A brief history of Crowd Duplication Effects In TV and Film
Since audiences have started watching scripted stories play out on screen, the creators behind them have pushed the boundaries of what’s possible. In the films of the early 1900s crews could hire thousands of extras to compose their crowds; but still, filmmakers wanted more.
In 1966 the film The Battle of Algiers featured few professional actors. The majority of the cast were locals, and in the final climactic scene, tens of thousands of extras were called in for an epic upheaval in the streets. Some were even guerrilla fighters in the battle on which the film was based.
Filming crowds of real people is an ambitious undertaking. They are difficult to choreograph, and large crowds pose obvious safety issues.
Luckily, technology has evolved to the point where thousands of real people are no longer necessary. All you need is a small team and the right software. However, the reason we’re able to digitally create realistic crowds today is no doubt because of the innovations of our predecessors.
Early days and In-Camera VFX
George Méliès is one of the earliest pioneers of Visual Effects. An experienced magician, Méliès began to experiment with film illusions during the late 1800s and early 1900s. He would manipulate the film roll while it was still in the camera, learning that he could duplicate the human body using double-exposure and superimposition. You can see his first successful experiments with body duplication in films such as L’homme Orchestre (The One-Man Band) in 1900 and Le Mélomane (The Melomaniac) in 1903.
Méliès played the main character in both films. He replicated his likeness by hand-cranking the same strip of film through the camera, changing his position with each take. In one of filmmaking’s earliest mattes, Méliès painted a pane of glass black, save for one section where his body would be in the frame. When he placed it in front of the camera lens, the unpainted section of glass exposed a small portion of the film. With each take he would re-paint the glass to expose a new portion of the film.
In Elizabeth Ezra’s book George Méliès, she writes of the artist’s frustration. Méliès said, “The actor, playing different scenes ten times must remember precisely to the second, while the film is running, what he was doing at the same instant in earlier takes…if his arm moves in front of a character photographed in the preceding take, it will be transparent and out of focus, which wrecks the trick.”
He continued, “You can see from this just how difficult it is and how angry you get when, after three of four hours of work and sustained attention, a tear rips through the film after the seventh of eighth superimposition, forcing you to abandon the film and do everything over again.”
Then, twenty years later, vaudeville actor Buster Keaton used the camera in a similar way to pull off his visual gags.
In The Playhouse (1921) Keaton takes the multi-exposure method a step further. In a dream sequence he not only plays every member of the band, but every audience member and stage performer as well. Keaton’s director Edward Levine created a camera surrounded by a lightproof box, with a set of shutters on the front. As each shutter opened, a portion of the action was isolated.
“To sustain this kind of storytelling during a time of hand cranked footage is nothing less than heroic,” says Molecule compositing supervisor Rick Shick.
The idea was similar to Méliès’s, but the results were much more precise. Keaton’s extensive training in vaudeville acting also allowed him to interact with his duplicates with perfect comedic timing.
LATE 1900s, Early 2000s, and Early GLimpses of Modern VFX
Although productions may have had the budget to hire extras, casting and corralling crowds of thousands was a logistical nightmare.
As a result, productions started looking for alternatives and turned to practical effects to make their crowds appear larger.
Filmmakers frequently used matte paintings to fill in stadium-style seating. In Ben Hur (1959), matte painter Matthew Yuricich painted a large crowd and a Roman background on glass, leaving a portion unpainted. The glass was placed in front of the camera lens, and the camera would capture the action through the unpainted section. The two images were joined during filming.
The crew dressed the dummies in period-appropriate clothing and gave them realistic (enough) looking faces. It was a great solution for crowds like theater or stadium audiences. Those crowds are often stationary and appear as part of the backdrop. If the cinematography is done well, the foreground action dominates the audience’s attention. Most of the time, viewers are none the wiser to the fake crowd in the background.
By 1995 we have one of the first glimpses into how digital compositing was working its way into filmmaking. Braveheart introduced a then cutting-edge method called crowd tiling to expand small groups of extras into two large armies. A group of actors would stand in formation on a hill, and directors would shoot the scene. The actors moved to a different location, and directors shot the scene again.
(Above, an example of a crowd tile and how it becomes part of a whole, from Hulu’s “The Path”)
VFX and CG Take Over
Today we use a several methods for creating a crowd. The main difference is that we do more in post-production now than ever before.
At The Molecule we have several approaches to crowd duplication in our toolbox. We like to blend techniques including practical, computer-generated animations, and compositing.
One of our go-to 2D solutions is to shoot plates of a small crowd in different arrangements within the frame, just like they did in Braveheart. With compositing software, artists stitch the plates together to fill all of the seats. VFX Supervisor Luke DiTommaso explains, “It’s an efficient way to combine practical with VFX.”
He continues, “A small number of extras can accommodate most tight coverage, but a wide shot would reveal empty seats. So we move the extras from section to section to shoot plates that artists composite together. Just like Buster Keaton, the extras have to time their reactions for each section so when they’re combined, the entire crowd is in synch.”
In addition, 3D programs like Golaem and Massive enable us to create battle scenes with armies of millions. Each has their advantages, but the common benefit is that artists can now build dynamic crowds in 3D.
For example, with 3D we recently augmented a shot from Amazon’s “Z: The Beginning of Everything” to fill the streets of 1920s New York City. The scene contained live actors, but the main thoroughfare still didn’t quite reflect the hustle and bustle that is iconic to NYC.
To fill in the sidewalks we turned to a program called Fuse. With it we dressed the 3D models in period-appropriate clothing, gave them props, and had them mingle with the live actors. The crowd in the finished render helped to sell the busy feeling we wanted from the shot.
Today, creating and compositing crowds in post gives us more flexibility and control today than ever before. Depending on the scale of the crowd, it may or may not always be cheaper to build it in post.
Although, you can’t deny that it’s much easier to wrangle a CG crowd of thousands than a real one.