LAIKA and Intel built a machine-learning tool to speed up rig and seam removal

befores & afters reports that LAIKA partnered with Intel's Applied Machine Learning team, working near the studio in Portland, Oregon, to build a prototype tool inside Foundry's Nuke that speeds up the rotoscoping work behind rig removal -- the standing practice, running since Coraline (2009), of painting every tie-down point, wire, and puppet-face seam out of the finished frame. Trained on as few as five or six of LAIKA's own finished shots, the tool tracks facial landmark points on a puppet face and outputs roto shapes already smooth across space and time, compatible with standard tools such as Mocha, rather than a blank frame for an artist to start from scratch. Lead roto artist James Pina put the gain plainly: 'What the tool outputs is fully finished roto. It's a lot faster than we could get if we were doing it by hand -- it's hours of work in a few minutes.' A companion befores & afters report on Missing Link (2019) and a Heroic Hollywood set visit put scale on the underlying job: more than 1,000 rigs painted out across the film's 1,486 shots, cosmetic clean-up on over 106,000 individual puppet-face frames, and head of rigging Ollie Jones's summary of the trade -- 'That's the sadness and happiness of a riggers life.' The practice began, per the same reporting, when director Henry Selick wanted Coraline's puppet-face seams left visible as evidence of the handmade object, and LAIKA founder Phil Knight asked for them to be painted out instead. This database's Wiki now carries the durable version of the workflow on a dedicated Rig removal page, since a single studio's tool partnership is exactly the kind of fact that link-rots once the announcement cycle moves on.