Eye/Face Detect Test - Canon EOS R, Fuji X-T3, Nikon Z6, Sony a9, Sony a7 III and Sony a6400

Feb 6, 2015
124
Central Ohio, USA
Andrew
Found this Youtube video. thought it interesting.

They tested the eye/face detect functions of the following cameras:
Canon EOS R, Fuji X-T3, Nikon Z6, Sony a9, Sony a7 III and Sony a6400

It is 20 minutes long, but if you scroll through the timeline bar, you'll see the thumbnails fear each camera as well as finding the side by sides.

If you don't want to do that, a results slide for specific tests are at the following times, so you can go and see the results for yourself without watching it all.
10:15, 10:34, and 13.56

SPOILER ALERT:
The Sony A9 did the best, with the Nikon Z6 not that far behind, then the EOS R and in last place the Fuji XT3.
 
Last edited:

Covey22

Hall of Famer
Feb 3, 2012
124
First problem: Studio lighting. :biggrin:

Real world doesn't give you strong even lighting like that. Some trends can be drawn, but I'd love to see how FD works under typical, dim and dynamic stage lighting where I find myself often.

Edit: That being said, the end of the video where it's totally dim says it all. Someday, it'll work as ubiquitously as AF in general does today, but not quite yet.
 

Luke

Super Moderator
Nov 11, 2011
214
Milwaukee, WI USA
Luke
to be clear, Tilman, I only meant that glasses did not make the "face" unrecognizable.

Obviously this only really matters at close distances with wide apertures, but I've seen plenty of poorly focused shots using traditional focusing methods as well. That problem being (I presume) that the frames of the glasses are a high contrast area for normal AF systems to look onto.

It would indeed be interesting to know HOW these face detect algorithms work. Does the camera "look" for what appears to be a "face" and then further look for the "closest eye" (which is what it SHOULD do, if you ask me)......or does it just lock onto a high contrast area somewhere in the box recognized as "a face"?
 

Latest posts

Latest threads

Top Bottom