26

I've seen several shots where an action sequence is halted and then the camera seemingly moves around while the action is frozen in place. An example of what I'm talking about can be seen in this YouTube clip.

I've also seen this used in sports action replays, where the play gets frozen and then the shot transitions to another camera's view of the play.

How are such shots filmed?

35

The technique is called "Bullet Time" and it is done with an arced bank of cameras surrounding the subject. By switching from one camera to the next in sequence the point of view swings around the subject creating the illusion of a single camera moving. In the example you cite, the director chose to use still frames from each camera to appear to freeze time.

enter image description here

In this clip » from "The Matrix," you can see how a director uses a green screen to shoot the subject and then drop in a background in post production.

  • Was the matrix to use this technology first or was there other movies who had done it too – war_Hero Mar 18 '18 at 9:22
  • 7
    I don't know about movies, but that Gap commercial for khakis predates The Matrix by over a year. – nitsua60 Mar 18 '18 at 21:05
  • Actually, the technique is called "slice of life" and predates Matrix by a few years. It was first seen in large release in a documentary series (by Robern Winston, I think), and there was a bonus segment that showed the multi-camera sync process. "Bullet Time" just added software interpolation of frames for smoother results, but that is not the main point. The underlying idea goes back to the 19th century. – JDługosz Mar 19 '18 at 9:09
  • A yes, I see Episide 6 of "The Human Body" which aired in June 1998 sounds right. – JDługosz Mar 19 '18 at 9:23
18

There's also the low-tech variation of the Bullet-time effect, with just one camera:

enter image description here

For this, you need :

  • a very bright scene to allow a short exposure time.
  • a camera with high framerate. There are very expensive cameras with insane fps (e.g. from Phantom), but there are very affordable mirrorless cameras (e.g. Sony or Nikon 1) with high fps, albeit with somewhat lower resolution.
  • Some device to rotate around the scene at high speed. It could even be just a coathanger attached to a string.
  • The camera should move much faster than the subject. It means that for a stop-motion scene, the camera can move as slowly as desired.

In the Grand Tour example above, it seems to be the more expensive variant though, since the water splash doesn't appear to be moving at all between the frames. Note that they used each frame twice (forth & back) in order to use half as many cameras.

  • I think the used this method in The Grand Tour example provided. You can see the water splatter change between frames. – Boondoggle Mar 18 '18 at 14:11
  • 1
    @Boondoggle, I don't see the water splatter changing between frames. – NeutronStar Mar 18 '18 at 14:45
  • @Joshua Yeah, you're right. – Boondoggle Mar 18 '18 at 15:06
  • IIRC, they used this same technique in Dragonball: Evolution. – ibrahim mahrir Mar 18 '18 at 22:18
5

There are two main techniques.

the first is to arrange a sequence of still cameras along a fixed track which is equivalent to a camera move. To completely freeze the shot these will all take an image simultaneously, for slow motion they can be fires in a sequence equivalent to the frame rate you want to achieve.

It is also possible to combine this with either a motion control video camera on the same track either simultaneously or in two separate shots which are later composited together.

Alternatively you can use a high frame-rate video camera with a fast motion control to achieve a similar effect with a single camera.

In some cases you will have an actor moving in real time with events around them eg a crowd or explosion happening in slow time or vice versa. This can either be done in green screen or by compositing different motion control shots together. In some cases you can get a wide range of frame rate by having one set of actors perform in actual slow time or keep as still as possible.

Its also not uncommon to use depth of field and possibly forced perspective camera effects to further enhance the sense of unreality.

As is often the case in film it is often a a matter of combining physical, camera, digital and post production effects to get exactly the result you want.

1

The answers here are good for the theatrical explanation. The sports replays, however, work differently.

A partial 3D model is generated, using images from multiple viewpoints and perhaps other sensor data (like a point cloud from laser scanning). The transition is thus CGI that moves the virtual camera from the position and orientation of one of the actual cameras to the position and orientation of another of the cameras.

This is much cheaper than installing the zillions of cameras that would be necessary to accomplish the same effect on something the scale of a playfield, and it's flexible enough to generate transitions between several key cameras.

A search for "interpolated camera positions in sports broadcasting" will turn up lots of references on this approach.

Not the answer you're looking for? Browse other questions tagged .