The atlas covers fourteen representative configurations across magnetic, inertial, magneto-inertial and electrostatic fusion. It is a conceptual anatomy collection, not a catalogue of every experimental machine. Model proportions, coil paths, field lines and time scales are simplified for inspection in mixed reality.
The plasma uses real-time GLSL filament and emission shaders. Green traces represent the prescribed magnetic-field geometry; blue points moving on the conductors indicate the direction of the displayed coil current. Gold ion and cyan electron markers follow analytic guiding-centre paths with oppositely directed gyromotion. The optional gas-injection view sends stylized particles from a fueling port only during the fueling stages. Pale neutral reaction products travel straight and appear only during the burn stage. Gyroradii, particle speed, emission and time scales are exaggerated for visibility. These are idealized relationships, not measured camera images or a solved magnetohydrodynamic, gyrokinetic, radiation-transport or nuclear-reaction calculation.
The thermal map is qualitative by design. Coil overlays are rendered cool/blue to represent the cryogenic or low-load magnet side of the system, while first-wall and divertor overlays move toward yellow/red as the model's heating and burn envelopes rise. This is not a calibrated temperature field and does not assign a false tesla, kelvin or heat-flux reading to the authored geometry.
In a tokamak, external toroidal-field coils and a toroidal plasma current combine to create winding field lines. The central solenoid can induce that current, while additional coils shape and position the plasma. Radio-frequency and neutral-beam systems provide extra heating. A spherical tokamak has a lower aspect ratio and a narrow centre column. The models show these different proportions.
Stellarators obtain rotational transform primarily from external coil geometry. The modular example shows fifty non-planar coils grouped around a five-period vessel, inspired by Wendelstein 7-X. It omits its full auxiliary coil and support system and is not an engineering replica. The quasi-symmetric example is a generic three-period concept. Its symmetry is a design objective, not a verified magnetic equilibrium.
The heliotron model shows two continuous helical coils with five windings each, illustrating the double-helix topology used in the Large Helical Device. The historical figure-eight model uses a continuous three-dimensional crossing with vertical separation. Field paths compensate for the rotating vessel frame to avoid counting geometric twist twice. The rotational-transform profile is prescribed, not calculated from the displayed coils.
A reversed-field pinch reverses the toroidal magnetic component near the plasma edge. The model reverses toroidal advance at outer radii while retaining the poloidal direction. A spheromak is a compact toroid with both toroidal and poloidal fields maintained by plasma currents. The field-reversed configuration follows closed contours of a prescribed poloidal flux function inside its separatrix, with no central material column. None of these analytic fields is an equilibrium reconstruction of a particular machine.
A magnetic mirror has open field lines with stronger fields near its ends. Many particles reflect, while a loss-cone population escapes. A Z-pinch uses current in the plasma to generate an azimuthal magnetic field and inward force. The animation illustrates compression; it does not solve stability or sheared-flow stabilization.
Mirror flux-tube area varies inversely with field strength. The displayed trapped-particle population bounces between mirror points and has smaller gyroradii in stronger fields; loss-cone transport and collisions are not numerically evolved. Inertial targets and electrostatic fusors have no artificial magnetic field traces: capsule particles compress with the target, while fusor ions accelerate inward. Individual collisions and reaction rates are not computed.
Every selection starts with fuel or target preparation, then proceeds without operator intervention. Stages last 12, 12, 14, 20 and 12 illustrative seconds. Every exhibit now shows the complete preparation, confinement, heating, discharge, exhaust and reset cycle before repeating. Stellarators, mirrors and fusors label their discharge interval as externally powered, but the demonstration still reaches recovery so the observer can see the cycle boundary. Target replacement is schematic, not an engineering demonstration. This is not a claim of ignition, net electricity, or unlimited operating duration. Tokamak examples demonstrate an inductive pulse; non-inductive steady-state tokamak operation is outside this sequence.
Hosted narration describes each stage using the Kokoro-82M neural speech model and its af_heart voice. This is synthetic narration, not a human recording. Each clip is written as an active observer cue: it directs attention to the inlet, coils, heating, plasma core or reset rather than reciting a passive label. The separate machine-audio channel combines procedurally synthesized pump, motor, power-supply and coolant ambience with a quiet sonification of the burn envelope. It is not a recording of any real machine. Plasma inside a vacuum vessel is not an ordinary audible roar: see ITER on perception inside a tokamak. Sound levels follow the exhibit stages and reduce during narration. Haptic stage cues are optional, default off, and require a held grip or touched trigger while the XR session is visible. Solid, translucent and plasma-only visibility are independent of assembled, cutaway, exploded and schematic views.
The optional transparent XR plots display the same normalized heating, field or drive, burn and compression envelopes used by the animation. Their horizontal axis is stretched exhibit time, not physical discharge duration. These are illustrative model values, not measured temperature, density, magnetic field in tesla, fusion power or an equilibrium calculation. A sustained trace means external power is maintained in the demonstration, not an energy-gain claim.
In the indirect-drive laser example, beams enter a hohlraum and strike its walls. X-ray emission drives capsule ablation and implosion. Twenty-four visible beam paths stand in for a much larger laser system. The capsule and hohlraum are deliberately enlarged relative to the chamber so their layered construction can be inspected. The five-stage cycle stretches an extremely short physical event into an observable sequence.
Magnetized liner inertial fusion combines an axial magnetic field, laser preheating and current-driven implosion of a metal liner. The disposable liner must be replaced between physical experiments. The electrostatic fusor accelerates ions toward a central grid; collisions may produce fusion, but electrode and other losses make this a research neutron-source concept rather than a practical power-producing reactor.
| Component | Role | What The Model Shows |
|---|---|---|
| Vacuum vessel | Isolate the fuel from air and carry ports. | Segmented metallic outer wall, flanges and cutaway. |
| First wall | Protect the vessel from plasma-facing heat and particle loads. | Separate inner lining exposed by the cutaway. |
| Coil system | Provide confinement, shaping and position control. | Distinct windings, cases, central solenoid and outer rings where applicable. |
| Divertor / exhaust | Receive outgoing plasma heat and particles and support pumping. | Separate lower targets on the toroidal examples. |
| Heating & fuel ports | Transfer energy or fuel into the chamber. | Radial ports and traveling energy markers. |
| Hohlraum & capsule | Convert laser energy into X-rays and compress fuel. | Open radiation enclosure and nested capsule layers. |
| Liner & electrodes | Deliver the pulse and compress a target. | Separate current terminals and a moving cylindrical liner. |
A future deuterium-tritium power plant would also require neutron shielding, a breeding blanket, coolant loops, heat exchangers, fuel processing and an electricity-generation system. These are not implemented as a full plant model here. Neither Wendelstein 7-X nor the generic exhibit should be read as a working electricity-generating plant.
The Quest view requests immersive-ar with a local-floor reference space and optional hand tracking, using a transparent WebGL background. See Meta's mixed-reality documentation. Physical room composition, controller tracking and hand tracking need verification on a headset.