New ways to explore space, corrections that make the science clearer, and notes about how the atlas keeps improving. These are verified release milestones, not a list of every code change.
For learners
Start by exploring.
Each release links directly to a lesson or interactive model. You don't need to understand software versions to use the new activities.
For educators
See what has changed in the explanation.
Review the teaching note and scientific limitations before reusing a model or a number in class.
For maintainers
Trace the release to the code.
Each entry links to its recorded commits. These links identify changes, not a claim that each commit was a separately numbered release.
Release history
Improvements in context
Newest first. Grouped around meaningful changes to the experience, not individual refactoring commits.
Mobile display resilience and reader control
Recover an oversized ad after changing browser width
When a previously loaded desktop-size Google display ad exceeds a newly narrowed window, Space Atlas now detects the actual overflow and presents a clearly labeled choice to reload the page at the new width or continue reading. No ad creative or frame is changed by the site.
For learners
Lesson content and the original model remain usable. A notice appears only when an already-loaded display ad no longer fits after the browser window narrows. Reloading restores the mobile-size layout but restarts lesson controls.
For educators
The isolated, ad-free local classroom preview is unchanged. Supervised real learner research remains not conducted; the four classroom browser tasks are technical checks and not student comprehension scores.
Actual public Chromium reproduced desktop1280→mobile390 with a served Google publisher ad stuck at970px and document width680; code release bb1a0ab displayed the accessible recovery notice. Clicking the explicit reload button returned document width390 and ad width320 with no notice. Initial320 loaded with no warning. The code does not hide, resize, or auto-refresh Google ad creatives or change account settings. Regression gate CI verified alongside 14 Python and 32 Node checks.
Classroom readiness and evidence-based visual controls
Moon eclipse controls and isolated classroom-model preview
The eclipse teaching model now opens with a tilted lunar orbit, making node-alignment controls meaningfully change the side-on view from first interaction. A separate local teacher-supervised preview removes Google Analytics and AdSense scripts and serves the existing 3D engine locally.
For learners
In the Moon Eclipses model, compare a New Moon at misaligned nodes with an aligned node. Moon phase prediction, transit brightness and galaxy viewing-angle controls continue to support manual scientific comparisons.
For educators
The separately built classroom preview is a local, noindex research-preparation tool, not the publicly indexed site; it does not include Google ad or analytics calls. Four automated browser interaction tasks passed at 320, 390 and 1280 pixels and with reduced motion, but no real school, educator-supervised pupil, guardian consent, or measured learning study has taken place.
Added CLASSROOM_PREVIEW and strict output isolation to build.py, local MIT-licensed Three.js 0.179.1 importmap, disabled GA/AdSense in preview, script-less student data collection, classroom-build TDD, eclipse default tilt regression and actual Chromium 4/4 task-flow probes. Public first-load AdSense units fit 320/390/600/800px; after a 1280-to-390px browser resize a loaded Google 970px iframe persists, and reloading at 390px restores a 320px unit. Provider account settings and actual pupil studies were not changed.
Mobile ad placement and SVG dimension visibility improvements
Corrected publisher-controlled AdSense slot sizes to fit 320–1280px screens without removing the ads or their identifiers. Extended vector-path browser checks across 16 lessons and made a previously hidden minimum gravity-radius guide visible in its teaching diagram.
For learners
The gravity radius model now shows a readable gold dimension guide even for the smallest modeled body. Educational diagrams and existing interaction models remain available at their original URLs.
For educators
The filled-ad layout was simulated under local and live browser conditions; a real Google creative was unfilled in the available test session. The independent Google Auto Ads overlay and any historical provider-origin horizontal overflow need separate account-side review. Grade 4–8 student studies are not claimed.
Maintained the exact existing AdSense unit IDs and Google loader. Used Google-approved CSS size breakpoints; all nine publisher slot viewport checks kept the unit inside screen bounds. Audited 16 SVG-bearing routes for nontext stroke visibility and intentional paint-order crossings; gravity radius guide went from 5/9 sampled points covered to 0/9. Added two mandatory regression gates to deployment and health workflows and recorded scope-limited evidence.
Five more visual lessons: tides, light travel, star pathways, light energy, 3D galaxies
Added new science-driven visual interaction to all five remaining backlog lessons: tidal forcing contribution comparison, a bounded light-travel demonstration, stellar stage walkthrough, wavelength-versus-photon-energy reference comparison, and an opt-in spatial 3D spiral galaxy with 2D fallback.
For learners
Move the tidal-force model from aligned (100%) to perpendicular (33%) and watch Moon/Sun contribution bars; play a normalized photon journey with linked elapsed time; step through a selected stellar path; save a wavelength and compare actual photon energy ratios; rotate a synthetic 3D spiral disk from face-on toward edge-on.
For educators
All model caveats remain on the page. The tidal bars are an idealized lunar/solar quadrupole with no coastal water heights; light-path motion and stellar steps do not represent real time; the spectral ratios follow physical wavelength equations; galaxy stars are generated conceptual coordinates, not observational reconstruction.
Pure tide resultant invariants and UI state gates; optional bounded progress controls with Pause, tab visibility and reduced-motion settings; lazy existing Three.js galaxy scene with fallback on unavailable WebGL; measured browser interactions and screenshots at 320,390,1280. Site-wide browser SVG audit checked 16 source routes, 27 visible SVGs and 68 text nodes at 320/1280 with no defined text clipping/occlusion flags. 13 Python +27 Node regression suite passed before release; provider AdSense mobile overflow remains independent.
Interactive science diagrams and explanatory motion
Four more lesson improvements: seasons, lunar face, gravity arrows and transit motion
Four previously under-connected model controls now update visible evidence: day length and solar noon altitude for north/south seasons; Moon surface-marker Earth-facing direction; Newtonian trajectory prediction and velocity/gravity vectors; and opt-in transit replay tied to the light curve. Existing already interactive lessons were independently checked rather than rewritten.
For learners
At 40° latitude, compare June, December and zero tilt; compare locked versus faster-spinning Moon markers; launch a cannonball with directional vectors; play one 12-second transit while watching the star brightness and graph cursor.
For educators
The seasons model assumes a spherical Earth, no atmospheric refraction and a fixed latitude of 40°. Gravity arrows show direction and are normalized for visibility; orbital speeds use model units rather than km/s. Transit animation is an illustration, not a measured planet orbital period. None are student outcome measurements.
Physical equations isolated in three pure modules; four RED/GREEN regression gates added to Pages and site-health. Actual Chromium desktop and mobile exercised model changes, 12s finite transit animation/reduced motion, and existing Tides, Light-Year, Stars, Spectrum, Galaxies functionality without duplicate feature code. All 13 Python and 26 Node site checks passed, exact deployment SHA live verified.
Moon eclipses: move the lunar orbit nodes and watch alignment or a near miss
A fixed orbital-tilt toggle could show only missed eclipses. Students can now rotate the line of nodes, compare a modeled alignment or miss at new/full Moon, and see synchronized 3D tilted orbit and side-on evidence.
For learners
Turn on the illustrated 15° lunar orbit tilt (real tilt ~5.1°), change node orientation from 90° to 0°, and watch the side-view Moon offset change from about 0.26 to 0.00 model orbital radii. A short orbit animation is optional and respects reduced motion.
For educators
An orbital-plane match means eclipse alignment is possible, not guaranteed. The model does not compute real eclipse dates, shadow cones or apparent-size effects. Mobile has a separate legible SVG, and no-JS static explanation remains.
Pure circular orbit/node-axis rotation model, relative offset, no perpetual Three.js rendering, 12-second opt-in cycle and TDD. Full 13 Python +22 Node site regression passed; actual live phone-width controls and NASA-backed caveats verified.
Black holes: compare viewpoints, mass and the apparent shadow in 3D
Replaced a static-only anatomy presentation with an opt-in Three.js classroom model of an accretion disk, viewing-angle-dependent projection, and Schwarzschild mass/radius comparison. The original labeled 2D diagram remains available without WebGL.
For learners
Change the viewing elevation between near edge-on and near overhead, inspect the projected disk ratio, compare one to one billion solar masses, toggle an illustrative horizon-size guide, and start/stop a short conceptual gas-motion cycle.
For educators
The glowing lensing arcs and disk texture are labeled educational illustrations, not GR photon ray tracing or Event Horizon Telescope photography; the event horizon and black-hole shadow remain distinct. Real measurements use the sourced nonrotating mass formula.
One pure physics model, lazy reuse of existing Three.js, finite opt-in 10-second motion, reduced-motion, WebGL failure fallback, responsive controls, scientific TDD and SEO/JSON-LD regression. Browser verified desktop and mobile, full site CI.
The 90° corner in the lamp-and-ball classroom diagram is visible again
The original 90-degree marker was hidden behind the opaque Earth observer circle, despite the caption describing perpendicular viewing directions. The desktop, mobile and printable teacher diagrams now mark the same observer angle with visible arcs outside the circle.
For learners
Look for the blue 90° arc outside the Earth observer. It shows the two viewing directions to the lamp and Moon ball forming a right angle, while the gold arrow still shows light travelling from the lamp to the ball.
For educators
Both web diagrams and the separate device-free paper activity now show the angle clearly. A geometry test checks the observer disk, the arc's distance from it, its layering and labelled angle; the original lamp/ball science caveat remains.
TDD RED on missing external markers, GREEN for desktop/mobile/print. Real Chromium rendered hit tests at 320, 390, 800, 1280 and in offline handout show the visible arc is the foreground element. The broader 35-SVG source survey is triage, not a site-wide no-defects claim. Existing Moon CER and SEO tests pass.
Your Moon-phase prediction is now separate from the real observation
A reader reported that all three answer choices appeared to produce the same Moon simulator output. First Quarter is physically 50% illuminated regardless of the guess, but the interface had not distinguished predictions from actual observations. It now shows both results side by side and explains any mismatch.
For learners
Selecting 25%, 50%, or 100% immediately displays your chosen prediction. Compare to reveal the scientifically correct First Quarter (50%) beside it; choosing another answer clears the earlier comparison. The model is never altered to match a wrong prediction.
For educators
Preserves the Claim–Evidence–Reasoning exercise and the device-free printable handout. Students can discuss why the scientific model stays at 50% but their own 25/50/100 guesses differ. No pupil results are stored, and no classroom outcome is claimed.
Clarified hypothesis versus physical model with native radio choices, selected-vector previews, hidden-before-check observation visual, choice-sensitive explanation and state reset. Added a new focused TDD regression to production and scheduled CI. Browser-verified all three options at desktop and mobile widths; preserved static no-JS answer, SEO and NASA sources.
Scientific accuracy and controlled visual learning
Moon phases: follow sunlight, drag the orbit, and see why the bright side changes
A review of the Moon phase simulator found a hidden Sun, missing Earth-to-Moon sightline, an unrotated Moon surface and misleading phase names. The model now connects the viewing geometry, correct phase label, illuminated disk and textured lunar orientation; the classroom lamp diagram now directs its light toward the ball.
For learners
Tap or drag the Moon along its orbit or select any of eight phases; watch the Earth sightline, visible illuminated area, Sun–Earth–Moon angle and day update together. Play an explicitly initiated, pausable 24-second illustrative model cycle, or use still native controls with reduced-motion preferences. A bright 67% disk is now called waxing gibbous, not First Quarter.
For educators
The first-quarter CER activity remains device-free as a two-page printable. The revised classroom top-down geometry distinguishes the actual lamp-to-ball light ray from the observer's perpendicular gaze lines, has a legible mobile diagram and explains why a nearby lamp approximates ideal parallel sunlight.
Single pure testable phase-geometry model with nine phase invariants, idealized tidal orientation and direct orbit inverse; idle WebGL no longer continuously spins Earth. Local browser tested drag, play/pause, mobile layout and reduced motion. Existing NASA citations, canonical/JSON-LD, ad boundaries and independent astronomy-educator CER evidence retained. This is not measured pupil learning or a real ephemeris.
Test a no-tilt Earth directly: predict, compare, explain
A science educator-editor suggested testing the zero-tilt counterfactual directly with a globe. The already-existing 0° simulator control now has an explicit teacher-led prediction-and-evidence sequence and a new printable classroom worksheet; the older toggle is not attributed to that suggestion.
For learners
Predict what happens to opposing hemisphere sunlight patterns with no axial tilt. Compare June and December at 23.5° and 0° and explain which model variable matters, without implying the model measures weather or climate.
For educators
A suggested 12–15 minute lesson can run with the existing simulator or with a cool lamp, a foam globe, and two marked orbit positions. A device-free two-page printable offers classroom setup, safety notes and a student prediction/evidence/reasoning table; no pupil outcome is claimed.
Added an accessible no-JS native details disclosure in the existing Earth Seasons teacher-use area; used scoped anonymous acknowledgment data and no new UI framework or JS bundle. Retains canonical, structured data, original model, ads and existing experimental teacher-share lane.
A classroom-ready Moon phase investigation informed by educator feedback
Feedback from elementary and astronomy educators led to a three-choice Moon prediction activity, a clearer lamp-and-ball demonstration, an offline two-page CER lesson, and a compact acknowledgement of practical improvements.
For learners
Choose whether First Quarter should appear one-quarter, one-half or fully bright, then check your prediction in the existing 3D Moon model. The explanation remains available without JavaScript.
For educators
A suggested 30–35 minute Grades 3–5 activity offers objectives, safe materials, a reproducible original setup diagram, student prompts, example reasoning, differentiation, and a device-free printable guide. Neither adoption nor classroom effectiveness has been measured.
Native radio inputs and a no-JS answer connect to the original phase model. Scoped static acknowledgments preserve private contributor identities until confirmed opt-in; the disclosure shows at most three improvements and leaves the lesson layout uncluttered. CI verifies privacy, bounded rendering, static assets, SEO, and offline print compatibility.
Astronomy-educator feedback identified an eclipse-like Moon-phase graphic and closely packed, faint solar-orbit paths. Both views now use more accurate geometry and more readable presentation while keeping the real-distance lessons separate.
For learners
Move the Moon and see right-side waxing crescents, half-illuminated quarters and left-side waning phases formed by a spherical day/night boundary, not a moving Earth shadow. In the home Solar System model, clearer orbit rings and more separation prevent oversized gas giants from colliding with nearby orbital tracks.
For educators
Compare new, first quarter and full Moon to challenge the misconception that the monthly phases are eclipses. The Sun-Earth-Moon diagram still compresses distance and flattens inclination. Solar System ring spacings remain illustrative scene units, not astronomical units or equal-size planets; use the actual distance-scale page for physical AU.
Replaced translated overlapping Moon disks with a lit surface-normal renderer; separated Saturn/Uranus/Neptune model radii and reconciled the 50-unit distance-scene static marker, made orbital paths visible and extended camera framing. Added physically targeted nine-phase/orbit-annulus CI, headless Chrome at 320/390/1280px, all prior quality gates and evidence audit. No change to existing AdSense, canonical paths or JSON-LD.
The universe-scale and light-year lessons now use actual quantitative diagrams, not text-only selection controls. The cosmic ruler maps eight distances or dimensions across logarithmic orders of magnitude; the light-year lesson locates eight sources on its own log distance axis and lets readers scrub a light signal from emission to Earth.
For learners
On the universe ruler, select Earth, the Moon or the Milky Way to see their true approximate order of magnitude and how many tenfold jumps separate stages. On Light-Year, choose a source and move a 0–100% photon-journey slider to see how much travel time has elapsed and remains.
For educators
Teach that a logarithmic scale can faithfully show enormous ratios, that different reference measures are not all diameters, and that a normalized photon-path bar is different from a ruler comparing multiple sources. The scale-jump quiz was corrected: 4,500 ÷ 30 is roughly 150, or hundreds of times.
Implemented dependency-free native SVG logarithmic axes and state-driven selectors, a source-to-observer time scrubber, static no-JS explanatory visual fallbacks, reduced-motion-safe manual controls and isolated mobile/desktop testing; added ci/verify_cosmic_visuals.mjs to both deployment and scheduled read-only Health. Preserved all existing indexed URLs, sources, teacher experiment and AdSense positions.
Explore why planet mass and size both affect gravity
The planet-weight comparison now includes a hands-on experiment: change mass and radius separately and see the gravity equation, a symbolic radius circle, and the explanation respond together. Published NASA equatorial gravity ratios were also reconciled with an explicitly labeled idealized model.
For learners
Try four times the mass at the same radius, then twice the radius at the same mass. See why the first makes gravity four times stronger, but the second reduces it to one quarter. Compare a Jupiter-like theoretical model with the real reference value.
For educators
Run a predict–change–explain activity with independent mass/radius controls, live numerical outcomes and an honest schematic. Explain that actual effective equatorial gravity differs from the simplified non-rotating GM/R² relationship, and gas giants lack solid surfaces.
Added a static-first native SVG radius symbol, four physical thought experiments and accessible sliders, corrected NASA effective equatorial data for giant planets in the shared fact source, kept scientific evidence links, and added independent ci/verify_gravity_mechanism.mjs to deploy and read-only Health. Verified 320/390/1280 responsive states and blocked-JavaScript fallback.
Planetary illustrations use accurate diameter ratios. Astronomy units, transit-candidate limits and the orbital teaching model now disclose their definitions and approximations. The image-evidence guide prioritizes authentic NASA observations.
For learners
Compare planetary diameters on one true scale. Manipulate an illustrative transit with live relative brightness and learn why a recurring dip alone does not prove a planet.
For educators
Use corrected linear/log scale labels, source-backed IAU definitions, false-positive warnings and clear model-only units during lessons.
Source-grounded 28-page learner-needs audit, corrected visible glyph ratios and model claims, and a new automated science-truth gate in build and health workflows.
The three image-reading examples stay closed until selected, and the planetary size comparison works with keyboard or touch. The mobile Search button remains separate from the lesson-navigation row. Moon phase and tide model presets now accurately show which mode is selected. Light-year, universe zoom and planet-size models now show readable reference values before interactive scripts finish loading.
For learners
Select “View an example” to read one explanation; select it again to hide the text. Planet size buttons update the comparison by touch, mouse, or keyboard. Moon phase and spring/neap tide shortcuts now also reflect the selected model state. The Sun light-travel reference, first universe-scale step and Earth–Jupiter comparison remain readable during loading; controls activate only when ready. The Moon remains fully visible through every position in the tidal-locking orbit, including on narrow mobile screens.
For educators
The illustration/observation/model examples now match their visible open state. Keyboard-operable planet controls and an announced fact readout support classroom accessibility. Moon phase and tide preset selection states now remain synchronized with displayed model values and assistive controls. Three interactive lessons now retain factual first-screen content even if JavaScript is unavailable, instead of starting with empty feedback. The tidally locked and hypothetical Moon marker demonstrations now retain the complete orbit in the camera frame at every slider angle.
Removed a competing CSS override and obsolete JavaScript card-toggle behavior, replaced nonsemantic whole-card click targets with native HTML disclosures, converted eight planet controls to buttons with aria-pressed, and added a 28-page UI structural regression gate. Verified against actual mobile and desktop browsers. Follow-up dfbeea2 synchronizes Moon Phases 0/90/180/270° preset aria-pressed with the slider and both Spring/Neap selected states without changing scientific calculations. Follow-up 593010b adds static-first HTML values, disabled-until-ready controls and ci/verify_ready_content.mjs to both Pages deployment and scheduled site health; production and blocked-JavaScript browser flows verified. Follow-up b2ee470 fixes OrbitControls silently clamping the camera inside the lunar orbit, adds aspect-aware camera fitting on resize and 360-degree orbital-frustum regression coverage for eight viewport shapes.
The search control now lives beside the Space Atlas logo instead of interrupting the horizontally scrolling topic links. The second navigation row is smaller and dedicated to browsing lessons.
For learners
Tap the magnifying-glass icon at the top right to search. The row below now contains only lesson navigation links, including All topics.
For educators
The full topic directory and direct lesson links remain available from mobile. The search dialog and keyboard focus return work as before without covering or rearranging the study material.
Following Android screenshot review, separated the utility-owned 42px SVG search trigger from the 44px-high mobile link strip, removed conflicting narrow-screen CSS overrides and emoji icon, preserved search dialog IDs and accessibility, and added a 28-page mobile-header CI gate with 320/390/768/980/1024px+ browser verification.
Topic breadcrumbs now lead to real category sections, mobile readers can find the full atlas directory and search immediately, and related lesson cards recommend distinct next questions instead of repeating the same destination.
For learners
From any lesson, use the linked topic breadcrumb to return to its subject directory, or choose an evidence-based next question in Keep exploring.
For educators
Guide learners between models, measured facts and adjacent concepts using explicit category and recommended-lesson links; the science citations and model limitations remain unchanged.
Repaired eight duplicated Go deeper targets, linked all 26 category breadcrumbs to four real homepage anchors, and exposed the topic directory and search earlier on mobile. Added a deterministic 28-page user-flow graph, canonical/sitemap and fragment validation gate; existing JSON-LD avoids invented indexable category hubs.
The shared footer has been reorganized so its science explanation, learning pathways, project resources and release history stay aligned and readable on phones and larger screens.
For learners
Follow the Solar System, Earth & Moon, and Deep Space links from the bottom of any lesson without hunting through a dense block of notes.
For educators
Model limitations, primary learning sources and project updates are now directly accessible from a labelled and keyboard-navigable footer.
Replaced an unstructured one-line footer and a grid placement rule on a non-grid child with one semantic component. Added a 28-page shared-footer hierarchy and route regression gate, alongside 320/390/1280px browser checks.
Search and motion controls respond to your choices
Keyboard search now closes without unexpected navigation; the solar-system model respects reduced-motion preferences, and shared quizzes provide assistive-technology feedback.
For learners
Search topics with a keyboard and follow your intended selection. Motion-sensitive visitors start with a still 3D solar-system scene and may choose to resume it.
For educators
Keyboard navigation, a reduced-motion default and quiz status announcements make common classroom interactions more predictable. Screen-reader behavior still benefits from device-specific testing.
Resolved global Enter interception, late search-index loading, hidden-tab WebGL frame scheduling, and live-region semantics. Added a dedicated browser-flow/DOM regression gate and a prioritized code-quality audit.
Explore temperature, tides and honest distance scales
New controls reveal signed planetary temperatures, changing lunar–solar alignment, and differences between physical AU distances and compressed model coordinates.
For learners
Compare two planets’ temperatures, vary the Moon’s alignment, or follow a planet across two distance rulers.
For educators
Read surface-vs-atmosphere temperature caveats and explain why the tide-force diagram is not a local tide forecast.
Space Atlas is an evolving educational reference. Some diagrams are scale-accurate, while others are deliberately simplified so their relationships can be seen. Each lesson states what is observed, estimated, reconstructed or illustrated. A correction in this log does not imply that an earlier visualization was scientifically equivalent to the new one.
Historical entries are reconstructed from identifiable changes in the source history. Dates reflect documented changes, not retroactive version numbering. Automated site checks verify public pages and critical files; they are not measures of improved test scores, Google indexing or learning outcomes.
🔍
Type any question, planet, or topic to instantly filter Space Atlas. Press ESC to exit.