mid90s

Mid90s is Jonah Hill’s 2018 coming-of-age comedy-drama for A24, following a 13-year-old boy drawn into the skateboarding scene of 1990s Los Angeles. The film — Hill’s feature directorial debut, which he also wrote — stars Sunny Suljic, Lucas Hedges and Katherine Waterston as the boy begins spending time with a mostly older group of skateboarders. It had its world premiere at the Toronto International Film Festival on September 9, 2018, and opened theatrically in the United States on October 19, 2018, earning generally positive reviews from critics, who called it a “promising first outing for Hill.”

SCANable performed 3D scanning and modelling for the film, providing LiDAR capture and photogrammetry-based digital assets to support the production’s visual effects work.

A location is available exactly once

Location shooting borrows a place for a day. The permit ends, the dressing comes down, and the street goes back to being a street.

Anything the visual effects work will need from that place has to exist as data before the unit moves on. There is no version of the process where a dimension gets recovered afterwards from the plate — photography records appearance, not measurement.

Set and location scanning is the step that keeps the place recoverable later, at whatever level of detail the shot eventually turns out to need.

Coverage is a plan, not a quantity

Photogrammetric capture is not solved by taking more photographs. It is solved by taking the right ones: sufficient overlap between adjacent frames, consistent exposure, angles that see into every recess, and a path that closes back on itself.

A subject photographed haphazardly reconstructs with holes and warping in exactly the places nobody thought to check. A subject photographed to a plan reconstructs cleanly on the first pass.

The plan is made before the first frame, because the site will not be available for a second attempt.

Registration: many scans, one model

No single scan position sees an entire location. Interiors occlude themselves, and exteriors run well past the range of any one setup.

The finished dataset is many captures aligned into a single coordinate space, and that alignment has to hold everywhere at once. An error that is invisible in one room becomes a visible offset at the far end of a building, which is why registration is checked globally rather than pair by pair.

What ships is one coherent point cloud and, where the production needs it, clean modelled geometry built from it.

More feature film scanning is in the portfolio, alongside the full service list.