Explanation for shallow depth of field with a fast-moving object

[JK6065]

Hi folks,

I was practising shooting cars in motion. I positioned myself right next to a 3 lane road. The cars were riding towards me on the lane on the other side of the road.


1/100 F10 ISO-100 @ 87mm

Now my question about this photo is: why is the back end of the car not that sharp, even at fstop 11.
(I do like the effect but I think it's a little too much and more importantly, I wonder why it happens)
I figgered out some kind of theory but it sounded a little strange and I wonder what you have to say about it.

[Dave Humphries]

Hi Jeroen,

It probably isn't a lack of DoF, I suspect it will be motion blur, the bit you focus on and (almost inevitably) pan with, will be sharpest. But everywhere else will not because it is moving relative to the lens/sensor - granted not as much as the background trees, but it happens nevertheless. The closer you are to the subject, the more pronounced this effect.

I was shooting little trains the weekend before last.
Have a look at this:
It is deliberately unsharpened, note how the cab area, where I was focused and panning with, is sharper than the "Henry" nameplate on the smokebox. Also how things get sharper as you go from left to right back along the boiler and running plate towards the cab.

If you cannot see the effect on the 800px image here, go see the full width image on PBase - beware, it is 4288px and 998kB.

Now it might be argued there is a focus error on this one, or that vibration played a part - all probably true, but I still think it is a valid point, if not the best shot to demonstrate it. I deleted all the ones that really showed the effect, as it made them useless!

Cheers,

[arith]

I found this in Wikipedia:

Distortion of fast-moving objects: although no part of the film is exposed for longer than the time set on the dial, one edge of the film is exposed an appreciable time after the other, so that a horizontally moving shutter will, for example, elongate or shorten the image of a car speeding in the same or the opposite direction to the shutter movement. For an example of such distortions see this image taken with a vertically traveling focal-plane shutter, showing extreme distortion of fast-moving helicopter rotor blades.

http://en.wikipedia.org/wiki/Shutter_(photography)

But it appears that part of the image is exposed before other parts; and with a moving object points will be in different places.

I don't know if this explains bluring though.

[arith]

Maybe in the future someone will invent an opaque lens which becomes opaque when a charge is applied; then combined with a shutter everything could be exposed at the same time.

[carregwen]

Arith

The problem here is a missing 'H' from the car number plate. Insert an H as the second letter and it will read SHARP VZ!

Seriously - if the focal plane shutter moves vertically across, I would have thought there will be a part of the edge of the shot that will be longer in view at the end of the exposure and will therefore blur a moving object. For a still subject this wouldn't make and difference, but a fast moving car could make a difference. The shutter doesn't just open and close, it;s a two-stage process.

Also, you are about 45deg to the side plane of the car. I know it was f10, but the angle could have affected the DOF slightly.

[Colin Southern]

if the focal plane shutter moves vertically across, I would have thought there will be a part of the edge of the shot that will be longer in view at the end of the exposure and will therefore blur a moving object.

Umm - would be interested to read how you came to that conclusion You might find this youtube clip interesting.

Also, you are about 45deg to the side plane of the car. I know it was f10, but the angle could have affected the DOF slightly.

It would be interesting to know the lens focal length and the camera-to-subject distance, but my guess would be that the DoF would be pretty much infinite past about 10m.

The other thought I had was could it just be poor lens performance towards the periphery?

[arith]

This is just a guess, and if it's stupid; well....

If you draw a line through the car in the direction of travel, you can measure distance qualitively at least by an angle to the lens.

The car is being tracked at the front and the angle at the back changes more rapidly relative to the front.

I haven't drawn any images to test it, it is just an idea since the car is getting larger as it moves closer.

[will_c]

This is just a guess, and if it's stupid; well....

If you draw a line through the car in the direction of travel, you can measure distance qualitively at least by an angle to the lens.

The car is being tracked at the front and the angle at the back changes more rapidly relative to the front.

I haven't drawn any images to test it, it is just an idea since the car is getting larger as it moves closer.

The car is an extended object moving across in front of you. In this case, from your position, up to the moment when you took the photo the front will have been sweeping through a greater angle per unit time than the back. So by tracking the front while panning you are panning too quickly for the back and hence blurring it. Even with infinite DOF and a perfect lens this would occur.

Will

[will_c]

The solution is to get the exposure time down. You're probably going to be better off with a somewhat faster aperture and higher ISO.

[arith]

The solution is to get the exposure time down. You're probably going to be better off with a somewhat faster aperture and higher ISO.

I thought that; but it might freeze the wheels.

[will_c]

I thought that; but it might freeze the wheels.

I figure that the wheels are probably rotating about 10 to 20 times per second, so I don't think that you would freeze them!

[wjh31]

Not sure if someone else already said this in slightly different word, but:

The car is moving in two ways compared to the lens. as it moves closer to the lens, it gets closer, but it also rotates. Consider in the distance you see the front of the car, as it passes you you get a side on view, once past you see the back end. When it is passing closest to you the apparent rotation is greatest. When you focus on the nose of the car, you kinda set the relative pivot point on the nose, so the front doesnt have much movement in this rotation, but the rear is further, so has a larger motion from this apparent rotation. If you look the extent of the blurring, it increases gradually from the front to the back. You can see this again with the train pic, the 'pivot point' is at the cab, so bluring increases towards the back, and towards the front as you move away from it.

To avoid this, try to focus on the center of the car rather than the front or the back, and/or stand on the centre of rotation on a bend so the the car is effectively orbiting you, staying side on the whole time with no rotation relative to yourself.

Hope this makes some sense!

[Dave Humphries]

Hi Will,

That's sorta what I was trying to demonstrate with the train pic up there in post #2 above.
I was on the out side of the curve, which made it worse

But next time ...