Author Topic: Nikon 14-24mm focussing difference at edge of frame  (Read 4630 times)

Gromit44

  • Jr. Member
  • **
  • Posts: 23
    • View Profile
Following on from this thread http://forums.hayobaan.nl/index.php/topic,20.0.html re the Nikon 14-24mm, I noticed that if I focus on the centre of the frame at 14mm, the centre is always sharp but the edges can be soft/distorted. If however I focus on the egde of frame, the centre is soft but the edge is sharp.

Here's an example showing the right edge of two different shots of the same subject. One is focussed at the centre, the other is focussed at the edge using Live View and the AF-ON button. The red square indicates the Live View focus point. (Nikon D800E, distortion control off, 14mm at f/9, mirror lock up, wireless remote).

« Last Edit: 23/12/2012, 18:04:05 by Gromit44 »

Hayo Baan

  • Professional Photographer and Workshop Leader
  • Administrator
  • Sr. Member
  • *****
  • Posts: 210
  • Professional Photographer and Workshop Leader
    • View Profile
    • Hayo Baan – Photography
Re: Nikon 14-24mm focussing difference at edge of frame
« Reply #1 on: 24/12/2012, 12:52:26 »
Hello Simon,

One thing your samples show (again) is that, when properly focussed, your lens is pretty darn sharp, even at the edges. So the lens is most likely OK.

So why the big loss in sharpness from side to centre? This, while Depth of Field of the lens at 14mm at f/9 indicates this should not be a problem.

A couple of observations that may explain this:
  • For the centre shot you focussed on the the building that is farthest from the camera, but the sides are much closer (any idea what the distances to the camera where?). For your side shot, the reverse is true. From a depth of field perspective, this is far from optimal and you'd best focus somewhere at a third in between.
  • DOF calculations are based on old rules of thumb with an old definition of what is acceptably sharp. With modern cameras, and certainly with the super high pixel count cameras like the D800E, what is acceptable sharpness at pixel level requires a much more stringent definition of acceptable sharpness. The DOF as a result will be much, much, shallower.

Further info regarding the second point: normal calculations for Depth of Field take 0.025mm or even 0.03mm for the “circle of confusion,” with a pixel size of only 4.88µm, this means 5 to 6 pixels (in each direction!), similar to viewing at 20%…

Try the same shot again, but now with focussing somewhere between the closest and farthest points of what you want to have in focus (you may also try going to f/11 though this already may introduce some diffraction effects).

Also, to rule out any environmental or vibration related issues, try the shot at daytime with a short shutter speed.

Hope this helps,
Hayo
Hayo Baan – Photography
www.hayobaan.nl

Gromit44

  • Jr. Member
  • **
  • Posts: 23
    • View Profile
Re: Nikon 14-24mm focussing difference at edge of frame
« Reply #2 on: 24/12/2012, 17:15:53 »
Hello Simon,

One thing your samples show (again) is that, when properly focussed, your lens is pretty darn sharp, even at the edges. So the lens is most likely OK.

I'd agree with that.

So why the big loss in sharpness from side to centre? This, while Depth of Field of the lens at 14mm at f/9 indicates this should not be a problem.

A couple of observations that may explain this:
  • For the centre shot you focussed on the the building that is farthest from the camera, but the sides are much closer (any idea what the distances to the camera where?). For your side shot, the reverse is true. From a depth of field perspective, this is far from optimal and you'd best focus somewhere at a third in between.

Here is an aerial shot of the building - and a layout plan:


The room marked 'H' on the plan (the one nearest 'G') is 180 feet long. The camera was in the centre of the courtyard where the two (pale coloured) walkways meet. From this I estimate the building was 210 feet from the camera at the centre of frame and 180 feet away at the edges. The part of the building at the edges (i.e. where the Christmas trees are) was well forward of the sensor plane though - so even allowing for the CoC calculation below, I'm still surprised that a 30ft difference would have such a large effect.


  • DOF calculations are based on old rules of thumb with an old definition of what is acceptably sharp. With modern cameras, and certainly with the super high pixel count cameras like the D800E, what is acceptable sharpness at pixel level requires a much more stringent definition of acceptable sharpness. The DOF as a result will be much, much, shallower.

Further info regarding the second point: normal calculations for Depth of Field take 0.025mm or even 0.03mm for the “circle of confusion,” with a pixel size of only 4.88µm, this means 5 to 6 pixels (in each direction!), similar to viewing at 20%…

Try the same shot again, but now with focussing somewhere between the closest and farthest points of what you want to have in focus (you may also try going to f/11 though this already may introduce some diffraction effects).

Also, to rule out any environmental or vibration related issues, try the shot at daytime with a short shutter speed.

Unfortunately the place is closed now until February so I can't try any further shots until then.  :( :(
Hope this helps,
Hayo
« Last Edit: 26/12/2012, 15:37:19 by Gromit44 »

Hayo Baan

  • Professional Photographer and Workshop Leader
  • Administrator
  • Sr. Member
  • *****
  • Posts: 210
  • Professional Photographer and Workshop Leader
    • View Profile
    • Hayo Baan – Photography
Re: Nikon 14-24mm focussing difference at edge of frame
« Reply #3 on: 26/12/2012, 14:36:00 »
Though I am quite surprised about your findings (I too would have though the DOF would have been enough at the relative distances we are talking about here), DOF is still the most likely explanation.
(Note: The plane of focus of the 14-24 may not be perfectly flat, but I don't expect that even if there is any curvature in the plane of focus, it will be big enough to explain the behaviour you're seeing.)

Pity you can't retake the shot and focus somewhere in between the closest and farthest point as I suggested. Hopefully you can still get the result you're after by blending the two images.
Hayo Baan – Photography
www.hayobaan.nl

Gromit44

  • Jr. Member
  • **
  • Posts: 23
    • View Profile
Re: Nikon 14-24mm focussing difference at edge of frame
« Reply #4 on: 26/12/2012, 15:49:22 »
Yes, the manual for this lens gives huge DOFs at 14mm and f/8. Here are figures from page 104:

  • 0.54m to infinity when focussed on a subject 1 metre away
  • 0.92m to infinity when focussed on a subject at infinity

So with a focus distance of 210 feet (64m) the DOF should have easily covered both the centre and edges of the building.

I wonder if the reason for the strange behaviour has something to do with the front element of the 14-24 being ashperical.


Hayo Baan

  • Professional Photographer and Workshop Leader
  • Administrator
  • Sr. Member
  • *****
  • Posts: 210
  • Professional Photographer and Workshop Leader
    • View Profile
    • Hayo Baan – Photography
Re: Nikon 14-24mm focussing difference at edge of frame
« Reply #5 on: 28/12/2012, 10:40:24 »
Yes, the manual for this lens gives huge DOFs at 14mm and f/8. Here are figures from page 104:

  • 0.54m to infinity when focussed on a subject 1 metre away
  • 0.92m to infinity when focussed on a subject at infinity

So with a focus distance of 210 feet (64m) the DOF should have easily covered both the centre and edges of the building.
Indeed, however, these figures are still using the “old” notion of DOF like I explained. So at pixel level these aren't applicable.

Quote
I wonder if the reason for the strange behaviour has something to do with the front element of the 14-24 being ashperical.
Don't really think so; this has been all designed in. But still, perhaps it does influence it a bit.

Anyway, best test this again in a more controlled situation with ample light and good shutter speeds, focussing in between, etc.
Hayo Baan – Photography
www.hayobaan.nl