Welcome to How’d They Do That? — a monthly column that unpacks moments of movie magic and celebrates the technical wizards who pulled them off. This entry explains how they shot the bullet time stunts in The Matrix.
In 1999, The Matrix kicked Hollywood’s door down in a torrent of leather kit, tiny shades, and philosophy-riddled sci-fi bullet storms. Here was a sci-fi action movie with two key ingredients in spades: kung-fu and a techno-apocalypse, together at last.
Directed by Lilly and Lana Wachowski, The Matrix tells the story of a bored computer hacker named Thomas Anderson (Keanu Reeves) who discovers the world as he knows it is nothing more than a simulation. A computer-generated daydream, known as “The Matrix,” orchestrated by machines harvesting the energy in our warm, goopy human juices like Duracell batteries.
Joining up with a group of underground insurgents, Thomas, who adopts the name Neo, learns that he may be “The One,” a messiah prophesized to break the machine’s vice grip on humanity. Armed with the knowledge that the “real” world is a barren wasteland full of humans waiting to be harvested, Neo must recognize his full potential if he is to liberate humanity from their mechanical overlords.
Bullet-time in The Matrix
In the third act of the movie, Neo and Trinity (Carrie-Anne Moss) enter the Matrix to rescue Morpheus (Laurence Fishburne). Their fearless leader is being tortured by the nefarious Agents for information about Zion, humanity’s underground city in the “real world.” During the mission, Neo and Trinity find themselves on a rooftop, duking it out with Agents and their minions.
Staring down the barrel of an Agent’s pistol, with no weapon of his own, Neo finally gains the confidence to lean into his abilities. The Agent releases a volley of bullets, and as the script puts it, “we enter the liquid space of — bullet-time.”
As Neo defies gravity in a backbend, the air hisses as angry lead rips through the air, each flight path leaving a wake like a skipping stone. Neo’s coat kisses the concrete as he twists and bends, narrowly escaping each shot until two stubborn grey balls shear his thigh and shoulder.
This kind of feat can only happen in a virtual reality where time and space can be manipulated by those in the know. It’s visual proof of Neo’s superhuman reflexes and his prophesized mastery of this virtually constructed space. Warping time and space, blending the digital and material worlds, “bullet time” is one of the most iconic effects shots in action cinema and is intrinsically tied to The Matrix and its themes.
So, how’d they do it? How is bullet-time different from slow-motion? And why does Keanu Reeves move at a different speed than the camera?
How’d they do that?
Long story short:
In The Matrix, bullet time is achieved on a green screen set with a series of still cameras surrounding the subject. The cameras are activated in rapid succession, and the resulting frames are displayed consecutively with additional CGI interpolated frames, creating the effect of the camera moving faster than the subject.
Long story long:
Before we dig any deeper, let’s make sure we’re all on the same page about what bullet-time means. In the most basic terms, bullet-time describes a slow-motion effect with dynamic camera movement, usually a pan that allows viewers to see 360 degrees of an action.
While “bullet time” is probably the term Warner Bros. would like you to use — considering they trademarked it and all — there are a lot of other ways to describe the effect. These include, but are not limited to: frozen moment, temps mort, time slicing, and my personal favorite, flow-mo.
Bullet time is characterized by an extreme transformation of both time (e.g. being able to see a fired bullet) and space (e.g. the camera moving at a “normal” speed while the subject’s movement is slowed).
This visual impression of detaching the time/space affecting the viewer’s point of view from that of the visible subject is not possible with conventional slow-motion. If the effect were attempted with regular slow-motion techniques, the camera would have to move implausibly — or at least impractically — fast.
So the problem to solve is this: how do you create the illusion of a continuous shot moving at a different speed than that of the subject?
Proof of concept tests were first conducted by Mass.Illusions in Lennox Massachusetts. “We did the bullet time tests to help producer Joel Silver raise money for the movie,” explained Pierre Jasmin to the VFX magazine before & afters in a 2019 interview.
According to Jasmin, the Wachowskis presented the team at Mass.Illusions with evocative storyboards for what bullet time would look like. John Gaeta, who directed the test shots at Mass.Illusions, would go on to supervise the visual effects for The Matrix via Manex, a VFX facility in Northern California.
The first step to realizing bullet time is blocking. Because in order to set up the all-important camera rig (well, cameras rig), you need to know how your subject is going to move. So, first things first, Gaeta and co. blocked out the action using a series of still image cameras. They then scanned the images into a computer so that the filmmakers could map and strategize the position of characters and the path the “virtual camera” would take.
Using these simulations as a guideline, cameras were placed side-by-side on a specialized rig that was set up using a motion-controlled laser-pointing system that could judge the correct angle and focal distance.
For the most famous iteration of bullet-time in The Matrix, where Neo dodges bullets on a rooftop, the set-up involved 120 still cameras and two film cameras. Placing the cameras close together was imperative to create the illusion of motion, as each camera only captured a single still photo. These cameras were triggered at extremely close intervals, so the action would appear to unfold slowly as the viewpoint moved at a “normal” pace.
The cameras were either fired sequentially or all at the same time depending on the desired effect. Single frames from each camera were then arranged and displayed consecutively to produce an orbiting viewpoint of a frozen or slowed action.
Once scanned into the computer, the resulting strip of still images is not dissimilar to animation cels. And speaking of animation, now is as good a time as any to talk about interpolation, the first of several techniques that give the final effect a fluid appearance. Bluntly put, “motion interpolation” is a process that uses an AI-powered algorithm to “animate” more frames in a sequence of stills.
Also known as motion interpolation, the technique effectively creates a perceived increase in frame rate by creating new intermediate frames. This makes the motion feel less jittery and allows parts of the image (e.g. Neo’s cape) to flow more smoothly.
The Matrix’s bullet-time sequences could not be shot on location, since most of the cameras could “see each other.” Scenes were shot on a green screen and the backgrounds were replicated using photos of the real locations as textures for the 3D modeled environment.
Actors were held in position by wires to both prevent them from falling over and to prolong their actions so that the final filmed movement looked effortless. Digital elements (bullets, light, etc.) were then added in post-production.
In the end, all of these elements (digital compositing, optical flow, frame interpolation, etc) work together to improve the fluidity of the apparent camera motion, creating the bullet time you know and love.
What’s the precedent for bullet-time in The Matrix?
For all its modern bells and whistles, the core idea of using a group of still cameras to freeze motion predates the invention of the moving image. French physiologist Étienne-Jules Marey first began experimenting with chronophotography in the 1880s as a means of better understanding animal movement, particularly flying birds.
Marey’s revolutionary idea (action photography) was to record several phases of movement on one photographic surface. In 1992, he commissioned “chronophotographic gun” capable of exposing 12 consecutive images on the same medium. Tell me there isn’t something a little poetic about the origins of bullet time going back to a literal camera gun.
Marey’s contemporary, Englishman Eadweard Muybridge, famously created sequential series of photographs of one of former governor Leland Standford’s race-horses (The Horse in Motion. “Sallie Gardner” running at a 1:40 pace over the Palo Alto track, 19 June 1878). Muybridge stationed 12 cameras along a race track. And each camera’s shutter triggered automatically as the horse tripped wires connected to an electromagnetic circuit.
In 1880, Muybridge commissioned a zoopraxiscope, an early projector, to display his images at public lectures. It’s wildly accepted that Muybridge’s zoopraxiscope served as primary inspiration for the Kinetoscope, the first commercial film exhibition system.
With that history lesson out of the way, let’s take a look at the direct inspirations behind The Matrix’s bullet-time effect. Conceptually, bullet time’s roots can likely be traced back to Japanese cel animation. An early example occurs in the 1966 anime Speed Racer (which the Wachowskis would later adapt in 2008). In the scene, a character leaps from a car and freezes mid-jump while the camera arcs from the front to the side.
As John Gaeta relayed to Empire magazine in 2006, the artistic inspiration for bullet time fell to Otomo Katsuhiro (the co-writer of 1988’s Akira) and French director Michel Gondry. “[Gondry’s] music videos experimented with a different type of technique called view-morphing,” Gaeta said, “and it was just part of the beginning of uncovering the creative approaches towards using still cameras for special effects.”
Gondry’s music video for the 1995 re-release of The Rolling Stones’ “Like Rolling Stone” has precedents itself, including Midnight Mover’s 1985 music video for “Accept” which was created with multiple cameras — around 17, if you count the frames. While far from fluid, “Accept” is an early example of a camera moving at a faster speed around a stationary-ish subject.
We also must acknowledge one of the earliest works of cinema to capitalize on the cool factor of dodging or out-maneuvering slow-mo bullets: the 1981 South African martial arts movie Kill and Kill Again.
And, while we’re here, it’s impossible to over-emphasize John Woo’s influence on The Matrix’s kinetic approach to gunfights. His work, particularly 1992’s Hard-Boiled, is largely credited with pioneering “gun-fu,” the action genre that combines the stylish swordplay of wuxia with, well, guns.
There was something in the air in the late ’90s that was inevitably driving towards bullet time as we know it. Commercials (like this 1996 Smirnoff ad and this 1998 Gap campaign) were doing it. And both Blade and Lost in Space feature bullet time-ish effects, seen a full year before The Matrix hit theaters.
However, despite its many precedents and peers, The Matrix’s take on bullet-time is unique because it combines three distinct elements: gunfighting, superhuman bullet-dodging, and time-slice effects.
As Gaeta puts it in the aforementioned Empire interview: the technique in The Matrix differed significantly “because we built it to move around objects that were themselves in motion. And we were also able to create slow-motion events that virtual cameras’ could move around – rather than static action.”
All to say: plenty of well-established aesthetic and photographic techniques went into making The Matrix’s bullet-time effects a reality. But in terms of execution, no one at that point had done bullet-time like The Matrix.