Autofocus issues

[Marconikon]

No, I haven’t checked with G series, I’ll definitely look into that though! I’m was thinking it could be a motor/screw issue, but if that is the case wouldn’t the camera be unable to confirm focus? I mean if it is a mechanical problem and the camera can’t put the focus where it thinks it should be, then it shouldn’t show the green dot. Even focusing manually at short distances the dot is green but shots are OOF... Here is the link to one of the tests I did before sending it to Nikon for a check: https://www.dropbox.com/s/31q7nfnirz...81%29.tif?dl=0 all the shots of a flat chart were taken at the same distance (~2.5m) on a levelled tripod and cable release with controlled continuous lighting, no adjustments in post apart from crop, tiff exported directly from raw. Pictures on the left are the reference (AF using LV) and the right pictures are the best of 3 shots taken refocusing each time using phase detection - no AF fine tune.

[dem]

What you describe sounds a lot like a focus shift problem.

https://photographylife.com/what-is-focus-shift

The autofocus is engaged when the lens is wide open, then aperture closes and the focus shifts. A workaround would be to force the camera focus with the aperture closed down (there might be a menu option or a "DOF preview" button on your camera).

[george013]

No, I haven’t checked with G series, I’ll definitely look into that though! I’m was thinking it could be a motor/screw issue, but if that is the case wouldn’t the camera be unable to confirm focus? I mean if it is a mechanical problem and the camera can’t put the focus where it thinks it should be, then it shouldn’t show the green dot. Even focusing manually at short distances the dot is green but shots are OOF... Here is the link to one of the tests I did before sending it to Nikon for a check: https://www.dropbox.com/s/31q7nfnirz...81%29.tif?dl=0 all the shots of a flat chart were taken at the same distance (~2.5m) on a levelled tripod and cable release with controlled continuous lighting, no adjustments in post apart from crop, tiff exported directly from raw. Pictures on the left are the reference (AF using LV) and the right pictures are the best of 3 shots taken refocusing each time using phase detection - no AF fine tune.

Both contrast and phase are using the same mechanical components to correct the focus.

George

[Marconikon]

What you describe sounds a lot like a focus shift problem.

https://photographylife.com/what-is-focus-shift

The autofocus is engaged when the lens is wide open, then aperture closes and the focus shifts. A workaround would be to force the camera focus with the aperture closed down (there might be a menu option or a "DOF preview" button on your camera).

I had never heard about this problem, looks like that is it... unfortunately the camera won't let you focus while holding the preview button, so i guess i'll have to manually focus at close distances. Thanks a lot for sharing!

[Marconikon]

Both contrast and phase are using the same mechanical components to correct the focus.

George

Yes, indeed. So that confirms it isn't a mechanical problem

[Marconikon]

What you describe sounds a lot like a focus shift problem.

https://photographylife.com/what-is-focus-shift

The autofocus is engaged when the lens is wide open, then aperture closes and the focus shifts. A workaround would be to force the camera focus with the aperture closed down (there might be a menu option or a "DOF preview" button on your camera).

I really am very grateful you stopped by, I put the camera in LV, focused with contrast detection at f2.8 and then changed my aperture to f5 and the focus shifted as suspected! At least now I know I'm not crazy or something

[Marconikon]

Is anyone aware of a way to use AF with aperture closed down (except the obvious LV)? The preview button doesn't seem to work (locks focus when pressed) and if i unlock the aperture ring i get the FEE error and can't shoot...

[Manfred M]

I really am very grateful you stopped by, I put the camera in LV, focused with contrast detection at f2.8 and then changed my aperture to f5 and the focus shifted as suspected! At least now I know I'm not crazy or something

Now that result in itself is a bit suspicious.

The "changing focal length" is generally something one sees in very fast (f/1.4 and faster) lenses. For a f/2.8 lens, usually not an issue, just because the wider depth of field will be somewhat self correcting.

It's not something that should only manifest itself in autofocus, but also something that should show up when manually focusing as the issue is inherent in the lens design. I've never owned a lens that is faster than f/1.8, so I have not seen this issue with any of my gear.

[pschlute]

Can I suggest again that until some controlled testing is done we may as well suggest that a fairy living in the camera is causing the op s issues

[Marconikon]

Can I suggest again that until some controlled testing is done we may as well suggest that a fairy living in the camera is causing the op s issues

If you happen to speak to her please tell her she should focus a little closer at shorter distances, I’d appreciate it

[george013]

Now that result in itself is a bit suspicious.

The "changing focal length" is generally something one sees in very fast (f/1.4 and faster) lenses. For a f/2.8 lens, usually not an issue, just because the wider depth of field will be somewhat self correcting.

It's not something that should only manifest itself in autofocus, but also something that should show up when manually focusing as the issue is inherent in the lens design. I've never owned a lens that is faster than f/1.8, so I have not seen this issue with any of my gear.

It's not the focal length that's changing.
Just some thoughts. You can get that green dot in a bigger range when focusing with the aperture wide open. I wonder if there is a difference when starting the focus from minimum or from maximum. Let's say set the lens manual on infinity and focus on something at 2m and take the shot. Then set the lens at the closet distance and do the same.
When using live view the focus system is using contrast. That means contrast in a RGB rasterimage?


George

ps. I found a youtube movie showing Marco's problem.
https://youtu.be/Tt6BXRsslVo

[Stagecoach]

Edit : I believe Manfred is referring to the point of best focus.

Now that result in itself is a bit suspicious.

The "changing focal length" is generally something one sees in very fast (f/1.4 and faster) lenses. For a f/2.8 lens, usually not an issue, just because the wider depth of field will be somewhat self correcting.

I undertook some controlled tests today Manfred, D800, lens 28/75 Tamron at 68mm, 0.85m SD with focus locked and shot through the range from f/2.8 to f/8 in 1 stop intervals. I then cropped and studied the subject area at 100%.

The focus plane that was clearly evident at f2.8 (narrowest) remained sharp through to f/8 and there was nothing to suggest that focus was moving away from the camera. If anything it appears it may have been moving closer to the camera contrary to the article, but determination has to take account of DoF and its fwd/rear % split with aperture change.

The article did mention that different lenses are affected differently.

I can post the results if anyone is interested.

Another test I undertook today was confirming that the AF fine tune on the D800 has a greater adjustment range in the 'Default' input than the 'Individual Lens' input. A useful quirk if you have a lens that you can not correct in the normal way.

[Stagecoach]

I wonder if there is a difference when starting the focus from minimum or from maximum. Let's say set the lens manual on infinity and focus on something at 2m and take the shot. Then set the lens at the closet distance and do the same.

George,

This is the normal recommended practice you should use when checking AF Fine tune or AF accuracy. For each stage in the range you should take one shot de-focused to front and one to rear.

The reason being that any control system has a dead band, if it didn't, focus would never stop hunting.

[george013]

Marco,

I've been thinking further on the focus shift idea. Does your lens have aspherical glass? Searching for the specs it's not mentioned so I assume no.
https://photographylife.com/what-is-...cal-aberration.


George

[george013]

George,

This is the normal recommended practice you should use when checking AF Fine tune or AF accuracy. For each stage in the range you should take one shot de-focused to front and one to rear.

The reason being that any control system has a dead band, if it didn't, focus would never stop hunting.

With dead band you mean a range where in the control system accepts the sharpness?
My idea was if that was so one might change front and back focus in this situation. But I must reject that idea.
Looking at the graphs the image distance is always decreasing so creating front focus, focus is moving away from the camera. Decreasing image distance=increasing subject distance.
As in my other post probably related to the absence of aspheric glass.

George

[Manfred M]

The reason being that any control system has a dead band, if it didn't, focus would never stop hunting.

Exactly!

Given that these are older "D" model lenses (which use the camera's internal focus motor, rather than an in-lens motor) AND he is working at distances close to the minimum focus distance, this thought certainly came to mind. I'm not sure that this is a problem though. I put a couple of my old (80-400mm and 105mm) D lenses on my tripod and tested them.

The focus seems to be work well if I point the lens some distance away and refocus and then return to the point I was focusing on and focus again. Where I get "failure" is when I move my focus point just slightly and the focus mechanism does not change the focus. This is the only time that the "dead band" problem crops up. I only played with the phase detect and did not test with contrast detect.

This of course brings up another consideration; are these test results he is getting due to his test methodology rather than something else?

[odds]

...are these test results he is getting due to his test methodology rather than something else?

I wonder if "something else" could be the AF-system drifting out alignment, and that it even may be a D800(E)-thing. A couple of months ago I noticed that my D800E (2012) did not do AF like it used to. After quite a bit of (time consuming) testing, I concluded there was a back focus issue beyond the AF fine tune capability. All lenses I cared to test apparently got back focus issue. Even the 200-500 was affected, and I felt it kept getting worse. The D800E went to Nikon with a 50mm f/1.4 for AF calibration and a check and clean while it was there.

The Nikon report was terse, "AF adjusted" or similar. It certainly made a difference, my D800E appears to be back to normal.

[Stagecoach]

This of course brings up another consideration; are these test results he is getting due to his test methodology rather than something else?

I have to admit I have some reservations having re-read the thread a number of times.

The initial concern/problem was that the lens/camera was back focusing with reference to portrait results (FL 70mm, 3 to 4m SD, roughly 1/2 body shot framing). This then moved on to the area of 'focus shift' that appeared to also be a problem. AF accuracy repeatability has also been mentioned.

The answers I would want to know if it were my problem lens (and starting at just the long end);

1. At 70mm 3.5m subject distance am I able to adjust the lens fine tune somewhere within its range -20 to +20 to get exact equivalent focus/sharpness on the target as I can with LV/manually?

I'm not sure if Marco can or has achieved this, and this comment in post 4 concerns me , " (at 2.8 it seems as if the back focus is much less, good results at: -10 default +15 lens value) Ignore the -10 def value it does not come into the equation. If we start off with a lens/camera that by inspection of the images suggests back focusing is evident why then would adding an AF fine tune value of +15 give a 'good' result, more back focus has been added.

What I'm thinking is that we should have the lens tuned spot on at f/2.8 (or possibly slightly front focused on the subject) to give us the minimum affect of any 'focus shift' when stopping down to say f/5. If the focus plane is placed at the rear of the target when AF focusing, any possible 'focus shift' will push the focus plane even farther back.

2. Another concern I have is that in the target example linked in post 14 Marco mentions the results on the right "are the best of 3 shots taken refocusing each time". Taking the f2.8 result example used could it be said that this was a 'lucky' one, not a very good testing protocol.

So, has Marco got a lens that has not been 'tuned' fully or can't be and is this exaggerating other situations, e.g. close focusing and focus shift?

[george013]

Exactly!

Given that these are older "D" model lenses (which use the camera's internal focus motor, rather than an in-lens motor) AND he is working at distances close to the minimum focus distance, this thought certainly came to mind. I'm not sure that this is a problem though. I put a couple of my old (80-400mm and 105mm) D lenses on my tripod and tested them.

The focus seems to be work well if I point the lens some distance away and refocus and then return to the point I was focusing on and focus again. Where I get "failure" is when I move my focus point just slightly and the focus mechanism does not change the focus. This is the only time that the "dead band" problem crops up. I only played with the phase detect and did not test with contrast detect.

This of course brings up another consideration; are these test results he is getting due to his test methodology rather than something else?

A D-lens is a lens with a cpu in it that gives information about the Distance
A AF-lens makes use of a camera focussing motor.
A AF-S lens has an internal focussing motor.

George

[george013]

I have to admit I have some reservations having re-read the thread a number of times.

The initial concern/problem was that the lens/camera was back focusing with reference to portrait results (FL 70mm, 3 to 4m SD, roughly 1/2 body shot framing). This then moved on to the area of 'focus shift' that appeared to also be a problem. AF accuracy repeatability has also been mentioned.

The answers I would want to know if it were my problem lens (and starting at just the long end);

1. At 70mm 3.5m subject distance am I able to adjust the lens fine tune somewhere within its range -20 to +20 to get exact equivalent focus/sharpness on the target as I can with LV/manually?

I'm not sure if Marco can or has achieved this, and this comment in post 4 concerns me , " (at 2.8 it seems as if the back focus is much less, good results at: -10 default +15 lens value) Ignore the -10 def value it does not come into the equation. If we start off with a lens/camera that by inspection of the images suggests back focusing is evident why then would adding an AF fine tune value of +15 give a 'good' result, more back focus has been added.

What I'm thinking is that we should have the lens tuned spot on at f/2.8 (or possibly slightly front focused on the subject) to give us the minimum affect of any 'focus shift' when stopping down to say f/5. If the focus plane is placed at the rear of the target when AF focusing, any possible 'focus shift' will push the focus plane even farther back.

2. Another concern I have is that in the target example linked in post 14 Marco mentions the results on the right "are the best of 3 shots taken refocusing each time". Taking the f2.8 result example used could it be said that this was a 'lucky' one, not a very good testing protocol.

So, has Marco got a lens that has not been 'tuned' fully or can't be and is this exaggerating other situations, e.g. close focusing and focus shift?

If the best out of 3 is shown, that means the result might be even worse.
If you want to correct AF for the close distance and wide aperture, the focus shift, then you might introduce problems with longer distances and/or smaller aperture.



I tried to find the specs of that lens.
From this site http://www.photosynthesis.co.nz/nikon/serialno.html I can see only 2 35-70 2.8D lenses.
A description of that lens https://www.mir.com.my/rb/photograph...0mm/index1.htm. I'm not sure if Marco has this lens for this lens has a macro switch for close distances. He didn't mention that.
That specific lens doesn't have aspherical glass so when one wants to compare it with other lenses do it minimal with a lens without aspherical glass.

George