Is GH2 AF as fast as a Base Ball?

M

mpgxsvcd

Veteran
I wanted to test how fast the GH2 Auto Focus really is. Therefore, I decided to test it while throwing baseballs at it. I set the camera up about 2 inches behind a chain link fence. Then I proceeded to chuck baseballs at it to see if it could focus on the ball throughout its path.

I think the results are pretty freaking cool. The shots that are extremely noisy were all done at 1/4000 of a second shutter speed with the 14mm @ F2.5 and the ETC mode. The quality of the HD video goes out the window when filming with High ISO and the ETC mode because the ETC mode emphasizes noise. However, the AF is spot on.

Just check this video out and let me know what you think. Is this good enough auto focus?


http://www.youtube.com/watch?v=BEJ-otHLOhw

Please don't critique my form. I haven't pitched since high school about 18 years ago. I am incredibly sore today though.
 
It sure does a pretty good job. Thanks for posting it.
Now get back out there Spring training is just around the corner.
 
I'm not sure I get how this tests AF. Were you expecting the incoming baseball to cause the camera to hunt? Or were you expecting it to track the ball? I do not see the AF coming into play at all in these shots. Looks like you're just focused at infinity on a 14mm and everything is in focus and AF is not doing anything. On a 14mm lens at f2.5 when hyperfocal everything is in focus from 8ft-infinity and when the ball is any closer it seems like motion blur may be greater than focus blur anyway for such fast action.
 
Ok lots of people have stated that they don't think the depth of field is shallow enough. I will redo the test today at various focal lengths with the 14mm, 14-140mm, 20mm, and 45-200mm. I will also do some with the 55mm F1.2 and the 85mm F1.8 to show you what it looks like without auto focus.

Hopefully, my arm won't fall off.
 
Isn't there easier way to verify AF speed by swapping infinity focus and close focus? Or is this specific to AF tracking feature?
 
Forget baseballs, why not put on your longest lens wide open and walk/run toward the camera? Then motion blur will be less of an issue. If the camera can track that accurately, then focus pullers everywhere should consider a new line of work. However, I would not hold my breath to see it pull that off. If you think the AF is going to track a small, white object moving that fast, you will likely be disappointed. You can actually clearly see in your video that the camera does not even try to track the ball in the slow motion clip. The ball gets close and goes out of focus, thats it.
 
Last edited:
I did not see focus shift at all on these shots. Seems like depth of field is enough to handle the distances involved.
 
Razz16mm said:
I did not see focus shift at all on these shots. Seems like depth of field is enough to handle the distances involved.
Click to expand...

correct! the AF is doing zilch! focus is not shifting at all! I doubt this test is possible unless you take a dolls head and teach the AF "face recognition" then trow it :-)
 
I totally disagree with you guys. Flip between 0:01 and 0:11 and you can clearly see the background blurring while the ball remains sharp...

Although I agree it's a very small change and might just be an accident by the camera... Or even attributable to the codec breaking up...
 
Justin Kuhn said:
nice angle, very dynamic, put that in an actual project if you have the opportunity. agreed not a great AF test.
Click to expand...

It is funny that I was worried everyone would just dog on the content. It looks like my test is what they have taken exception to. When I shot it I thought everyone would think that watching me pitch a baseball was boring because most people think even watching a real major league pitcher is not very exciting.

I have seen that angle a million times and once I even used my face to stop the ball instead of the fence. Therefore, it didn’t even occur to me to use it as a piece. It would be cool to do a film on high school sports and really get in the middle of it. Maybe shoot some from on field or right in the action. I guess it is rare that you get to see sports video from the player’s perspective?
 
Jester2138 said:
I totally disagree with you guys. Flip between 0:01 and 0:11 and you can clearly see the background blurring while the ball remains sharp...

Although I agree it's a very small change and might just be an accident by the camera... Or even attributable to the codec breaking up...
Click to expand...

Nah. Nothing happenning. I agree about the suggestion to realistically test it by running towards the camera. There isn't a chance in hell of the camera even noticing this ball, let alone tracking it.
 
Lucas Adamson said:
Nah. Nothing happenning. I agree about the suggestion to realistically test it by running towards the camera. There isn't a chance in hell of the camera even noticing this ball, let alone tracking it.
Click to expand...

There is a big difference between “noticing” the ball and “tracking” it. For the camera to track the ball it would either have to recognize it as a face or have the ability to do the focus tracking mode that only the G2/G10 can do while recording video. Since the GH2 cannot do focus tracking at all during video that is not an option.

I stuck my GF1 behind the GH2 while recording it to see what face detection mode does. It always picked up the pitcher’s face as the focal point instead of the ball. It is actually quite good for “Face detection” but not useful in the baseball scenario.

However, in order to “Notice” the ball(or in other words set it as the focal point) all it would have to do is determine that the ball is the object that should be in focus. Since the GH2 validates its focal point 120 times a second(As opposed to 60 times per second for the GH1) it would have about 60 opportunities to readjust the focus while the ball traveled from the pitcher’s hand to the plate(75 mph for 60 feet 6 inches).

The GH2 has the ability to do an extremely small spot focus if the object is not near the edge of the screen. I have no doubt the GH2 could pick out the ball as the focus point and it would easily be able to adjust the focus during the flight of the ball.
 
mpgxsvcd said:
There is a big difference between “noticing” the ball and “tracking” it. For the camera to track the ball it would either have to recognize it as a face or have the ability to do the focus tracking mode that only the G2/G10 can do while recording video. Since the GH2 cannot do focus tracking at all during video that is not an option.

I stuck my GF1 behind the GH2 while recording it to see what face detection mode does. It always picked up the pitcher’s face as the focal point instead of the ball. It is actually quite good for “Face detection” but not useful in the baseball scenario.

However, in order to “Notice” the ball(or in other words set it as the focal point) all it would have to do is determine that the ball is the object that should be in focus. Since the GH2 validates its focal point 120 times a second(As opposed to 60 times per second for the GH1) it would have about 60 opportunities to readjust the focus while the ball traveled from the pitcher’s hand to the plate(75 mph for 60 feet 6 inches).

The GH2 has the ability to do an extremely small spot focus if the object is not near the edge of the screen. I have no doubt the GH2 could pick out the ball as the focus point and it would easily be able to adjust the focus during the flight of the ball.
Click to expand...

Nonsense. Prove it! Do a shallow depth of field version of this test.
 
Lucas Adamson said:
Nonsense. Prove it! Do a shallow depth of field version of this test.
Click to expand...

So what AFC lens should I do that shallow depth of field test with? There really isn't an m4/3s lens that can do a shallow depth of field at 60 feet 6 inches and still keep the ball in the frame. I need something like the Canon 55mm F1.2 that will auto focus. I don't see that happening anytime soon.

Here is what it looked like at 140mm F5.8. The issue was that the path of the ball was outside of the frame until the last second. It was too close to the camera to focus at that point.

548D874369EF443D81516E6102DD2D10.jpg

http://g2.img-dpreview.com/548D874369EF443D81516E6102DD2D10.jpg
 
mpgxsvcd said:
So what AFC lens should I do that shallow depth of field test with? There really isn't an m4/3s lens that can do a shallow depth of field at 60 feet 6 inches and still keep the ball in the frame. I need something like the Canon 55mm F1.2 that will auto focus. I don't see that happening anytime soon.

Here is what it looked like at 140mm F5.8. The issue was that the path of the ball was outside of the frame until the last second. It was too close to the camera to focus at that point.
Click to expand...

OK, any fast lens will do it. Test it on the 20mm f1.7 Lumix ideally. Test it on the 14-140 at about 25mm at it's widest. It will frame. It will be short dof. It still won't track it.
 
first.. replace the baseball with a snowball. Then paint a face on it. Then throw it in hell :) Its bound to work.

To the OP, good job on taking a relatively scientific approach and posting it for the scrutiny of your peers. In bright light GH2 has some pretty amazing autofocus. I would say it is almost comparable to a55. Not quite as ferverent but when I use it I almost get a feeling of intelligence. I am sure this is the result of panasonic algorithms that have duplicated my neural processes in silicon.

In anycase. it seems that you set the highest possible bar in terms of AF. Fast, small, white object moving towards the camera.
 
You must log in or register to reply here.
Back
Top