Roadmap

What has shipped, in which release, and what is proposed next.

The vibe-sketch/vibe-view scope decision records the agreed division between 2D publication drawing and 3D calculation work, including basic cleanup in both applications and the intended geometry round trip. It is independent of the historical release snapshot below.

Last reviewed 2026-10-24 against main, released tags through v3.0.0, changelog fragments and the canonical development tracker. Update shipped scope when a release is cut, following the release process. The tracker carries live defect status; this page records release history and proposals.

Issue numbers. Bare #N references below identify the canonical development tracker. The public issue tracker uses a separate numbering namespace. Historical monorepo issue numbers are identified explicitly.

Future releases are proposed, not scheduled. Registered codenames commit to nothing about when, or whether, a release is cut, and a name does not bind the scope. v2.19.0 and v2.20.0 carried the names Phong’s Pheasant and Levoy’s Lemur but shipped the #34 audit follow-ups. The lighting proposal’s publication mode then shipped in v2.20.2, and the first version of direct volume rendering shipped in v2.20.3; both are recorded under Proposals alongside the remaining work. The maintainer decides the scope of the next release at cut time.

Where things stand

Latest release

v3.0.0 Levinthal’s Lynx, 2026-10-23; the existing released baseline

Next release

3.0.x: focused user-facing regressions and their missing acceptance, through the normal release owner. No date is set

Current defects

Consult the canonical development tracker; public reports are tracked separately

Published installers and documentation

Public macOS Apple Silicon and Linux x86-64 desktop downloads and a source archive exist for 3.0.0. Public-source mirror and any remaining deployment closeout are tracked separately under #110; see Desktop distribution

Current development and the next release

The delivered feature base includes direct volume rendering, publication figures, playback filmstrips, ORCA result inspection, calculation preparation and queue submission, and detailed 3D editing. These are existing capabilities, not work that must be reimplemented for a new release.

3.0.0 is released. Its immutable tag, passing candidate CI and public desktop downloads already exist. All twenty work packages of its recorded scope (#111–#114 and #120–#135) are accepted. Remaining #110 publication closeout reuses that evidence; it does not recut the release or reopen accepted work packages.

Development targets 3.0.x corrections. Prioritize user-facing regressions and complete their applicable acceptance before delivery through the normal release process. Keep Python 3.12 as the minimum, the section-renderer extension contract compatible throughout 3.x, and the documented platform limitations. Broader editor, representation and Sketch round-trip enhancements remain proposals for later compatible releases unless they demonstrably block Sketch 0.6; they are not new 3.0.0 gates.

The 2.21 milestone was closed without a release: its one item, excited-state tables (#109), shipped in v2.20.19. 2.x fixes are cut on demand, decided 2026-10-23: only security or data-correctness fixes qualify. Each lands on main first and is cherry-picked to a maint/2.20 branch made from the v2.20.20 tag when the first such fix is needed, then tagged as a 2.20.x patch.

Released in v2.20.19 and v2.20.20:

  • Excited-state tables in the CLI, terminal and browser (#109); linked NTO and density identifiers are shown, but drawing them from the table is not included. Transition densities are a supported signed volume (#115) and open with difference-density defaults (#119).

  • The first 3.0 work packages, additive and compatible with 2.x: client-side label followers (#111), the client-side silhouette pass (#112), renderer extension API v1 (#113) and the client performance harness with the WebGL renderer string in the About dialog (#114, matrix not yet complete).

  • Export screenshot saves the browser frame with its outlines (#116); browser file opening and downloads work again (#116, #117); normalized spectrum CSV peaks (#118).

  • Self-contained Linux x86-64 desktop packaging. As for macOS, this is packaging support in the release, not proof of a published installer.

Released in v3.0.0:

  • All sixteen result-inspection and parity work packages, #120–#135, are implemented and verified on their issues: spectra and modes from an ORCA output alone (#120), Gaussian/Lorentzian line shapes with ASCII export (#121), mode arrows and GIF/MP4 export (#122), ORCA densities through orca_plot (#123), exact geometry and bond editing (#124), molecular surfaces and interactions (#125, #126), fetched biomolecules, mmCIF and sequence superposition (#127), experimental maps (#128), fchk and Molden orbitals (#129), other programs’ outputs and XYZ trajectories (#130), the molecular builder (#131), the coordinate editor and Z-matrices (#132), input writers for eight more programs (#133), publication figures and morph movies (#134), and the selection language, model scenes and command line (#135).

  • The Python 3.12 floor (#110), and the Linux client performance matrix with tightened candidate limits (#114).

  • Fixes found while integrating them: saved geometry keeps its charge and multiplicity (#137); figure text contrasts with the background (#138); builder and scene inputs start empty (#139); auto_open applies to files opened later (#140); compare explains a missing partner spectrum (#143); parser, renderer and accessibility contracts repaired from the candidate gate (#142, #144).

  • vibe-viewer open --view/--mode/--orbital (#141).

  • The manual: chapters on clusters, supported programs and the vibe-sketcher and vibe-qc workflows.

On main after v3.0.0, for a later 3.0.x release (status 2026-10-24):

  • Animation format follows the output suffix and contradicting explicit formats are refused before rendering (#148).

  • Wrong section identifiers name the available sections (#147, #149).

  • Vibration tables retain per-mode intensities, activities, masses, force constants and symmetry, and join companion spectra (#150); the browser mode picker labels imaginary modes and uses the same IR intensities (#151).

  • vibe-viewer animate starts on the first real vibrational mode and refuses an out-of-range --mode (#152).

  • The qvf 0.6 kinds without a panel report skipped, not yet rendered instead of unsupported, guarded by a test against the vendored schema (#153); volume.laplacian renders as a signed isosurface (#154).

  • Existing Turbomole/xtb coord support (#145) and xtb trajectory logs (#146).

  • Example vq configurations and the Build, inspect and present tutorial with a runnable Python tour and navigation links.

These are source landings; their containing patch release and installed-package acceptance belong to the release owner. They are not part of the 3.0.0 tag.

The maintainer decided on 2026-10-23 that feature parity means capabilities, not native file formats; see Feature parity.

The open tracker also retains verified fixes already included in tags, such as #100–#102 in v2.20.17. Open/closed and release labels alone therefore do not measure unfinished implementation: compare the release notes and source. The release owner qualifies each later candidate and its distribution artifacts under the release process. Accessibility and cross-application proposals are not automatically gates for every intervening release.

Milestones and weekly reconciliation

The 3.0 - interactive viewer milestone records the accepted 3.0.0 scope (#110, work packages #111–#114 and #120–#135). The tag, candidate CI and public desktop delivery are complete; the loop and existing release owner coordinate only the remaining publication and milestone closeout. Accepted work packages stay accepted. 2.21 - bounded feature release is closed. Later regressions are tracked for compatible patches rather than added to the frozen baseline. Issues hold individual work and acceptance evidence; milestones group a release’s selected scope.

Main now targets 3.1 for the next feature release. This permits the planned one-based animate --mode interface correction (#155) alongside compatible post-3.0 work. The published 3.0.0 tag and accepted installers remain immutable; any 3.0.x maintenance release must exclude this interface change. Package version updates and release qualification follow the normal release process.

Review this roadmap every week against main, tags, changelog fragments, linked issues and applicable acceptance evidence. Update stale statements and record implementation, validation and released status separately. When the selected milestone passes acceptance, mark the target ready and advance the next planned release number; update package versions and tags only through the normal release process. Semantic major increments require an agreed contract/scope, not just elapsed time or a paper deadline. Preserve historical snapshots as dated history and explicitly record deferrals.

Shipped

What v2.15.2 already contained

All Avogadro-parity feature work was built in the private monorepo, before the first public commit. Its full evidence and reasoning is in the frozen parity roadmap.

Milestone

Scope

State at v2.15.2

M1 vq cockpit

Job Manager, submit from the viewer, live monitor with log tail, queue overview

Complete

M2 file watching

Content-fingerprint watcher, per-section diff, hot reload that keeps the camera

Complete

M3 live optimization

MSINDO relaxation while editing, MACE engine, frozen atoms

Complete

M4 live results

Streaming checkpoint QVFs, run-status chip, Watch live

Complete

M5 biomolecules

Residues and chains, secondary structure, flat ribbons with sheet arrows, residue selection, colour by chain, secondary structure, residue type and B factor, camera-facing labels

Complete

M6 formats (F)

[ase] extra, 16 import extensions, twelve export formats round-tripped

Complete

M6 builder (E)

SMILES to 3D, Miller slab spacing, periodic bond images, supercell lattice

Partial, see Not yet placed

M6 platform (G)

Electron build targets, Python SDK page, served-viewer accessibility baseline

Partial, see Continuous tracks

Since the split

The first post-split releases made vibe-viewer a standalone product through repairs, a documentation site, release machinery and artwork. v2.17.0 adds TREXIO geometry and molecular-orbital interoperability, v2.19.0 and v2.20.0 work through the #34 viewer audit, and v2.20.2 ships the Publication figure look and volume payload support; v2.20.3 completes the first direct volume rendering workflow. The detailed release notes are in the changelog.

Release

Date

Codename

What shipped

Issues

v2.15.3

2026-09-09

Roothaan’s Roadrunner

First gated tag. QVF drift guard, skip audit, [queue] extra, VTK 9.7 test fix; documentation site and its CI; codename catalogue; release branch and procedure

#1, #2, #3, #5, #6, #7, #11, #12, #13, #14

v2.16.0

2026-09-09

Sayle’s Starling

Product manual with real viewer figures; wordmark, favicon and the seven-image codename series; release extra; trame<4 bound; QTAIM and spectrum-panel fixes

#9, #13, #20

v2.16.1

2026-09-10

inherits

The vibe-qc availability probe degrades instead of raising

#18

v2.16.2

2026-09-12

inherits

Missing-RDKit tests run in every environment; patch checklist; contributor guidance

#20

v2.17.0

2026-09-13

Lorensen’s Loon

TREXIO geometry and molecular Gaussian orbitals; packaging-metadata checks; standalone agent guides

#23

v2.17.1

2026-09-15

inherits

Complex coefficients and Gamma-point lattice tails for MOs; scene replacement that avoids stale viewports; command-palette Escape handling during presentation

—

v2.18.0

2026-09-17

Richardson’s Robin

Biomolecule polish; metadata-preserving slice; Fermi-relative DOS; background and export fixes; energy windows; docs extra; Bash 3.2 compatibility and candidate-only product gate

#19, #21, #22, #24, #25, #26, #27, #29

v2.18.1

2026-09-22

inherits

Full-tree lint and strict source typing; general Miller slab orientations; tutorial and documentation updates; stable real-UHF gradient fixture

#8, #20, #30, #33, #41

v2.19.0

2026-09-25

Phong’s Pheasant

#34 audit consolidation: one QVF writer, one bond rule, one unit-cell module, shared CLI options and headless scene setup; hardened readers and importers; periodic-surface, terminal-chart and archive-merge fixes

#34–#40

v2.20.0

2026-09-29

Levoy’s Lemur

#34 audit follow-ups: reader, renderer, browser, terminal, CLI and converter correctness fixes; display units for orbital energies, band/DOS axes and lengths in browser and terminal; signed volumes keep both lobes along reaction paths

#43–#63

v2.20.1

2026-09-30

inherits

Open-time errors clipped (#64); validate separates layout errors from hash mismatches (#65); malformed spectrum, GTO and MO payloads refused by the reader (#66); batch-compare reports reasons (#67); GRO and PDB first-frame and cell validation (#40); typed browser composition root (#48)

#40, #48, #64–#67

v2.20.2

2026-10-06

inherits

Publication preset as the one-click figure look, with outline and hide-chrome switches and capture --publication (#68); dark-theme accent at WCAG AA (#68); large unsigned volumes open again through the reduced-detail path (#69); Python 3.14 CI lane and trexio extra guard; the direct-volume payload module, step 1 of the volume design

#68, #69

v3.0.0

2026-10-23

Levinthal’s Lynx

Python 3.12 floor; accepted browser, result-inspection and 3D work packages; 3.x section-renderer compatibility. Public macOS Apple Silicon and Linux x86-64 downloads; Windows GPU qualification remains deferred

#110–#114, #120–#135, #137–#144

The subsequent v2.20.3 to v2.20.14 releases added direct volume rendering, filmstrip playback, library thumbnails, broader display units, ORCA result workflows, vibe-sketcher launching, corrected calculation inputs and queue integration, QVF 0.3 compatibility, and 3D dragging, displacement and FIRE relaxation. See the changelog for the release of each change. v2.20.15 adds single and combined normal-mode geometry edits, the ORCA mode-editing tutorial, stricter input handling and the QVF 0.4 reference validator. v2.20.16 fixes CIF block isolation and updates format documentation. v2.20.17 adopts the QVF 0.5 toolkit and fixes long GIF export, linear-molecule symmetry and species-preserving rigid alignment. The QVF archive format remains version 1; toolkit releases are not archive-version changes. The detailed sections below retain the remaining proposals and measured limits.

Correctness fixes shipped in v2.18.0

These corrections are included in the v2.18.0 tag.

Issue

Defect

Released behaviour

#24

slice drops section fields and root metadata

Preserve original metadata and kept bytes; prune removed viewer hints and validate before replacing output

#25

Combined bands/DOS uses one Fermi shift for both panels

Shift bands by their own reference; preserve the DOS grid; separate axes when the bands’ reference is absent

#27

Scene rebuilds discard the light-background choice

Keep the explicit background through rebuilds and reloads

#21

Headless export creates an unawaited reset coroutine

Schedule a callback only when an event loop is running

v2.17.0: Lorensen’s Loon, TREXIO orbitals

The maintainer selected TREXIO interoperability for this release. HDF5 files and text datasets now supply geometry and real molecular Gaussian s/p/d/f orbitals to the existing isosurface workflow. The importer preserves units, normalization, AO ordering and available spin/energy/occupation metadata; the format guide describes its limits.

v2.18.0: Richardson’s Robin, biomolecules

Shipped scope: polish of M5, whose initial biomolecule support shipped in v2.15.2, plus the correctness fixes above.

  • Isolate or hide selected residues without connecting hidden ribbon spans.

  • Select residues by clicking the viewport.

  • Share chain, secondary-structure, residue and B-factor palettes across ribbons, atoms and bonds.

  • Retain PDB HELIX/SHEET ranges and distinguish supplied alpha, pi and 3₁₀ helices and beta bridges. The CA-only geometric assignment stays H/E/C.

  • Draw crisp strand edges and independently shaded end caps; include the final residue in a chain-terminal strand assignment.

  • Show dashed geometric hydrogen-bond contacts using explicit H and N/O candidates in the supplied coordinates.

The biomolecule guide documents the display controls and scientific limits. No new core dependency or QVF schema change is needed.

Minor releases and remaining proposals

The proposals below are not a release commitment. Recheck their current code and tracker evidence before selecting a release scope.

PyChem-Pro and ReShiki assessment (2026-10-30)

This is a source/documentation assessment, not runtime or scientific validation of either tool. Sources are pinned to PyChem-Pro f011960 and ReShiki a115382. The proposals below have no release assignment and do not reopen accepted work.

PyChem-Pro combines molecular viewing, chemistry services and analysis plugins. Its useful concepts are UI-independent service protocols, inspectable numerical methods, linked structural diagnostics and bounded background work. Its feature overview also describes descriptors, chemical-space projections, QSAR splitting and docking. Those claims do not establish scientific accuracy or production readiness. View should extend its existing services and renderer API, not replace Trame/VTK or reimplement a chemistry toolkit to avoid RDKit.

ReShiki centres on editable chemical drawings and publication figures. Its overview describes templates, cleanup, editable interchange, offline chemistry and optional agent control. The transferable View concepts are atom-linked result inspection, explicit recovery choices and preservation of model identity. Its crystal workflow also distinguishes display copies from editable atoms and preserves imported geometry. Keep that distinction in View’s existing periodic representations; do not copy distance-inferred bond orders into an authoritative chemical graph.

Capability or idea

Existing View base / disposition

Owner and next step

Surfaces, contacts, representations, high-resolution figures

#125, #126, #134 and publication presets already cover the core workflows

View; reuse existing rendering and export paths

Structure building, cleanup and conformers

#124, #131, #132, optional RDKit and live FIRE/L-BFGS already exist

Both apps retain basic preparation; share reusable logic without importing each other’s UI

Sequence alignment and rigid superposition

#100/#127 already fit structures; current sequence fit explicitly performs no structural outlier rejection

View #177: optional robust fitting with auditable correspondence and both retained/all-pair RMSDs

Ramachandran diagnostics

Residue selection, peptide building and trajectories exist; a linked phi/psi plot is missing

View #178: residue/frame-linked torsion inspection

NMR interaction

spectra.nmr already shows metadata, shifts, couplings and raw shielding tensors

View #179: linked calculated-result table/sticks/3D atoms; Sketch retains drawing annotations and prediction workflows

Recovery

View sessions already save/autosave appearance; their payload is not a complete unsaved-geometry snapshot

View #180: recover selected geometry drafts while protecting live owners and original sources

Chemical-space maps and similarity

#163 is accepted on main: producer fingerprints, projection, neighbours and channel comparisons

Reuse #163 and its guide; no duplicate projection or descriptor engine

Selection commands, SDK and plugins

#113/#135 and the Python SDK already provide extension and automation foundations

Extend only for a concrete consumer; no second event bus, agent stack or renderer rewrite

2D templates, schemes, cleanup previews, editable clipboard and page composition

Specialized drawing/publication work under the scope decision

Sketch roadmap; View supplies structures/results and existing figure exports

QSAR training/splitting and docking search

Broader modelling workflows, not required for detailed result inspection

Defer; no new engine, dependency or native program reader proposed

Feature requests and acceptance boundaries:

  • #177 — robust protein superposition. Inspired by PyChem-Pro’s alignment implementation. Preserve the existing all-pair default. An opt-in bounded fit must report rejected pairs, thresholds, convergence and both RMSDs under the final transform, refusing underdetermined fits. Verify known transforms with displaced residues and preserve both original structures.

  • #178 — linked Ramachandran diagnostics. Inspired by PyChem-Pro’s plugin. Link plot, residue selection and trajectory frame; export angles and identity. Missing atoms, chain breaks and insertion codes must be handled explicitly. Angle-only inspection can ship before reference contours: classification requires a documented, licensed reference dataset and published method, not copied KDE data or adjusted thresholds presented as validated quality scores.

  • #179 — atom-linked calculated NMR inspection. Inspired by ReShiki’s NMR interaction. Add decreasing-ppm sticks and two-way table/atom selection, disambiguate overlapping shifts and retain producer provenance. Equal-height sticks show positions, not integrations or simulated multiplets. Never invent atom mappings, shift references, uncertainty or missing coupling data. This is inspection of supplied results, not a new empirical prediction engine.

  • #180 — unsaved geometry recovery. Inspired by ReShiki’s recovery manager. Extend existing sessions with atomic geometry snapshots and selective recovery into separate work. Verify interrupted writes, concurrent owners, missing/changed sources and failed restore/discard. Keep graph, state and origin metadata; recovering a draft must not automatically send stale edits back to a sketch.

Coordination and reuse. The parallel vibe-sketcher assessment owns drawing cleanup previews (Sketch #162), publication pages (#163), command discovery (#164), reviewable agent edits (#165), batch property tables (#166) and chemistry-aware drawing selection (#167); all six numbers in this sentence refer to the Sketch tracker. Those are distinct from the four View requests above. Sketch recovery qualification remains under Sketch #69; View #180 addresses the separate geometry snapshot gap. Reuse View #136/#141 and Sketch #67/#88 for geometry return and result selection; #136 has implemented View support and paired acceptance, so this assessment creates no replacement exchange issue. Descriptor/prediction producers retain ownership of numerical definitions; View consumes their values and records method/version/units. QVF owns any needed format extension, released and vendored through the normal pinning path. Shared numerical contracts are desirable; a new shared package is not assumed to exist or required as a preliminary migration.

Reuse constraints. PyChem-Pro’s pinned license is Polyform Noncommercial 1.0.0. These proposals use workflow ideas only and do not import its code, force-field parameters or datasets. ReShiki’s original work is MIT OR Apache-2.0, with separately licensed components and data listed in its license scope and notices. No upstream code/data, new dependency or license choice is adopted here.

Materials and lighting

Originally proposed for v2.19.0, Phong’s Pheasant; not shipped there.

Today: material presets, Publication, Presentation and Analysis representation presets, toon shading, OSPRay ray tracing, and light and dark themes.

Proposed:

  • Publication mode: white background, thick outlines, no gizmos, one click. Shipped in v2.20.2 (#68). The Publication preset outlines every atom and bond with an inverted hull, hides the viewport chrome, remembers the look it replaced so clearing the picker restores it, and vibe-viewer capture --publication renders the same look headlessly.

  • A theme and accent pass. Partly shipped in v2.20.2 (#68): the dark theme’s accent now meets WCAG AA as text. A wider pass over contrast and focus stays open under Accessibility.

Constraint: ambient occlusion and fog are VTK render passes, and render passes never reach the vtk.js client (test_ambient_occlusion_is_server_side_only). Silhouette outlines depend on the view, so an outline computed on the server goes stale when the client rotates. The shipped outline avoids this by being geometry: a slightly larger, front-culled black copy of each actor (Raskar and Cohen, 1999), correct from any camera. The true silhouette pass for isosurfaces and ribbons runs in the client since #112, below.

Transfer functions and volume rendering

Originally proposed for v2.20.0, Levoy’s Lemur; not shipped there.

Today: the first version shipped in v2.20.3. Render as → Volume in the volume controls draws a stored section through a transfer function: a uint8 grid capped at 128 voxels along its largest axis, the section’s colormap, a preset opacity ramp edited through a window and six sliders, per-section persistence through bookmarks, sessions and the render and transfer_function keys of viewer_defaults, and vibe-viewer capture --render volume for the same look headlessly, with capture and batch following a section’s render hint by default. A skewed grid is resampled onto its bounding box, since vtk.js places a volume axis-aligned whatever direction matrix its image carries (checked in Chrome on 2026-10-06), and the opacity curve is edited node by node in a page widget with pointer and keyboard. v2.20.2 itself carries only the payload module. The isosurface stays the default.

Specified in QVF. The volume-mode render and transfer_function hints are documented in the current QVF section-hint contract. They are advisory hints; malformed values do not invalidate an archive. The original request was filed as qvf#5 on 2026-10-06.

Measured 2026-10-02. Rendering stays on Trame and PyVista (decision 6), so volume rendering needed the same payload and interaction check cartoon rendering got. Both scripts are in docs/.

  • Payload (bench_volume_payload.py). A volume sends the whole grid, so it grows with N³ where an isosurface grows with N². On the 19 volume sections in ten example archives (largest 75×75×30, most under 0.3 MB) a float32 volume sends 0.26 MB or less and uint8 0.06 MB or less, below the 0.60 MB cartoon ribbon. Synthetic grids show the risk: a float32 volume is about 8 MB at 128³ and 67 MB at 256³, against 1.0 and 4.0 MB for the isosurface. Quantising to uint8 and capping the grid at about 128³ keeps it near 2 MB.

  • Frame rate (bench_volume_fps.py). In Chrome on an Apple M4 Pro, a volume cost about 1 ms more per frame than today’s surface (2.7 to 3.1 ms against 1.8 to 2.0 ms) from 64³ to 256³, far above 60 fps.

Conclusion. Neither measurement blocks a first version. Send a capped, quantised grid by default.

Design note, 2026-10-06. The direct volume rendering design sets out the first version and records the six decisions taken that day. Of its five-step rollout, step 1 (the payload module) shipped in v2.20.2, and steps 2 to 5 (the actor and switch, the editor and persistence, the manual, and the measurements above) shipped in v2.20.3. The rollout is complete; its last follow-up, the [for qvf] ask, is filed (above), and the cases under “Still unmeasured” below remain.

Measured 2026-10-06, step 5 of the design (bench_volume_viewer.py). The three cases the first measurements left open, against the real viewer at 1440×900, dragging the mouse across the canvas, with a synthetic water density on 128³, 160³ and 200³ grids (production cubes run 100 to 200 points per axis):

  • Production-sized cubes. Building the payload takes under 0.15 s at every size, serialising it 0.005 s at the 128³ cap and 0.02 s for a 200³ grid sent uncapped; the voxels are 2.1 MB raw at the cap and 8.0 MB uncapped (0.13 and 0.41 MB after compression, an optimistic figure for a smooth synthetic field). Switching Render as to Volume took 0.5 s from click to drawn on both GPUs, 0.8 s uncapped.

  • The hardware GPU (Apple M4 Pro in Chrome, refresh cap off): a drag step costs 2.3 ms with the volume against 2.4 to 2.7 ms with the isosurface, at every size and also for the uncapped 200³ grid. In the full viewer the volume costs no more than today’s surface; the extra millisecond of the bare-view measurement is lost in the rest of the scene.

  • The software renderer (SwiftShader, headless Chromium): a drag step costs 100 to 170 ms with the volume (6 to 10 fps) against 340 to 410 ms with the isosurface (2 to 3 fps); the uncapped 200³ grid draws at 7 fps. On a GPU-less machine the whole viewer is slow, and the contour is the slower of the two, so volume mode adds no new floor.

Measured 2026-10-07, the remaining cases (bench_volume_viewer.py with --dpr, --vsync and its new --archive, on the same M4 Pro; the 160³ baseline re-measured the same day gave 3.1 ms per drag step with the volume against 3.3 ms with the isosurface):

  • Retina pixel ratio (2): 5.3 ms per step with the volume against 5.4 ms with the isosurface. Four times the pixels cost about 1.7 times the step in both modes, and the volume stays no dearer than the surface.

  • The refresh cap on: both modes tie at the display’s 120 Hz, 8.3 ms per step, so a drag feels the same in either; only the cap-off numbers tell them apart.

  • A computed cube: vibe-qc’s water density example, a 68×88×76 grid with nuclear cusps to 101 e/bohr³ against a 99.5th percentile of 0.26, imported with vibe-viewer import and sent whole, being under the cap. Its 0.45 MB of voxels compress to 0.04 MB on the wire, 12:1 against the smooth synthetic field’s 16:1 at the cap: a real field costs a little more wire and stays far under the cartoon ribbon. The switch took 0.9 s, a drag step 2.8 ms as a volume against 2.7 ms as a surface, and on the software renderer 235 ms against 343 ms (4.2 against 2.9 fps), beside 262 against 610 ms for the synthetic 128³ grid the same day; the volume stays the faster of the two on a GPU-less machine. A five-second drag gives too few steps there to trust; the probe’s --seconds 20 does.

Still unmeasured. Other operating systems and GPUs.

v3.0.0: Levinthal’s Lynx, the interactive viewer

Scope recorded 2026-10-19 on the 3.0 milestone and its umbrella issue #110, at the maintainer’s request. The release is about the browser client: work that happens in vtk.js rather than as geometry rebuilt on the server, plus the first public extension contract beyond importers. Released 2026-10-23.

What stays compatible. The QVF format stays version 1; qvf owns it (qvf#11). The Python SDK on the Python API page, the CLI commands, the session file and create_app(readers), which vibe-qc’s contract tests call, keep their 2.x behaviour; 3.0 removes nothing. A 2.x user needs no migration beyond the Python 3.12 floor below.

What makes it a major version. A versioned extension contract with a compatibility promise for the 3.x series, and a change in what the browser guarantees: labels and outlines become continuous client-side behaviour rather than something re-rendered when an interaction ends.

Work packages, one issue each under the milestone, each accepted on its own issue with served-browser evidence:

  • WP 3.1 Client-side label followers (#111). Atom, measurement and axis labels keep facing the camera during an interaction through a vtk.js follower; the server-side billboard basis stays for captures and the terminal. Before this, labels re-faced the camera only when an interaction ended. Released in v2.20.19: the label mesh carries each glyph’s anchor and reading frame as point data, trame-vtk is told to send them, and a vertex-shader replacement places every glyph for the frame’s camera, so the mesh is never rewritten in the page; the bounds grid’s labels already followed in vtk.js, and measurements are a card, not scene labels. Evidence: the trusted-drag profile of tests/test_design_refresh.py reads back a frame mid-drag and matches it to the glyphs rasterised for the camera-facing basis; docs/bench_label_orientation.py --browser for the drag frame time; the browser manual.

  • WP 3.2 Client-side silhouette pass (#112). Outlines for isosurfaces and ribbons as a screen-space edge pass in the client, which the inverted hull of #68 cannot give; the geometry outline stays the fallback without WebGL2 and for headless captures. Released in v2.20.20: a vtk.js render pass after the scene’s own renders the depth of the isosurfaces, Fermi surfaces and ribbons the server marks, and draws their depth and crease edges over the frame, depth-tested, with figure outlines on; translucent surfaces included. Evidence: the silhouette profile of tests/test_design_refresh.py (frame readback against the surface boundary, before and during a trusted drag, and a measure-mode pick with the pass installed); the browser-against-headless figures in the headless page; the limits list and docs/bench_silhouette.py timings in the browser manual; the audit note in docs/design_refresh_2026.md.

  • WP 3.3 Extension API v1 for section renderers (#113). A documented, versioned vibeview.renderers entry-point group modelled on importer plugins: a renderer with load() and render_to_html(), a sidebar title and icon, and a plugin kind that classifies as rendered. A sample plugin ships in the manual and runs in candidate CI. Released in v2.20.20: vibeview.renderer_plugins is the contract, at API version 1; plugin kinds open as details panels in the browser, are reported by info, the banner and doctor, satisfy critical, and fall back to the generic pane in the terminal. The sample package in docs/examples/renderer-plugin is installed by the candidate test job and checked end to end by scripts/smoke_renderer_plugin.py; tests/test_renderer_plugins.py holds the contract tests; Extending vibe-viewer documents both groups and the 3.x compatibility promise.

  • WP 3.4 Client performance qualification (#114). The in-page frame-timing harness extended from volume rendering to drags, representation switches and label followers, a measured matrix for Apple Silicon and Linux, and the thresholds that gate the 3.0 candidate. Released in v2.20.20; required matrix qualified: scripts/client_qualification.py drives the real viewer through eight cases (structure drag, two representation switches, label-follower drag, isosurface drag plain and with the outline pass, the switch to volume rendering and a volume drag) and judges them against scripts/client_thresholds.json per renderer class. The candidate job client-performance runs it on Linux with SwiftShader, gating 3.x candidates and reporting on 2.x; the About dialog shows the WebGL renderer string. The client performance page holds the matrix: Apple Silicon with Chrome is measured on the GPU and on SwiftShader; Linux SwiftShader is measured from the v2.20.20 candidate, with software thresholds tightened to about twice that measured cost. Windows GPU qualification is deferred until users request it and does not gate 3.0; no Windows test PASS is claimed.

Result inspection from one viewer

Scope extended 2026-10-22 at the maintainer’s request: new features go to the next major release. A user request asked for one program covering geometry with a basic editor, orbital and density plots, normal modes as arrows and animation (saved as GIF), and calculated UV-Vis, IR and Raman spectra with a chosen line width and ASCII export. Most of this exists (ORCA GBW orbitals, Hessian vibrations and spectra, property-file electronic spectra, video export, the 3D editor); these work packages close the gaps. Each is accepted on its issue. Status 2026-10-23: all five are implemented on main and verified on their issues.

  • WP 3.5 Spectra and modes from an ORCA output alone (#120). Geometry, normal modes and the IR, Raman and absorption/CD tables read from .out, checked against the Hessian/property route when both exist.

  • WP 3.6 Line shapes and ASCII export (#121). Gaussian or Lorentzian envelopes with a FWHM typed in physical units, broadened in energy, and a documented whitespace-delimited export from the browser and the CLI.

  • WP 3.7 Mode arrows and GIF (#122). Static displacement arrows for the selected mode, in the client, captures and exports, and one-step GIF/MP4 export of a mode from the Vibrations panel.

  • WP 3.8 ORCA densities (#123). Correlated, excited-state, difference and spin densities as cubes through the user’s orca_plot, folded into the calculation when the geometry matches and kept in a saved QVF.

  • WP 3.9 Internal-coordinate editing (#124). Set a measured distance, angle or dihedral to a value, and add, remove or set the order of a bond. 2D drawing stays with vibe-sketcher under the scope decision.

Feature parity for 3D work and result inspection

Scope extended again 2026-10-22 at the maintainer’s request: the goal is that vibe-viewer and vibe-sketcher together cover what users now do in the established desktop programs for 3D structure viewing, building, quantum chemistry result inspection and 2D chemical drawing. The gap analysis below was checked against main on that date. Status 2026-10-23: every work package in it, WP 3.10–3.20, is implemented on main and verified on its issue; the table is kept as the record of the gaps they closed. It lists only capabilities that were missing or partial; the delivered base is described in capabilities. The frozen 2026-09 parity plan remains the record of the earlier builder and live-job work.

Area

vibe-viewer today

Missing

WP

Molecular surfaces

Isosurfaces of stored or evaluated fields; a stored potential mapped onto a density isosurface

VdW, SAS and SES surfaces from atoms, property colouring, mesh/dot styles, area and volume

3.10 (#125)

Interactions

Dashed geometric hydrogen bonds

Configurable H-bond criteria, contacts and clashes, salt bridges, halogen bonds, π-stacking, tables

3.11 (#126)

Macromolecules and databases

PDB with chains, residues, B factors; cartoons

Fetch by PDB/AlphaFold/PubChem identifier, mmCIF, ensembles, alternate locations, sequence panel, sequence-based superposition

3.12 (#127)

Experimental maps

Cube files, isosurfaces, volume rendering

MRC/CCP4 maps, sigma levels, mesh, zone around a selection

3.13 (#128)

Orbitals from other programs

ORCA GBW/JSON, TREXIO, vibe-qc

Formatted checkpoint and Molden files

3.14 (#129)

Outputs from other programs

ORCA result files; Gaussian input geometry

Optimisation steps, energies, modes, spectra and charges from common program outputs; multi-frame XYZ

3.15 (#130)

Building

Atom editor, fixed fragments, hydrogens, SMILES, crystal builder, live relaxation, symmetrize

Bond rotation, coordination templates and ligands, peptide/nucleic-acid builder, conformers, larger fragment library

3.16 (#131)

Coordinates as numbers

Measurements; XYZ export

Editable Cartesian and Z-matrix tables, Z-matrix import/export

3.17 (#132)

Inputs for other codes

vibe-qc and ORCA, vq submission

Gaussian, Psi4, NWChem, Q-Chem, GAMESS, CP2K, xtb, VASP structure

3.18 (#133)

Figures and movies

Publication presets, outlines, screenshots, turntable/trajectory/vibration video

Scene measurement labels, 2D annotations, colour keys, colour/radius by attribute, morph movies

3.19 (#134)

Selection and scripting

Click selection, Compare Files overlay, Python SDK

Selection language, command line with replayable log, multi-model scenes, selection measurements, desktop Python console

3.20 (#135)

Each work package is accepted on its own issue. Optional parsers and readers (for example cclib or gemmi) are lazy extras with an install hint; no GPL dependency is added. Network access (WP 3.12) happens only on an explicit user action. Formats vibe-viewer needs beyond the current QVF kinds go to qvf first.

Decided by the maintainer on 2026-10-23: parity means capabilities, not file formats. Matching what the established desktop viewers can do does not mean reading every quantum-chemistry program’s native files. The readers that exist, including WP 3.12–3.15 and the Turbomole/xtb coord and xtb log readers (#145, #146), are kept and fixed, but no further native readers for other programs’ formats are planned or tracked. A program reaches vibe-viewer by writing QVF; every vibe-viewer surface then reads its results with no change here. Third parties may still ship an importer plugin.

Owned by vibe-sketcher, tracked on its tracker under the scope decision: skeletal drawing, reaction schemes, publication styles, formula and mass already exist there. Its 2D parity gaps, filed 2026-10-22 for its 1.0 milestone, are vibe-sketcher#90–#99: R-groups, repeat units and abbreviations; shapes, highlights and TLC plates in figures; CIP and enhanced-stereo labels; elemental analysis and descriptors; name to structure; CDXML export; office formats and clipboard; calculated NMR shifts on atoms; peptide and oligonucleotide sequences; a stoichiometry table. Spectra line lists (vibe-sketcher#87) share the broadening function of WP 3.6 (#121). vibe-viewer’s part of that workflow is the 3D round trip, deferred out of 3.0 below.

Not planned for 3.0: immersive VR/XR viewing, fitting models into maps, and homology modelling.

Conditional. Translucent isosurfaces through client-side depth peeling join the scope only if WP 3.2 lands with headroom; otherwise they stay under Not yet placed.

Decisions taken by the maintainer on 2026-10-23:

  • The Python floor rises from 3.11 to 3.12, the floor of vq and vibe-sketch. This retires the 3.11 CI lane and lets the queue and sketch extras resolve on every supported interpreter. It is the one compatibility change in 3.0.

  • Representations and 3D actors join the extension contract in a later 3.x minor, not in 3.0, as WP 3.3 (#113) shipped.

  • The codename is Levinthal’s Lynx.

Deferred out of 3.0: metal–metal bonds in clusters, bonds in live playback frames, the vibe-sketcher round trip (cross-repository, see the scope decision), representations and 3D actors in the extension contract (a later 3.x minor) and the trame 4 migration, which is dependency-driven and stays a continuous track.

Acceptance for the milestone: every work package accepted on its issue; the extension contract and the client performance page published in the manual; the candidate pipeline green on the Python 3.12+ matrix; release notes stating the floor change as the only migration; then the normal release process.

Status 2026-10-24: every work package is accepted on its issue; the extension contract and client performance page are in the manual. The immutable v3.0.0 tag names the passing candidate, its release notes record the Python floor, and public desktop downloads exist. Remaining publication obligations are tracked separately on #110. Corrections ship in 3.0.x; accepted packages and artifacts are reused.

Carbon nanotubes and large carbon nanostructures

Proposed 2026-10-28 for a prospective theory-support collaboration with the Alvarez group (University of Cincinnati). Their published work covers vertically aligned nanotube growth from premade metal-oxide nanoparticle films, covalent Cu–C and Au–C bonds at open ends and sidewalls of aligned tubes, nanotube microelectrodes and supercapacitor electrodes, and molecular linkers on Au. The theory we would supply is periodic and cluster DFT of those interfaces. The papers are filed in the paper library (maintenance/survey-alvarez-lab-2026-10.md). Not a release commitment.

Today: measured on 2026-10-28 with four nanotube scenes (a tube, a five-wall tube, a 19-tube bundle and an aligned forest) from 10³ to 10⁶ atoms on an Apple M5 Max (client performance). Before instanced glyphs the ceiling was about 10⁴ atoms, which took two to three and a half minutes to open in the browser. Now the browser opens 10⁶ atoms in 1.5 to 2.2 minutes and turns them at a median drag step under 30 ms; headless capture handles 10⁶ atoms in about a minute; the terminal’s CPU rasterizer stays interactive to about 10⁴ atoms (10⁵ for a single tube) and takes about a second a step at 10⁶.

Shipped on main (2026-10-28), in the order the collaboration needs them:

  • Large-system rendering. The budget above, measured by docs/bench_nanotube_scale.py. Atoms and bonds from 500 atoms are instanced glyphs (vtk.js vtkGlyph3DMapper, sent by a serializer trame-vtk lacked), bonds are built vectorised from the cell’s bond list, and from 10⁴ instances the browser swaps in coarse glyphs while the camera moves. At-rest sphere quality is retained after an inconclusive lower-resolution experiment. All of it runs in the vtk.js client.

  • Nanotube and interface builder. Single- and multi-wall (n,m) tubes, periodic, open, hydrogen-terminated or capped with a C60 half ((5,5) and (9,0)), with diameter, chiral angle, metallicity and wall spacing read out; bundles and forests in Python. A fcc slab ((111), (100), (110)), a nanoparticle or an aminophenyl-amide linker (with an optional slab) goes at an open end or the sidewall at a chosen M–C distance and top, bridge or hollow registry, as one edit-overlay step (nanotubes).

  • Cut-aways and axial views. Half-wall, cross-section and axial cuts that keep atoms near the axis, and orthographic down-the-axis and side views. Browser-only depth cueing now runs in the vtk.js fragment shader; cut-aways and property colours also reach headless and CLI capture.

  • Per-atom property colouring. Coordination, distance from the axis, partial charge (IAO, Hirshfeld, Mulliken, Löwdin) and bond order to a metal, plus a C–C bond-stretch proxy and categorical open-end, coordination-defect and heavy-adduct highlights. These are geometry heuristics, not a strain tensor or complete ring-defect identification.

  • Interface analysis. Metal–C contacts with lengths and bond orders, PDOS summed over the interface atoms, and a density difference integrated over planes along the axis with the cumulative charge transfer, or a potential averaged over them. No QVF change was needed.

  • Growth and reaction-path playback. Live trajectory/reaction frames reuse instanced glyph inputs and draw inferred bonds; automatic playback strides slow frames, manual seeking keeps every frame, and large filmstrips sample four frames. The coordinate editor shows 100 rows per page. A synthetic 100,200-atom trajectory was exercised in Chrome; diagnostic frame updates took 2.8 s under concurrent load.

  • Large-scene loading. Replicated atom arrays, large bond partitions and neighbourhood distances are vectorised. Frozen million-atom forest browser diagnostics measured 91.2 / 84.4 s before/after under concurrent workloads. Schema validation remains the dominant profiled cost; no schema or validation contract was changed.

  • Terminal large scenes. Subpixel atoms use depth-tested point sprites from 10⁴ atoms, with off-screen culling and covered subpixel bonds omitted. Frozen forest diagnostics improved median orbit steps from 74 to 50 ms at 108,000 atoms and from 1.11 to 0.64 s at 1,008,000 atoms. These are not controlled performance guarantees (measurements).

Still proposed:

  • True archive streaming instead of decoding the complete coordinate array, and faster connectivity for real-time growth/NEB playback. Current strides reduce rendering work, not archive memory.

  • Further million-atom loading work, particularly the profiled schema validation cost, and smooth terminal interaction at that scale.

  • Full local strain tensors and ring-topology defect classification beyond the shipped bond-stretch and coordination heuristics.

Constraint: keep the work client-side where vtk.js allows. Impostors and level of detail must live in the vtk.js client, because server render passes never reach it (see Materials and lighting above).

Not yet placed

No registered codename covers these:

  • Translucent isosurfaces without vtk.js depth-peeling stalls; orbitals currently default to opaque as a workaround (design refresh).

  • Metal-metal bonds in clusters and organometallics, which the bonding heuristic cannot see.

  • More fragments and templates: shipped in v3.0.0 (WP 3.16, #131), with a searchable 42-fragment library, coordination templates and sequence builders. The structure library’s thumbnails are shipped in v2.20.3: a preview of the picked molecule and a picture in every pinned and recent chip, rendered in the background from the library’s geometry.

  • Unit-preference coverage beyond #58, which gave the browser’s orbital energies, measurements and band/DOS axes and the terminal’s geometry, orbital, band and DOS views their persisted preferences in v2.20.0. Shipped in v2.20.3: COOP and COHP charts in the browser and the terminal, the band structure capture writes and the capture API’s chart and bond-order functions, the Bond Orders panel, the shift Symmetrize reports, the compare legend, and diff and batch-compare follow the saved units; JSON output and CSV downloads keep the archive’s units. Spectra retain their own picker.

  • A trajectory and vibration scrubber with thumbnails. Shipped in v2.20.3: a filmstrip of frame thumbnails under the trajectory slider, and a phase slider with a filmstrip of one period for a vibration, both rendered off-screen in the background and seekable by click or by a drag across the strip; a click on the energy profile seeks too, and the terminal’s playback keys (Space, [ ], < >) work in the browser. Live playback bonds shipped on main on 2026-10-28; they follow each displayed frame using the shared covalent-radius inference.

  • Rendering QVF excited_states tables. Shipped in v2.20.19 (#109): the states are tabulated by vibe-viewer table --kind excited_states, the terminal and the browser details panel, with energies in eV and nm, oscillator and rotatory strengths, orbital contributions and the linked NTO and transition-density sections. Drawing those from the table is not part of it.

  • Returning edited 3D geometry to its originating vibe-sketcher species with atom/species identity, graph, state and provenance preserved. Implemented on main under #136, with paired Sketch return acceptance on 2026-10-25; result selectors are implemented under #141. Reuse this transport, while tracking Sketch producer work separately under Sketch #67/#88. The broader shared-service direction in the scope decision does not imply a delivered shared package.

  • A vq job handle in QVF provenance (provenance.vq), which needs the producer’s agreement.

Continuous tracks

Accessibility

The served-viewer baseline landed before the split, in project 19 #179–#182, #314, #321, #323, #326, #328, #331, #332 and #341. Still open: systematic contrast, focus order and visibility, complete keyboard reachability, reduced-motion behaviour, and how often high-frequency status messages are announced.

Release and CI infrastructure

The candidate-only product gate and declared documentation extra shipped in v2.18.0 (#22, #19). v2.18.1 completed the full-tree ruff check . gate and strict source-module inventory (#8, #33); tests/test_type_ramp.py prevents new source modules from escaping that inventory. The browser composition refactor (#48) registers callback methods after construction and removes the former untyped-decorator exemption.

Continuing work includes optional-dependency coverage, supported Python versions (the Python 3.14 CI lane shipped in v2.20.2), and evaluating trame 4 once the viewer dependencies support it. Use the tracker and current CI configuration for evidence and status.

Desktop distribution

Source tags and installer publication are separate steps. Availability of a source release does not imply that every platform’s installer or update feed has been published. Distribution, signing and publication remain maintainer operations; see electron/PUBLISHING.md for the portable product contract.

Cross-repository coordination

Producer-dependent contract coverage belongs with vibe-qc, QVF schema and format decisions with qvf, and queue packaging with vibe-queue. Confirm the owning repository’s current issue status before planning a dependency here. Changes to another repository require its maintainer’s coordination.

Decisions carried forward

The decisions locked with the maintainer on 2026-07-02 (parity roadmap §0) still hold, with one exception:

  • Compute: vq first, with a local vibe-qc subprocess for small jobs.

  • Live optimization: vibe-qc semi-empirical methods in a background subprocess.

  • Running jobs: streaming checkpoint QVFs with hot reload.

  • Format breadth (amended 2026-10-23): the existing hand-written readers, with ASE optional, are kept but not extended to further programs. New programs are supported by writing QVF, not by native readers here; see Feature parity.

  • Rendering: stay on Trame and PyVista.

  • Cross-repository work: formal asks to the owning repository.

  • Delivery (superseded): decision 7 is replaced by the release process. Work lands on main and is proved by a release-candidate/* pipeline before a tag, and release only fast-forwards to tags.

History

  • The Avogadro parity roadmap was frozen on 2026-09-13, with a later dated general-slab note retained. It records M1–M6 around the split, with the evidence behind each item. Its issue numbers belong to project 19, and some of its links point into the pre-split layout.

  • ROADMAP_V2.md, at the repository root, is the original v1.1–v2.0 proposal. Its version numbers never matched a release, and its codenames appear nowhere in this repository’s code or catalogue.