Contact us
Install Your LFAM System Install Your LFAM System
3D Print Parts With Us 3D Print Parts With Us
contact us contact us
The Microfactory Behind Justin Bieber’s SKYLRK Seating at Coachella 2026 image

Decibel printed 30 large-format benches for SKYLRK Oasis at Coachella 2026 on Caracol’s robotic LFAM technology, running the same digital file in New Jersey and Texas at the same time. Roughly two weeks from first print to finished set. A technical breakdown of how a distributed microfactory model, one of the clearest advanced manufacturing use cases to date, handles fully custom geometry under a hard deadline by moving the file instead of the tooling.caracol-robotic-lfam-coachella-skrylrk-oasis-justin-biener1

Production context: the constraint set

SKYLRK Oasis was an invite-only installation for SKYLRK, Justin Bieber’s brand, sited on the eastern edge of the Coachella grounds in Indio, California. The brief called for 30 fully custom benches, organic and multiplanar, engineered for heavy public use and outdoor exposure well past the festival weekend.

Two constraints pulled against each other. The geometry consisted of multiplanar organic surfaces that a molding or conventional tooling route would only reproduce at significant cost and lead time. The schedule left no room for either. Design began in early March and all 30 units were produced over the following two and a half weeks. In fiberglass or concrete that turnaround is not realistic, because the tooling itself has to be designed, built, and cured before you can produce a single part.

Robotic LFAM changed the sequence. Since there is no mold in the chain, the tooling investment never needs to be committed before the geometry is final, which is what usually forces an early design freeze in molded and fabricated workflows. Decibel iterated a new version per day and moved the approved file straight into production, with no retooling step between the last revision and the first printed part. The design stayed open almost until production started, and the deadline was still met.caracol-robotic-lfam-coachella-skrylrk-oasis-justin-biener2

Process: distributed printing on Heron AM

Production ran on Caracol’s Heron AM platform. Robotic LFAM is what made the multiplanar surfacing printable at all, and how the team used it is what made the deadline. Rather than relocate tooling or funnel everything through one plant, they moved the CAD file. Decibel printed 15 benches in New Jersey. Caracol printed 15 in Austin. Both cells ran the same validated file in parallel, and the finished set was consolidated and shipped to Indio.

This is the microfactory model in its literal form: identical cells run one validated toolpath, sited close to where the parts are needed, and take a digital input straight to a finished part with no intermediate production steps. The payoff here was arithmetic. Splitting 30 units across two cells running in parallel roughly halved the print time compared with a single-site run, which is exactly what made the two-and-a-half-week delivery possible on a full set of custom parts.

Production specifications:

  • Technology: Heron AM – HF extruder
  • Material: rPETG + 20% GF
  • Production time: 30 benches in 2.5 weeks[ES1.1]
  • Dimensions: 7 ft length per bench

For an engineer the interesting part is what the process removes. Printing the surfacing directly kills the draft angles, split lines and mold-release constraints a mold would impose on forms like these. The hollow wall strategy keeps mass and material consumption low while still meeting the structural requirements of public load-bearing use, and the rPETG GF makes the parts reprocessable at end of life instead of landfilled, which is the relevant contrast with the way most temporary festival structures are built.caracol-robotic-lfam-coachella-skrylrk-oasis-justin-biener3

Repeatability: the open question

Underneath the benches sits a manufacturing argument. In a tooling-based process the mold is the fixed asset and it can only exist in one location, so scaling output means building more or larger tooling. In distributed LFAM the asset is the file, so capacity scales horizontally: you add validated cells instead of building a larger plant, and each new cell that can run the file adds throughput without duplicating any physical tooling.

There is a condition. One file has to yield an equivalent part whoever prints it, which means design intent, material behavior, toolpath generation and process control have to be integrated tightly enough for a development-validated process to survive the move into production. A design signed off in New Jersey and printed to spec in Austin is a small version of the problem. On this job it held, across two sites and 30 units, on a festival deadline.

The microfactory model’s practical limit sits there today. Holding that consistency across more cells and more materials, rather than one project at a time, is where the next gains come from.caracol-robotic-lfam-coachella-skrylrk-oasis-justin-biener4

Ready to unlock your next project?

LET'S TALK

Related Case Studies

Check out our success stories