VFX Supervisor

What they do

Starts in prep, breaking the script into shots and settling methodology for each one: practical, in-camera, LED volume, full CG, or a two-dimensional fix. Bids and schedules with the VFX producer, then works with previs to test sequences before anything is built. On Solo the train heist was previsualised against a photo-model of the Dolomites assembled from roughly 25,000 photographs, and the plates were later shot from a helicopter nose mount in the same real canyons. On set they are the person who says the crane needs to move, the tracking marker belongs on that wall, and this take is unusable because the eyeline missed a creature that is not there. They run data capture: HDRI and lighting reference, chrome and grey spheres, colour charts, lens and camera data, LiDAR and photogrammetry of sets, cyberscans of cast and props, clean plates and witness cameras. The Solo team captured tens of terabytes of on-set reference and scanned over 500 characters, props and sets. In post they run reviews, first internal dailies with artists and then approvals with the director, giving notes on light, integration and story, and renegotiating scope every time a sequence changes after lock. A large part of the job is protecting the shot list against the schedule and telling a director what a change will cost.

What they decide

The calls that belong to this role and no other.

• How each shot is achieved: practical, in-camera, volume, CG or a 2D fix
• What must be captured on set for a shot to be possible, and whether a take is usable
• Which vendor or internal team gets which sequence, agreed with the VFX producer
• What the shot count and chosen methodology mean for the bid and the schedule
• Which notes go to artists and which get escalated to the director
• Whether a problem is solved in visual effects, in the grade, or in the edit

Principles of the craft

• The plate decides the shot: almost every expensive visual effects problem started as a cheap on-set problem. Frame lines, eyelines, marker placement, a clean plate and one lighting reference cost minutes on the day and weeks afterwards.
• Choose the cheapest method that survives the cut: a shot on screen for eleven frames and a shot the audience stares at deserve different approaches. Methodology is a budgeting decision as much as an aesthetic one.
• Capture more reference than you think you need: you cannot go back to the set. Lens data, lighting reference and scans are insurance whose premium is paid in minutes on the day.
• Give notes about story, not pixels: telling artists what a shot must communicate produces better work than a list of pixel corrections, and it lets experienced people finish the last, fiddliest part of a shot without second-guessing.
• Name the one thing that must stay right in each sequence: identify what the director actually cares about, say it out loud, and defend it through every version and every schedule cut.
• Show editorial something real early: rough comps burned in on the day let the cutting room work with the shot instead of imagining it, which changes the notes you get months later.

Common mistakes

• Agreeing a shot count without agreeing methodology, then absorbing the difference in overtime
• Letting the day's schedule squeeze data capture, and paying for it in matchmove and lighting
• Approving shots on a small monitor rather than at the size and in the colour pipeline the audience will see
• Signing off a plate without checking the eyeline and the interaction the actor could not see
• Giving only technical notes, so artists deliver clean shots that do not read dramatically
• The competent supervisor's mistake: silently absorbing every change instead of naming its cost, which becomes unfinished shots at the end of the schedule

How to learn it

Exercises and routes in, not a reading list.

• Break a five-minute sequence from a finished film into a shot list with a methodology and a difficulty rating for each shot. Then find the film's published breakdown and compare your calls with what was actually done.
• Shoot one shot that needs a sky replacement and a set extension. Capture a clean plate, a lighting reference with chrome and grey spheres, the focal length, the camera height and some measurements. Then shoot it again with none of that and try to comp both.
• Direct someone to react to something that is not there, shooting three takes with different eyeline references. Composite a simple element into each and see which take is usable.
• Run a review on three versions of one shot at full size in a proper viewing setup, and write notes about the story point rather than the edges. Then write the pixel notes separately and see how short that list actually is.