4
\$\begingroup\$

When panning, I find that I can keep the entirety of subjects in focus when they're moving perpendicular to me -- straight L-R or R-L, as in this shot: JPM at Indy, 2017

However, I notice that when I'm panning in tight corners, I almost always wind up with a very small target area in focus. I try to track the driver when I can, so in this example, the helmet area is sharp, but the front and the rear of the car aren't. At F/5 and a distance estimated at 7m or so, I know I don't have a huge DoF, but I'd expect at least 1.5m of the car in focus, and it's not even close.

Tight corner panning - Felix Rosenqvist

The biggest difference between these two (and other examples like them) is that the car is moving straight across my FoV in the first shot, but it's moving in an angle in the second.

panning diagram

I've seen this in dozens of panning shots in tight corners (and really only in panning shots), but I'm at a loss to explain why it's happening.

\$\endgroup\$

2 Answers 2

4
\$\begingroup\$

Panning only freezes motion that is parallel to the motion of the camera. Which means panning is most effective when the subject is travelling *in a straight line, and when the image is taken directly opposite from the camera position.

Additionally, relative speed changes along with changes in distance. So at shorter distances the relative speed increases (increased SS blur) and any additional change in distance is proportionally greater (DoF blur).

Using your situational drawings I added identical panning arcs. You can see that the straight line pan is more parallel to the camera path with less change in distance and less change in speed (required camera movement).

Whereas the converging corner pan shows much less parallelism; with a much greater change in relative distance and relative speed on the left side as opposed to the right side. Because of this you should see more blur on the left side than on the right side, with the greatest effectiveness (sharpness) near the center where the two paths/arcs most closely match.

Also note that, due to the greater convergence of the two paths, there is a greater change in parallax (angle of perspective). Which means details cannot overlay each other exactly. E.g. a wheel that appears more oval at a distance becomes rounder when closer and more perpendicular.

enter image description here


(*actually most effective when the two paths are exactly parallel; but that almost never happens)

\$\endgroup\$
1
  • \$\begingroup\$ I'm really happy your answer touched on the "more blur on the left" bit - something I'd seen anecdotally, but hadn't included in my question because I thought maybe my focus point wasn't quite as dead-on as I'd thought. \$\endgroup\$
    – D. Lambert
    Commented Dec 3, 2021 at 21:13
3
\$\begingroup\$

Geometry. The car rotates, so it's not a focus blur but a slight motion blur, which is minimal near the rotation axis.

enter image description here

enter image description here

Note that even if it goes in a straight line the car can be seen as rotating (you see more of the front at the beginning of the pan and more of the back at the end), but the global angle is much smaller so the induced blur is much less visible, possibly below your panning accuracy.

PS: There could also be a slight zoom blur, but applicable on the whole car.

\$\endgroup\$
6
  • \$\begingroup\$ The axis of rotation (in terms of motion blur) is not around the center of the vehicle. As the rear of the car rotates away the vehicle is simultaneously moving nearer... the axis of rotation would be where the net change in distance/velocity is zero (somewhere just in front of the rear wheels I would guess). \$\endgroup\$ Commented Dec 3, 2021 at 15:19
  • \$\begingroup\$ True, but the camera is panning, keeping the pilot's head at the center. \$\endgroup\$
    – xenoid
    Commented Dec 3, 2021 at 17:14
  • \$\begingroup\$ If the pan arc matched the vehicle arc and the vehicle was held stationary in the FOV, none of that would happen... your drawing assumes the angle of view is held stationary (which it does not). drive.google.com/file/d/1icHB55Unu6goQBZJ2kPrzY1SlwCLcJ8p/… \$\endgroup\$ Commented Dec 3, 2021 at 17:27
  • 1
    \$\begingroup\$ If the camera is at the center of the bend there is indeed no visible rotation since camera and car rotate together. But if you are shooting from quite far away your shooting direction doesn't change much, while the car orientation still changes a lot. \$\endgroup\$
    – xenoid
    Commented Dec 4, 2021 at 1:10
  • \$\begingroup\$ If the shooting direction doesn't change much it's not really panning is it? \$\endgroup\$ Commented Dec 5, 2021 at 16:04

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.