Loading real stars from HYG 4.1
Exact source CSV ↗-> · CC BY-SA 4.0 ↗->
Local contours count intrinsically bright Gaia DR3 stars (absolute G < −2) in a ±200 pc disc slab, smoothed over 125 pc. The G < 10 sample reaches 2.51 kpc ignoring dust; contours fade at that boundary. Dust, astrometric selection and evolved bright stars still bias this sample: this is not a complete map of spiral arms. In the ramp palette, blue → cyan → green → yellow → orange → red means increasing density within each layer, speed for arrows, and distance for shells.
Horizontal layers sample real tracer heights with shared density levels, revealing the thin nearby disc and the outer Cepheid warp without filling gaps.
Disc contours combine 2,168 Cepheids, 198 unique maser regions and 16,500 Gaia tracers after cross-matching. Sample 95th-percentile reaches are 16.9, 10.5 and 2.4 kpc respectively. Sources: Skowron et al. (2019), Science 365, 478 classical Cepheids (mid-infrared period–luminosity distances and individual RMS errors), Reid et al. (2019), ApJ 885, 131 VLBI masers (inverse parallaxes, propagated 1σ errors), and the local Gaia DR3 luminous sample. VizieR/CDS provides the published tables. Catalogue redistribution licences are not explicit in those tables; consult their authors and CDS terms. Gaia derivatives retain CC BY-NC 3.0 IGO.
In the ramp palette, blue → red shows increasing weighted tracer density, not mass density. Each far family has equal total weight; Gaia has 0.03 total weight to avoid overwhelming the small local area. Horizontal Gaussian smoothing is 700 pc for far tracers, 400 pc for Gaia; vertical smoothing is 120 and 75 pc respectively, broadened by vertical distance error. The 21 slices span ±1.5 kpc at 150 pc spacing and share eighteen density levels. Layers off restores the original flat map. Different populations and survey selection prevent a complete or unique arm reconstruction.
By default, blue contours fade with distance above or below the cursor height. With depth shading disabled, plate palettes encode density through brightness; crosses replace dotted evidence at sampled low-support grid cells, without filling unmeasured gaps. Solid lines have stronger support; dashes, dots and disappearance mark falling counts and increasing distance error. This is a heuristic support score, not a completeness probability. Small neutral dots mark far tracers; faint sightline streaks span ±1σ (linear error propagation for masers; near endpoints clipped at the Sun). Rings show each sample’s 95th-percentile distance, not completeness. The dotted low-support boundary marks unmapped areas. Dust, the bulge, distance errors and survey footprints leave real gaps; darkness does not establish an absence of stars.
The side view exaggerates measured heights by the labelled scale factor.
Tracks show straight-line backwards extrapolation over the labelled time, ignoring the galactic orbit.
Gaia DR3 proper motions and radial velocities (ESA/Gaia/DPAC); HYG proper motions and radial velocities (David Nash / Astronexus). Arrows show km/s scaled by a time interval, not propagated catalogue positions. Unknown RV uses zero and is marked as a tangential estimate in the selected speed. Default LSR velocities add the Sun’s peculiar motion (U,V,W = 11.1,12.24,7.25 km/s; Schönrich et al. 2010) to heliocentric velocities. The Star motion panel can switch to the Sun frame.
Gaia DR3 and HYG stars are mapped observations. The soft stars in the “Modelled” layer are deterministic samples of a published-model-inspired Milky Way: discs, an old warm bar/bulge, and a nuclear disc and cluster around Sgr A*. They are not observed objects, never appear in search or star facts, and can be hidden with M (or ?model=0). Resolved points keep one star’s core. Every sample carries the expected number of stars in its population; unresolved stars and the population omitted by the drawing budget contribute soft combined light in the same layer. Local budgets follow model density, and broad overlapping coverage shares light between levels without distance shells. Counts are samples, not a census. Tune → Modelled stars → Far Glow and Exposure control this integrated light. Local sectors replace the distant sample as you fly closer.
The analytic approximation follows Bland-Hawthorn & Gerhard (2016), Jurić et al. (2008), Wegg & Gerhard (2013), Wegg, Gerhard & Portail (2015), Portail et al. (2017), Launhardt et al. (2002), Sormani et al. (2022), Schödel et al. (2014), and Reid et al. (2019) arm fits. Compact illustrative luminosity functions follow Besançon-style populations (Robin et al. 2003), including an old red clump at absolute G ≈ 0.5, BP−RP ≈ 1.2. These are new generated samples, not redistributed catalogue data or the authors’ simulation products.
To avoid filling the bright catalogue domain, a generated star is kept only when its apparent G from Sol is at least the shipped Gaia cutoff (G = 10). This includes a simple sightline extinction of 0.7 mag/kpc in the plane, falling with a 100 pc height scale and capped at 12 mag. This brightness exclusion is not a catalogue cross-match or a claim of survey completeness; HYG and Gaia selection effects remain. Model rendering applies the same approximate extinction from the camera. No dust volume is rendered. The existing 8,200 pc centre distance and 20 pc solar height are retained rather than 8,178 pc (GRAVITY 2019) and 20.8 pc (Bennett & Bovy 2019).
The dust clouds and dark lanes come from a measured 3D dust map: Vergely, Lallement & Cox (2022), A&A 664, A174, built by inverting Gaia and 2MASS extinctions toward millions of stars (EXPLORE G-Tomo data). It covers a box 10 × 10 kpc wide and 0.8 kpc tall centred on the Sun, with 50 pc detail, shown here averaged to 40 pc cells across the plane. Only cells whose density is at least twice its stated error are drawn; weaker cells fade out. Dust both glows faintly and dims stars behind it, by a tunable amount.
Nothing is drawn beyond the box. Its dashed outline marks where the dust map stops: outside it, dust is unmapped here, not absent. Within the box, the map is smoother far from the Sun, where fewer stars constrain it. The optional distance haze and core glow are a smooth model, not data, and are off by default. The Zenodo record states no redistribution licence; consult the authors before reuse. Add ?volume=0 to the address to see the atlas without dust.
The optional model contours and particle backdrop are illustrative, off by default, and remain available in the dev panel. The centre lies 8.2 kpc toward galactic longitude 0°, with Sol 20 pc above its plane. Default disc radius: 15 kpc; full bar length: 4.5 kpc at 28°; arm pitch: 12°. Perseus and Scutum–Centaurus are stronger arms; Sagittarius and Norma are weaker. The Local Arm passes near Sol. Arm phases, dust, scatter and colours are illustrative.
Structure references: Bland-Hawthorn & Gerhard (2016), Portail et al. (2017), Vallée (2015).
One scene unit = one parsec. +X points to the galactic centre, +Y to galactic north, +Z to longitude 270°. Ring radii: 10, 100, 1,000 light years. Brightness is a tunable display of inverse-square visual flux, without extinction. Twinkle is an optional artistic effect.