Trying to understand contrast detection AF on X-Pro1

Discussion in 'Fuji X Forum' started by flaxseedoil1000, Apr 28, 2012.

  1. flaxseedoil1000

    flaxseedoil1000 SC Regular

    Apr 28, 2012

    I am coming from a GH2 which is fast and furious with it's AF. I have caught a single bee in flight with it and the Oly 45mm wide open, no problem.


    I understand the X-Pro1 has a shallower DOF so it's a more difficult challenge, both for it and me.

    But I also see times when it chooses to focus on the most contrast, regardless of where the focus area is:


    In that shot the focus area (set small, EVF) was completely inside the dog's body, yet the X-Pro1 back-focused on the more contrasty gravel.

    Is it possible that if the contrast inside the AF area is low it searches outside the AF area for something to lock onto?

    NOT BASHING, just trying to understand what's happening so I can adjust.

    btw, these are not beauty shots, just me trying to learn the camera.

  2. Armanius

    Armanius Bring Jack back!

    Jan 11, 2011
    Houston, Texas
    Real Name:
    No bashing taken Tia. "Most" XP1 users in this forum aren't Fuji fanboys.

    I've had false AF locks before too. But primarily in more challenging lighting conditions. One thing about the XP1 that all XP1 users have to accept first is that its AF system ain't state of the art! On the other hand, there's no AF system in the market that is infallible. I recall out of focus photos even when using the Nikon V1.

    I take it that you are already using single point AF right? I've read in the past somewhere that the size of the AF box may not necessarily be 100% accurate. In your situation, how much of the AF box was inside the dog's body? You said completely. But was it barely inside, or had lots of room before going beyond the outline of the dog's body?

    Personally, I found using C-AF plus the EVF/LCD to be the most accurate way to AF. My subjective testing led me to the conclusion that the C-AF is able to AF on less contrary subjects with a higher chance of success than S-AF.
  3. flaxseedoil1000

    flaxseedoil1000 SC Regular

    Apr 28, 2012

    Yeah, AF area was completely inside the body, single point, small, EVF.

    Let's try this so I can wrap my brain around it.

    Let's say the AF area is completely accurate.

    90% of what's inside the AF area has little contrast. 10% has lots of contrast.

    A typical CDAF is going to lock on the 10%, correct?

    Maybe that's what happened above, the AF area is a little bigger than indicated, part of it hit the contrasty gravel and that's what it locked onto.

    Making sense?
    • Like Like x 1
  4. flaxseedoil1000

    flaxseedoil1000 SC Regular

    Apr 28, 2012
    Just did a test, X-Pro1 and GH2, to see how they behaved.

    XF35mm, f/2.8
    Panny 20mm, f/1.7

    Focus area is as large as the X-Pro goes, EVF. Made the GH2 focus area the same size.

    Lens are about 18" from the wood, 7' from the gravel.

    I slowly moved the focus area from the wood to the gravel and then back.

    Only posting the X-Pro1 pics for this

    Changing from front to back focus

    X-Pro1 - Kept focus on the wood until 100% of it's focus area was on the gravel.

    GH2 - Kept focus on the wood until 110% of it's focus area was on the gravel.

    Changing from back to front focus

    X-Pro1 - Kept focus on the gravel until 110% of it's focus area was on the wood. *

    GH2 - Kept focus on the gravel until 70% of it's focus area was on the wood.


    I would think the gravel has more contrast and thus they would lock onto it sooner than 100%. In this case, the X-Pro1 actually beat the GH2 because it needed 110% to change focus.

    * I think this may be what happened on my previous dog shot that was OOF. The focus area was on the dog but not 110% so it never changed focus from the previous shot.

    I think the lesson here is to not assume the focus area is 100% accurate. Give it an extra 10% cushion to be sure.

    Sorry for the long post, it helped me figure some things out, hopefully it does the same for you.
    • Like Like x 1