How do they do this special film effect?

There’s a special effect I see mostly in Television commercial advertisements, and it is hard to imagine how they do it.
In this effect, the action freezes, and then the viewer floats around the subject. It’s as if time stands still, and you’re hovering around the car or the waitress or whatever is on screen. Things clearly change in proper perspective, passing in front of one another and so forth, as your viewpoint moves. Sometimes, action takes off again after the effect.
I had an idea how they did this, which is that they set up a zillion cameras along a carefully planned line, each one aimed properly considering its location. The first camera in the line films action up to the moment of the effect, and then each camera takes a picture just slightly after its predecessor, no doubt oh-so-perfectly coordinated by a computer managing the tiny time delays. If the action continues, it’s filmed by the last camera. But there are two reasons to doubt this explanation. 1) It would take maybe 10 to 30 cameras per second of effect, which seems impractically difficult for shooting some dumb commercial advertisement, and 2) I think some of these shots float so far around the subject that some of the cameras would see one another (but I am not sure of this, as it’s not easy to judge so quickly).

The effect is accomplished with the use of many cameras set up around the subject, all filming together.
I’ve got the perfect website for you to check out:

I stumbled across it a couple weeks ago, not even meaning to find the very informative information there. Fun stuff!

This affect was originally created for the film “The Matrix”. How it is done, is beautifully demonstrated in the extra stuff on the DVD. If you really want to see it in detail, get that… but here is a synopsis.

Actually, you don’t really need one… the method you describe is just about perfectly correct.

Two movie cameras run simultaneously and a BUNCH of regular still cameras (the matrix used 35mm SLR’s) with remote shutter control. The still cameras either fire all at once (totally freezing time for the transition between the movie cameras) or sequentially at some frequency calculated to give the right amount of passage of time during the shift from motion camera one to motion camera two.

When done, they take the frames from motion camera one, the all the still cameras between and tack motion camera two footage to the end.

Actually, that effect was developed by [url=“”]Digital Domain*, and it’s first application was a commercial for a bank. Doesn’t sound as cool, but there it is.

The famous Gap Khakis “Swing” ad also predated the Matrix.

Digital Domain.

I don’t know if there was a proper buzzword for the process, but for The Matrix it’s been dubbed “Bullet Time”.

*Originally posted by scotth *

The first film I saw this effect in was the fantastically awful movie Wing Commander, which came out before The Matrix.

I believe they also use computers to interpolate frames between shots, so they can use fewer cameras.

There is another way to create this effect using a bunch of cameras (but not as many cameras), plus a lot of computer processing. I saw a presentation on it at SIGGRAPH '97 (annual ACM computer graphics convention), which featured a movie clip showing a snowboarder riding a halfpipe. While the boarder was in mid-air, the action froze, and the camera started moving around him. The cool part was that the camera wasn’t just following a predetermined track – the person moving the camera was just using a mouse to move it around arbitrarily.

What happens is that you take footage from several different camera angles (I think they said they used about 6) whose positions are known precisely, and the extremely cool software processes those images and builds up a 3d model of the scene, complete with texture maps for every visible surface. From that information, you can render a new view of the scene from arbitrary camera locations – even places where it’s impossible for a camera to have been. Depending on the number and locations of cameras you used to record the scene, and the complexities of the shapes in the scene, you can end up with “undefined” regions that your 3d model can’t represent (imagine a group of cameras surrounding a guy holding a donut on a plate, but none above him – none of the cameras can see inside the donut, so you would have an undefined spot if you tried to generate a scene from that point of view), but placing the cameras more strategically reduces that.

I think the low-tech approach (lots of cameras, like they used in the matrix) is easier to do right now (for one thing, it doesn’t take enormous amounts of number crunching), and the 3d-izing process produces artifacts, so the low-tech approach generally looks better (for the time being).

The Matrix won the Special Effects Academy Award because of that damned effect, even though they did not invent it, and they barely could be said to have perfected it. They merely popularised it.

I’m happy to see so many people point out the mistaken belief that it originated with that darn movie.

By the way, the first time I saw the effect was in the British documentary series The Human Body but I don’t think they invented the effect for that series, I’m sure it had been used elsewhere before then. However, they used it really effectively, in particular of an image of a metalworker grinding something and sparks flying everywhere.

I’ve also seen the effect where they freeze or slow down time but one person or object in the scene moves in normal time. If I remember correctly, a commercial I saw did that and even managed to move the camera through the scene as described in this thread. While they could have shot the frozen time scene in the way described above, how would they put the normal moving person into the scene?

And what’s with those talking M&M’s???

In that case they use either a blue or green screen. Check out this site to see what I mean-

How Blue Screen Special Effects Work- Marshall Brain’s How Stuff Works.

They accomplish what you’re talking about by isolating the image, or thing they want slowed down and stopped, and put it in front of the screen. Then, the behind action, as it were, is added in and can move any way they want it to, i.e. fast, or regular speed.

Basically, they’re combining two techniques- one old, one new- together, to get the overall effect you’re talking about.

You’re right, it absolutely did NOT originate with The Matrix; the first time such an effect was used, I believe, was around 1980, when a British commercial was shot using several dozen 35-mm cameras strapped to a board at ground level firing simultaneously. The pictures were shown in sequence, creating a right-to-left “tracking shot” of a perfectly still scene.

However, even though I didn’t like the film very much, I feel I do have to defend their Oscar. I believe they won not for the simple “camera moves around a central point” shooting method, but for all the equipment and software they created. The “bullet time” system allows for some very active and creative shooting possibilities. Shots can track forward and back again, zoom in, create “slow action” rather than stop action, start and stop film at any point along the camera track…it’s pretty hardcore. Consider, for example, the bullet-time scene where they’re rescuing Laurence Fishburne from the skyscrape. Film continues, although slowly, as the camera tracks rapidly from the back of the room to the front, turning 90 degrees in the process, then picking up speed slightly as we see Fishburne get shot in the ankle. Traditional “zoom the camera around the subject” systems couldn’t do stuff like that.

The computerized method was also used last year during the Super Bowl, in real time. It was fun to watch, but there were definite discontinuities.

I think the system set up for the super bowl was more like the “bullet time” method. I don’t think it was able to interpolate between the many cameras they had set up, nor did it actually build up a 3d world. The neat thing about that was that the officials were able to rotate the camera angle in realtime, but every shot they could choose from was an actual video picture generated by a camera.