Pages: [1]   Go Down

Author Topic: AF points  (Read 2415 times)

ben730

  • Full Member
  • ***
  • Offline Offline
  • Posts: 234
    • www.benhuggler.com
AF points
« on: January 31, 2015, 07:22:36 pm »

Hi all
Can somebody explain me why there is (AFAIK) still no Camera available with AF points in the corners of the picture?
Is it not possible at the moment, what is the technical challenge?
Thanks,
Ben

dwswager

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1375
Re: AF points
« Reply #1 on: February 01, 2015, 12:35:58 pm »

Hi all
Can somebody explain me why there is (AFAIK) still no Camera available with AF points in the corners of the picture?
Is it not possible at the moment, what is the technical challenge?
Thanks,
Ben

Phase detection technology works best with good light.  The further out and into the corners, the light fall off makes phase detection less effective.  That is why focus and recompose is the best method when you can use it.  Also why most cameras have cross type sensors in the middle of the frame.  And why the 7DmkII focus system with all cross type sensors is actually an advancement.
Logged

Slobodan Blagojevic

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 18090
  • When everyone thinks the same, nobody thinks
    • My website
Re: AF points
« Reply #2 on: February 01, 2015, 12:50:05 pm »

Samsung NX-1's 205 focusing points cover 90% of the viewfinder.

dwswager

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1375
Re: AF points
« Reply #3 on: February 01, 2015, 12:56:38 pm »

Samsung NX-1's 205 focusing points cover 90% of the viewfinder.

How much coverage versus how well it works!  I have no experience with the Samsung, but my money would be on the AF performance of the 7DmkII.
Logged

Bart_van_der_Wolf

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 8914
Re: AF points
« Reply #4 on: February 01, 2015, 01:45:14 pm »

Hi all
Can somebody explain me why there is (AFAIK) still no Camera available with AF points in the corners of the picture?
Is it not possible at the moment, what is the technical challenge?

Hi Ben,

It depends on the type of AF.

For Contrast detection based AF there is not much of a restriction, just that there is enough light and some subject contrast at the region where one wants to measure it. It's strictly a local contrast operation. Of course, due to field curvature, the edges and corners may focus at a different distance than in the optical center.

For Phase detection based AF, it is necessary to have at least two optical paths that can be compared (for perfect alignment if focused) and that requires to use the edges of the lens for the maximum path angle difference. That also means that closing down the aperture, or focusing closer than infinity, will cut off the image forming rays from those edges. In addition, the additional optical system that focuses those edge rays from opposite sides of the optical axis on the AF sensor has a limited diameter, a max aperture of itself.

So theoretically one could make a phase detect AF that could focus closer to the edges when the aperture is wide open but that would be limited by the widest aperture of the lens. That would be confusing with interchangeable lens cameras, or zoom lenses that have a variable widest aperture, and still vary with focus distance. The field curvature issue also plays a bad role.

The current compromise will usually allow to use Phase detect AF with lenses that have a widest aperture of f/5.6 or sometimes f/8, and that limits the maximum diameter for the AF system.

There is a fundamental expression that relates the half-angle θ of the light cone in image space to the F-number N of a well-corrected lens:
N = 1 / (2 * sin(θ))

That also explains why smaller image sensors will allow to have Phase detect sensors closer to the edges than larger image sensors, it's an angle from the optical axis limitation.

Cheers,
Bart   
« Last Edit: February 01, 2015, 01:49:23 pm by BartvanderWolf »
Logged
== If you do what you did, you'll get what you got. ==

Telecaster

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 3686
Re: AF points
« Reply #5 on: February 01, 2015, 03:47:49 pm »

The Sony A7r doesn't quite offer 100% AF coverage but it gets close. This is contrast-detect AF.

-Dave-
Logged

AFairley

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1486
Re: AF points
« Reply #6 on: February 01, 2015, 04:49:11 pm »

The Sony A7r doesn't quite offer 100% AF coverage but it gets close. This is contrast-detect AF.

-Dave-

So does the D800 in Live View.  However, last week I shot a corner-placed resolution target with the Micro-Nikkor 55mm f2.8 AF stopped down to 5.6, put the LV focusing square over it and used the LV (contrast detect) focus to focus (doing 3 focus and shoot and picking the best).  I was shocked at how bad the resolution was.  So I reran the test, this time manually focusing on the target in magnified LV and the result was much much sharper.  So there was something that was keeping LV autofocus for working accurately, but not sure what it was.
Logged

skimasks

  • Jr. Member
  • **
  • Offline Offline
  • Posts: 96
    • http://www.justinpoulsen.com
Re: AF points
« Reply #7 on: February 02, 2015, 01:10:56 am »

I'm going to play devil's advocate and mention Hasselblad's True Focus system.

<.<
>.>
 :)

ben730

  • Full Member
  • ***
  • Offline Offline
  • Posts: 234
    • www.benhuggler.com
Re: AF points
« Reply #8 on: February 02, 2015, 06:58:45 am »

Hi Ben,

It depends on the type of AF.

For Contrast detection based AF there is not much of a restriction, just that there is enough light and some subject contrast at the region where one wants to measure it. It's strictly a local contrast operation. Of course, due to field curvature, the edges and corners may focus at a different distance than in the optical center.

For Phase detection based AF, it is necessary to have at least two optical paths that can be compared (for perfect alignment if focused) and that requires to use the edges of the lens for the maximum path angle difference. That also means that closing down the aperture, or focusing closer than infinity, will cut off the image forming rays from those edges. In addition, the additional optical system that focuses those edge rays from opposite sides of the optical axis on the AF sensor has a limited diameter, a max aperture of itself.

So theoretically one could make a phase detect AF that could focus closer to the edges when the aperture is wide open but that would be limited by the widest aperture of the lens. That would be confusing with interchangeable lens cameras, or zoom lenses that have a variable widest aperture, and still vary with focus distance. The field curvature issue also plays a bad role.

The current compromise will usually allow to use Phase detect AF with lenses that have a widest aperture of f/5.6 or sometimes f/8, and that limits the maximum diameter for the AF system.

There is a fundamental expression that relates the half-angle θ of the light cone in image space to the F-number N of a well-corrected lens:
N = 1 / (2 * sin(θ))

That also explains why smaller image sensors will allow to have Phase detect sensors closer to the edges than larger image sensors, it's an angle from the optical axis limitation.

Cheers,
Bart   

Thanks for the explanation! It seems that this problem won't be solved for DSLRs in the near future.
Ben
« Last Edit: February 02, 2015, 07:04:25 am by ben730 »
Logged
Pages: [1]   Go Up