has some good info but it's description of the dynamic AF-Area modes is inaccurate regards initial focus lock. The points surrounding the primary AF point are not used to acquire initial focus lock.

--
My opinions are my own and not those of DPR or its administration. They carry no 'special' value (except to me and Lacie of course)
 
Last edited:

Keyboard shortcuts

Back
Top