D800 and Tamron 28-75mm, possible lens decentering?
9 months ago
Hey all, I got a D800 in mid July but I've waited until now to do AF-Tuning and check the infamous autofocus issue. I have a Tamron 28-75mm f/2.8 and a nikkor 50mm f/1.8G for now.
I used this test chart which seemed especially good for af-fine tuning ( http://regex.info/blog/photo-tech/focus-chart ), and followed the instructions carefully (I made especially sure that you could not AF on the light gray areas and reset focus to infinity after each shot). After fine tuning my lenses (chart at 45 degrees, camera on tripod with exposure delay, bla bla, yadda yadda) at max aperture I found out that the best values were +1 for both lenses which lead to simmetrical focus (upwards and downwards with respect to the 0) just like in Live View mode (although I can't use Live View that much on my Tamron since it doesn't autofocus in live view, it just hunts untill it reaches infinity).
I proceeded to test Left/Right AF with the 50mm f/1.8 at max aperture and 3 of the same charts with the centers aligned to each of the outmost AF points (the central one aligned to the center one, obviously). Took 1 shot in LV + 3 shots with the viewfinder for each point with the camera both at 45 degrees from the charts and ortogonal to them (meaning with the sensor parallel to the charts). In the first case I would compare which numbers were on focus (just like in a normal AF-Fine Tuning test), in the second case I would see if the outmost points caught focus properly (comparing focus/sharpness between live view and viewfinder shot of the same focus point used).
The first set of shots lead to a very slight back focus (out of focus area started at 11 above 0 and upwards and at 9-10 below zero and downwards) but it is more than acceptable (considering that the area where you are actually focusing on, 0, is totally in focus).
The second sets of shots lead to no apparent difference from LV to viewfinder shots (checked at 100%).
So ok, I am a lucky one (?) and I have no Left AF issue... good. However I also tried the 28-75mm with the same procedures just because it's wider at one end (even if slower). Here the problems begun... (please remember that I could not use LV to compare to in these cases)
on the 45 degrees shots at 28mm (still with 25x the focal length so less than 1 meter) central point would render in focus the area between +15 (using + for above and - for below, obviously) and -15 and right point would achieve a slight back focus (+17 to -13) but left point would completely back focused (focus area between 0 and the upper limit of the chart).
The other set of tests showed similar results, with the test charts parallel to the camera sensor the central and right points would show sharp images (with the right one being less sharp but still acceptable) while with the left point the image would be completely out of focus.
Very suspicious of the results I tried again at 50mm and 75mm (still 2.8) adjusting the camera distance accordingly and found the same results...
I thought it very strange that a 50mm lens at 1.8 would focus properly with all points while another lens (albeit a zoom) at 50mm at 2.8 (so with a wider DoF as well) would show completely different results... so I tried the same setup with this Tamron and my "old" D3100 (at 28mm and 50mm)
While the parallel (or orthogonal, depending how you want to view camera orientation) tests lead to acceptable results in all points (only 11 AF points, and DX format so it only uses the central part of the lens from what I understand) the shot with the left-most AF point was way less sharp than the right and center one (which were comparable in sharpness).
The 45 degree test lead to interesting results as well (reporting only values at 28mm, but 50mm was more of the same)
central point: focus area between -10 and +4 (front focus)
right point: focus area between -8 and + 6 (slight front focus)
left point: focus area between -5 and +9 (back focus)
Although a lot more subtle the problem is still present on the D3100, which leads me to believe that it's 99,9% a lens defect, and from what I am understanding from the web it might be related to decentering (probably tilting) of one of the lens elements.
What are your opinions about it? How much does "fixing" the lens would cost (based on past experience with tamron or even nikon lenses with the same problem)? Obviously the lens is no longer covered by warranty (otherwise I would have sent it in due to AF not working in LV), but I got it really cheap (200 euro) so even if I to pay a bit for repairs it would be probably worth it. If it ends up being unrepairable or too expensive I'll just keep it as it is, I shoot mostly with AF-S with central or right points anyways.
I might post the pictures later even if I see no need.
Thanks for everyone who will share their thoughts xD
|Post ()||Posted by||When|
|9 months ago|
|9 months ago|