by Eric Jacobus
I spent years doing traditional pre-vis (or “previz”) for action scenes in films and shows like Altered Carbon, Black Panther, and A Good Day to Die Hard. I took this overseas for Heart of a Champion and Man Who Feels No Pain. Previz is a video blueprint for a movie. They did one of the first pre-vizzes for The Karate Kid, essentially a walkthrough of the entire movie. You can see it on YouTube [EDIT: looks like the video is gone. Pre-viz became more advanced with digital filmmaking, which Yuen Wo Ping employed in The Matrix‘s previz, and with 3D tools in the sequel. Serenity‘s previz, done by 87Eleven, employed sound effects, props, and crowds of stuntmen wrecking on concrete and wooden stairs. This was probably when the previz market exploded. Every indie action filmmaker had learned camera and editing skills over the previous 10 years. They scored big Hollywood jobs, but those filmmaking skills sat dormant. Now they could be employed to full effect to sell a coherent action vision. (The industry term, in the action world, is “stunt-viz”.)
High quality stunt-viz became its own selling point. It became common to work stunt-viz into the budget. The market began to demand that stunt-viz include, besides choreography blocking, all camera angles, editing, sound design, visual effects, music, color correction, and if possible wirework.
The limitations of a live-action stunt-viz required constant re-shoots, repeated falls and reactions for the stunt performers (unwelcome wear-and-tear), and many late-night re-edits. The result was the equivalent of a short action film created over the course of a week or two, which the production could use to demonstrate its high-quality action team.
When it came time to shoot, it was anyone’s guess whether they would actually use the stunt-viz. Most of the choreography would inevitably be thrown out due to time or performance constraints. They threw out the entire rooftop stunt-viz for A Good Day to Die Hard, but I never found out why. They might use some camera angles from the stunt-viz, but the DP will have his own vision. (Forget even asking him to come to the stunt-viz session. He won’t.) And if they do use the stunt-viz edit, then you’ve found a unicorn, or the production just wants everyone to be happy so they can traffic heroin on the side.
At any rate, whether or not any of the stunt-viz was used in the project might not matter. The team still got a long gig out of it, and the stunt coordinator got a high quality demo reel with the stunt-viz.
Since then, previz has become an entire market. Halon, Third Floor, and every stuntman on earth has the means to create high quality previz. Some of them are ridiculous in their production value and have so much gloss they could almost be short films. Almost…
What exactly can you do with stunt-viz? Or live-action previz in general? You can polish it up and add vfx and try to make it into a short film. At best, it’s a bunch of stuntmen in workout pants doing choreography in a gym. Live-action pre-viz just doesn’t carry very far beyond:
- Reference for the production
- A demo reel to pitch for the next production
- Fun behind-the-scenes material that you hope will get a couple thousand views on YouTube
The Process Is the Problem
Live-action pre-viz is a process problem. The reason stunt-viz loses value is because film productions are linear by their nature. They weren’t always that way. Chaplin and Keaton films were, in a way, non-linear. They were like live performances when everything happens at once, only with a camera. Set-construction was Keaton’s specialty, and his gags hinged on this process. Chaplin would rearrange entire scenes to get a gag right. Jackie Chan, with the same live performance background as these Vaudeville performers, used the same process to make his great works. These productions were vertically integrated. The auteur‘s vision had perfect continuity because he exercised control over the elements of production. And that’s how you made good comedy. The auteur was a busy guy because he had to ensure every department carried his vision to completion. Or he just did it himself.
The studio system commodified film by making the process linear. Set construction was graciously taken off Keaton’s back so he could focus on things the studio deemed more important, like learning his dialog lines to take advantage of the new sound capabilities of film. This was the death of Keaton and the rise of screwball stars who could say funny things in funny ways. Fortunately, Chaplin had a good voice… and his own studio. The filmmaking process would become more like an assembly line.
The linear studio system is what we have today. It’s Netflix, Universal, WB, and Disney. Production departments have relative autonomy over their processes. There’s some oversight, but generally these teams are free to do what they need to do, but they do it with caution. Camera team will overshoot (just to cover themselves), and editors will edit the mountain of footage. When a single edit might work perfectly in a scene, the editor might use ten, because ten angles were shot. And you don’t want to throw stuff away. The camera operator doesn’t edit the film, and the choreographer doesn’t shoot the action. The common result is the “Blockbuster style” – lots of camera angles, lots of editing, lots of money. Bollywood and Chinese blockbusters are the same. The linear process is the antithesis of the Chaplin, Keaton, and Jackie Chan genius.
You got a job at Marie Calendar’s because they tasted your grandma’s signature apple pie recipe. Now you work the line building pies. There are fifteen stations of the Marie Calendar apple pie. Your job is to cut the apples. The guy down the line puts marshmallows in the apple pie, because they sold 80 million applemallow pies in China last year. You wish you could make your grandma’s apple pie, but hey, it’s a job.
The Non-Linear 3Viz Process
I did some motion capture for God of War and some other games. One day I walked past a sound booth on the way to the mocap stage, where a sound designer was working on the sound for today’s mocap shoot. This broke my linear filmmaking brain. How can you predict what the sound design will be for something you haven’t even shot yet?
Game design, and 3D filmmaking in general, is not a linear process. It’s a spiral. I snapped this photo of the game design process during a Unity presentation:
This isn’t a revolutionary way of thought. This is how great action and comedy were made almost a century ago. Sometimes, great ideas are very old. The process also applies to virtual production, which is when 3D engines and filmmaking cross paths. In a virtual production, you can motion capture animal movements and stream it live onto an LED wall or into a green screen, composited on the fly, and tracked with the camera movement. Once disparate processes of filmmaking suddenly collide into the same moment. The auteur’s vision can be executed at every second, but only if he can grasp the tech.
That’s the moment I took a right turn from the traditional, live-action world and began learning Unity, Unreal, MotionBuilder, and the Xsens system. We created action scenes using these tools, pitching them as high-end 3D pre-viz, which I dubbed 3Viz. With 3Viz, we could shoot and edit the pre-viz, or we could ship it to the production and let them do it. Reshoots and re-edits in 3D were as simple as moving some camera icons around, altering the timeline a little, and re-exporting. A reshoot might take a couple hours for a single person. The alternative was the live model, which meant getting our 15 stuntmen together again at the gym and re-shooting and re-editing everything.
The goal of 3Viz is to get the director’s action vision solidified before post-production, before cameras roll. The director might want to change the environment to accommodate the action. He might want a character to be 30 feet taller. Or move the sun 90 degrees west. All of this is 3D modeling 101 and requires a few clicks. Finishing the 3Viz mocap, shoot, and edit requires a team of 4-8 people, who can work remotely from cruise ships or hot springs at the same time and see live updates. The film is pre-finished this way.
The 3Viz is sent to all relevant departments. Art department replicates the textures when painting the set, carpentry builds the set that the director devised for the action, wardrobe is looking at the asset costumes, and camera team has a very defined shot-list. Any shots they can’t accomplish practically have already been sent to VFX, who are using the camera animations created in the 3Viz to build those VFX shots. VFX also have the motion capture files, character assets, and anything else for creating VFX shots in post. Publicity are using the assets for creating posters and social media posts to promote the film. Sound department is designing a soundscape based on the 3Viz edit. The composer has already playing to the 3Viz, and his music can be played on set like Morricone’s.
Cameras haven’t even rolled yet, but the film is almost done. During shooting, production can change the lighting on the fly to reflect the 3Viz using LED walls, and other lighting setups were pre-programmed weeks ago. Dailies are passed to the editor, who edits to the same 3Viz edit that he’s supervised for the past few months.
Post-production? What post-production? Clean it up, take a week-long vacation, and release the movie a month after shooting completes. The result is an action vision that is exactly the way the director planned it.
Cabin Fever is a demonstration of the 3Viz process which allows continuity of vision throughout the entire production. 3Viz can be a pre-viz and just stay that way. Only shoot what you need. Mocap for a day, edit for 2 days, and it’s done. Iterate away. But all the assets acquired during pre-viz can be kicked up to production and be worked into the final product.
In the case of Cabin Fever, the 3Viz IS the final product. With some facial capture, finger capture, asset creation, additional lighting and all that, we could have made it look as good as a Pixar film, but the action and the comedy, the bread and butter of the project, remain the same through all this. We hope you enjoy the short, but even more we hope you like the process. We cover that at the 5-minute mark.
Email me at firstname.lastname@example.org if you’d like more info, or if you wanna give it a shot.
WRITTEN & DIRECTED BY ERIC JACOBUS
PRODUCED BY ZAC SWARTOUT
ERIC JACOBUS AS “THE MAN”
DENNIS RUEL AS ZOMBIE, ROBOT, AND “BERNIE”
SET TECH/BEHIND-THE-SCENES/TITLES MARK R. JOHNSON
MOCAP SUPERVISOR/UNREAL TECH/MOTION EDITOR MIKE FOSTER
MOCAP TECHNICIAN CORDERO ROCHE
LIVE VCAM OP JEREMY LE
PA’s – CHRIS CORTEZ & DANIEL SHEPHERD
MUSIC: “HOME ON THE RANGE” BY CHRISTIAN LABRECQUE
“DAWN OF THE DEAD” THEME BY GOBLIN
“A HAPPY DAY” BY Z80
“GODZILLA THEME” BY ARTIFICIAL FEAR
“TERMINATOR THEME (COVER)” BY NEON FRONTIER
POST-PRODUCTION BY ERIC JACOBUS
BUILT IN UNREAL 4.23
SHOT USING GLASSBOX DRAGONFLY
SOUND DESIGNED IN ADOBE
MOTION EDITED IN AUTODESK MOTIONBUILDER
SPECIAL THANKS TO TJ GALDA AND ALINA KLINAEVA
CREATED WITH 3VIZ
COPYRIGHT SUPERALLOY INTERACTIVE 2020 ALL RIGHTS RESERVED