Upload a front-facing portrait and mark five landmarks. The system scales an anthropometric face model to your measurements, adds surface relief recovered from the photograph, and grows a sealing shell onto it — then hands you a printable STL in millimetres.
It is a design study in respirator form, and it is sold as art. The geometry is genuinely engineered — real parameters, real physics, real printable output. The air-treatment side of the idea does not yet work, and rather than hide that, the failing number is on this page in three places. Read the mission and the current status.
What this is not: single-image capture cannot recover true metric depth. Landmarks set the proportions; photometric relief adds surface detail. A validated fit needs a depth scan or a plaster cast. Nothing here is a respirator, PPE or a medical device, and nothing here protects you from anything.
Exports are part of a purchased tier. Buy on the pricing page and your files arrive by email — or enter the unlock code from your order confirmation to export this browser's generated mask directly. STL is written in millimetres with only the visible parts included — use the face form and visor chips to choose what prints. New to it? The print guide and the assembly walkthrough cover everything downstream of the download.
These files produce a decorative wearable-art piece. Not a respirator, not PPE, not a medical device; no protection of any kind, and no electronics included or supported.
Buy · unlock exportsOne chin canister, one pass, six PZT roles. Click a stage to drive the model.
These recompute on every change. The dose budget is the honest one — read it before anything else.
Inactivation model: N/N₀ = e^(−k·D) with k ≈ 1.2 cm²/mJ for enveloped RNA virus at 265–280 nm, and a 55% LED-to-chamber coupling loss. Dwell time assumes plug flow at the design peak — a real reactor has a residence-time distribution, so treat the figure as an upper bound.
PulseMask is a research prototype inside Aurelius Dynamic, sold as wearable art while the engineering question stays open. Nothing here is a certified respirator, PPE or a medical device. Phases are gated on evidence, never on a calendar — nothing is marked complete because a quarter ended.
Readiness, stated honestly: the device sits at TRL 2 — concept formulated, nothing built, no experiment run. The generator sits at TRL 5 — validated across two independent hosts with matching output. The full status breakdown, with every number unrounded.
A full-face shell with a wraparound visor has a large internal volume. Breathe into it directly and you rebreathe CO₂ and fog the lens. The inner cup isolates the breathing circuit: treated air enters the cup, exhaled air leaves through the cheek valves, and the visor cavity is swept by a small bleed. Dead space drops from roughly a litre to the figure in the readout.
PA12 is tough where photopolymer is brittle. A resin shell that cracks at the visor aperture during a fall is a failure mode with the user's eyes behind it. SLS also prints the organic shell without supports, so no support scars land on the sealing face — that surface has to be smooth or the bead leaks.
At 85 L/min peak inspiratory flow, air crosses a 28 mm reactor in about 27 ms. No practical LED array delivers a germicidal dose in that window. The honest options are a long folded labyrinth, a blower that decouples flow from breathing, or treating UV-C as a filter-media steriliser between breaths rather than an inline kill stage. The sliders let you find the boundary yourself. Note what the physics says: for a uniformly lit tube the dose reduces to P·η·ρ·r / 2Q — reactor length alone buys nothing unless you add emitters with it, which is why the slider is LED density.
Five roles, one driver: acoustic agitation in the reactor, ultrasonic coalescence ahead of the filter to grow sub-micron aerosol into capturable droplets, seal-contact sensing for live fit telemetry, blower assist, and breath energy harvesting at the exhale valves. Capsid disruption by ultrasound is theoretical here — it needs bench validation before it is claimed.
The bore is about 11 mm at the base, so no 18650 and no 14500 fits — it takes a curved pouch cell, and both horns carry one. An antenna sitting on top of a cell gets detuned, so the radio moved to the crest keel: centreline, clear of both cells, long enough to radiate. The keel stops being decoration. Runtime is still the weak number — gating the reactor to inhalation only is what turns an afternoon into a working day.