Every word on screen
These are the visual captions and annotations in the 9m45s film. Narration is off by default; the optional spoken script is separate. The film plays at its authored pace.
| Film time | Text or visual |
|---|---|
| 0:00–end | UNIVERSE AGE [CURRENT AGE] BILLION YEARS [ERA TITLE — SEE BELOW] |
| 0:00–end | GAS TEMPERATURE COOLER → HOTTER ≤10⁴ K → ≥10⁸ K GAS COLUMN DENSITY LESS GAS → MORE GAS ≤3.2×10⁴ → ≥5.0×10⁶ M☉ / ckpc² |
| 0:03–0:18 | THE EVOLUTION OF THE UNIVERSE |
| 0:15.2–0:28.5 | SUPERNOVA-DRIVEN WINDS One shared label points to three measured wind regions. The central region appears first; the upper and lower examples fade in from about 0:20, once source measurements are available. |
| 0:30.5–0:39.5 | BLACK HOLE FEEDBACK One shared label points to two black holes with recorded thermal energy injection. The stellar-feedback label is absent during this sequence. |
| 0:50–0:59 | GALAXIES One shared label points to two catalogued galaxies using tracked stars. |
| 1:06–1:25 | EVERY SECOND IS 23.5 MILLION YEARS OF EVOLUTION |
| 2:10–2:25 | [1.15–1.21] Mpc PHYSICAL SEPARATION ≈ [3.8–3.9] MILLION LIGHT-YEARS |
| 9:10–9:25 | ILLUSTRISTNG · TNG100-2 COSMIC STRUCTURE AND GALAXY FORMATION BY THE ILLUSTRISTNG COLLABORATION MAX PLANCK · HITS · HARVARD · MIT · FLATIRON |
| 9:28–9:42 | TNG100-2 · 2,688 CPU CORES TNG PROJECT COMPUTING: HAZEL HEN / HLRS · MPCDF PILLEPICH ET AL. · MNRAS 475, 648–675 (2018) Smaller closing credits; the project's computing facilities are acknowledged separately from the run-specific core count. |
General captions fade in over 1.5 seconds and fade out for 1.5 seconds after the listed end. Markers fade inside their listed windows and never persist beyond measured epochs. Pausing preserves the current frame and annotation. The pace caption follows the early feedback sequences so they have room to breathe.
The age clock, era title, and calibrated colorbars are available from the beginning when controls are hidden. The “Text, markers & scientific keys” switch turns these off. Markers and their labels stay attached to the simulation as you look around; labels do not turn or tilt with your head. General captions sit about 18° below the initial reading direction in 360°. Captions, the age clock and corner keys stay in place until you choose “Place text here” in the immersive controls. The same controls show the decoded video dimensions and let you choose 8K detail or lighter 4K playback. Playback speed is fixed.
The era beside the age clock
These are broad chapters for context, with approximate editorial boundaries. They are not sharp transitions or claims that the film has detected the first object in the universe. The first stars and reionization are not individually resolved here.
| Cosmic age | Era title |
|---|---|
| Before 0.2 billion years | THE DARK AGES |
| 0.2–0.4 billion years | THE FIRST STARS FORM |
| 0.4–1 billion years | THE FIRST GALAXIES FORM |
| 1–3 billion years | THE COSMIC WEB GROWS |
| 3–5 billion years | COSMIC STAR FORMATION PEAKS |
| 5–8 billion years | GALAXIES AND CLUSTERS GROW |
| 8–13.75 billion years | AN ACCELERATING UNIVERSE |
| 13.75 billion years onward | THE PRESENT-DAY UNIVERSE |
Context: NASA’s cosmic history and Madau & Dickinson’s cosmic star-formation history. The latter places the global peak near 3.5 billion years. These chapters describe the wider universe, rather than measuring the star-formation rate of this particular subvolume.
Early feedback
The first circle locates model wind particles at about 15 seconds, when the universe is approximately 0.4 billion years old. TNG uses these to represent collective supernova feedback. Three early measured epochs contain no black holes anywhere in this saved subvolume. This establishes a stellar-feedback example without calling it the universe’s first supernova.
At 30.5 seconds, the black-hole sequence identifies two separate sources. Each black hole has increasing cumulative thermal feedback energy in three saved outputs, with no change in its progenitor count. The pointers identify energy-injection locations, without attributing every surrounding puff to them.
The circles locate source regions, not expanding shock boundaries or individual explosions. Gas along other depths also contributes to the image. TNG physical model · Wind and feedback-energy field definitions.
The galaxy pointers
The two markers identify galaxies in the TNG100-2 catalogue at snapshot 11: subhalos 2098 and 1222. Their positions follow the mass-weighted centers of the same stellar particles in the film’s saved subvolume. Wind particles are excluded. The film renders the galaxies’ gas; their individual stars and spiral disks are not visible at this scale. The locator circles do not measure galaxy diameters.
What the scale measures
The two end marks identify black holes inside the visible gas concentrations, measured directly in eight TNG outputs. The line denotes their three-dimensional proper separation, including depth, not a universal distance per pixel or the diameter of either galaxy. The number changes from about 1.15 to 1.21 Mpc (3.8–3.9 million light-years) while the marker is visible.
The enclosing simulation cube is 11.07 comoving Mpc wide. The scale line instead uses physical distance at the displayed epoch: comoving separation multiplied by the scale factor. Positions are interpolated between measured outputs, independently of the film's gas reconstruction.
Reading the keys
Both axes are logarithmic and fixed throughout the film. Temperature runs from 10,000 to 100,000,000 kelvin. Gas column density runs from 104.5 to 106.7 solar masses per square comoving kiloparsec. The on-film bars show these endpoints and their units; the column endpoints are rounded. One M☉ is one solar mass. The “c” in ckpc denotes comoving distance.
Hue represents column-density-weighted stored temperature; intensity comes from the integrated gas column. Values outside the endpoints clip, and dense cores become pale. These keys describe the rendering transfer functions; they do not let you infer a unique physical value from a compressed video pixel. Temperatures in star-forming gas are effective model values. These are scientific colors, not visible-light emission.
Simulation credit and history
This film uses TNG100-2 Subbox 0, a densely sampled region of the intermediate-resolution TNG100 run, computed with the AREPO code. It follows dark matter, gas, star formation, chemical enrichment, magnetic fields, and stellar and black-hole feedback. The images show its gas density and temperature.
The collaboration includes Volker Springel, Lars Hernquist, Annalisa Pillepich, Dylan Nelson, Rüdiger Pakmor, Rainer Weinberger, Federico Marinacci, Jill Naiman, Mark Vogelsberger, Shy Genel and Paul Torrey. The original project brought together HITS, the Max Planck Institutes for Astronomy and Astrophysics, Harvard, MIT and the Flatiron Institute's Center for Computational Astrophysics. Team and computing acknowledgements.
The TNG100-2 entry in Pillepich et al. (2018), MNRAS 475, 648–675, Table 1 lists 2,688 CPU cores and approximately 0.6 million core-hours. The core count is not a count of processor sockets. Project computing used Hazel Hen at HLRS through the Gauss Centre for Supercomputing, with additional resources at MPCDF and other partner facilities. The film render is a separate computation.
The flagship TNG100 run finished in August 2016; its presentation papers appeared as preprints in 2017 and in journals in 2018, and the TNG100/TNG300 data became public in December 2018. That project milestone does not establish the precise finishing date of the lower-resolution TNG100-2 run. The 26-month computation completed in April 2019 was TNG50, a different simulation. The 2019 data-release paper is also distinct from the simulation's execution date. Project timeline.
The five TNG100/TNG300 presentation papers are Springel et al. (2018), matter and galaxy clustering; Pillepich et al. (2018), stellar mass in groups and clusters; Marinacci et al. (2018), radio haloes and magnetic fields; Naiman et al. (2018), chemical evolution; and Nelson et al. (2018), galaxy colour bimodality. Data access: Nelson et al. (2019), The IllustrisTNG Simulations: Public Data Release.
Measured positions, source checksums & method · Optional narration transcript