EM 1.5 — Long-Term Support

Recommended for all new Extended Matrix projects. EM 1.5 is the current Long-Term Support release — if you are unsure where to start, start here. EM 1.4 LTS remains available for legacy projects already authored against it.

Blender pin. EM 1.5 works on Blender 4.4 LTS and later — the download picker offers two builds, one for the Python 3.11 ABI (Blender 4.4 LTS or 5.0) and one for the Python 3.13 ABI (Blender 4.5 LTS or 5.1). Mac Intel (pre-2020) users should stay on Blender 4.4 LTS or 4.5 LTS, since Blender 5.x ships no Intel build. The official LTS binding for the 1.5 line will follow the next Blender Foundation LTS in the 5.x series — Blender 5.2 LTS is planned for July 2026; see blender.org/download/lts for upcoming LTS announcements.

What’s new in 1.5

In 1.4 you could describe a stratigraphic sequence and tag its sources, but the rest of the reconstruction — the chronology, the documents that justified each 3D piece, the geometry of the proxy itself — had to be threaded together by hand, across tables, side-notes and folders. 1.5 closes those gaps. The graph now carries the reconstructive narrative, documents are tracked alongside the 3D objects they describe, the proxy geometry is drawn directly from the evidence that supports it, and a published model can be opened in a browser by anyone you share the link with — with the sources behind each piece visible inline.

axisLanguage

The graph language gains room for things that, in 1.4, you had to record outside the graph or work around.

Surface areas — the projected portions of a wall or floor that an archaeologist actually measures and discusses — become first-class graph citizens via the new Working Unit (UL) node type. Re-used architectural pieces (spolia) get their own node, the Reused Special Find (RSF, DP-26 — the last development project before the 1.5 cut), so that the second life of a stone in a later wall is recorded explicitly rather than narrated in commentary. Paradata themselves carry richer metadata, with additive family and is_series fields per subtype: when a single inscription or document covers a series of stratigraphic units, the graph reflects that continuity instead of scattering identical references (built on the s3dgraphy v1.5.2 datamodel patch — additive only, no breaking changes to the EM formalism).

Container nodes (VSF, US, USD) and a dotted transformation edge (changed_from) let the graph describe instance chains and reuse end-to-end across yEd, s3dgraphy and EM Tools (DP-36, DP-39). Group nodes — Time Branch, Activity, Paradata — are formalised in the palette and drive things like the Activity Manager’s “By Activity” filter and the bundling of evidence behind each source in Heriverse (DP-43). Authors, licenses and embargo terms become typed nodes you connect with their own dashed edges (has_author, has_license, has_embargo); the graph can then resolve, for any given unit, who claimed what under which license and under which embargo, by walking up from the node to its swimlane to the graph as a whole (DP-32, DP-51).

Most visibly for daily work: the paradata that used to clutter the area around each unit can now be collapsed into a single, openable cluster — the Paradata NodeGroup (DP-60). Open the cluster and you see the same scattered nodes as before; closed, the graph stays readable as projects scale to hundreds of units. Old graphs continue to render without change.

A LocationNodeGroup for spatial membership decoupled from time and intention is in the pipeline for s3dgraphy 0.1.41, just after the 1.5.0 cut.

(For the formal definitions of these node types, see the Extended Matrix manual and the development projects listed on the dev tracker.)

axisTools

In 1.4 a modeller could describe stratigraphy and pin sources, but the reconstruction itself — the 3D pieces, the documents behind them, the proxy geometry — was authored across panels that didn’t quite talk to each other. 1.5 fills in the missing ones. The documents you cite have a manager of their own. The 3D objects of the reconstruction become first-class graph entities with their own epoch handling. The proxy geometry can be drawn in a way that already encodes the measurements that justify it, so you are not authoring shapes and then bolting evidence to them in a second pass.

axisMethodology

The 1.4 workflow had places where the modeller had to bridge between disconnected steps — exporting tables, hand-copying data between tools, holding mental state across passes from chronology to publication. 1.5 fills those gaps.

The chronology of a reconstruction is no longer something you describe in commentary: with CronoFilter (DP-10) you slide through the horizons of the project and watch the model evolve epoch by epoch in the viewport, and the same horizons drive what a published Heriverse scene shows when a visitor moves a time slider. Landscape mode lifts a project out of the single-monument register into territorial workflows, where multiple graphs co-exist in one Blender scene with their own publishable flags and graph-code prefixes.

The sources behind the work live in a structured XLSX template that you import into the graph through a wizard with conflict resolution (DP-45), instead of being a list at the end of a document. The reconstructive steps that EM has long described are now operations you perform in the tool — the propagative metadata, the container groups and the dotted transformation connector all let the graph carry the steps explicitly, instead of relying on external commentary. A self-learning example dataset ships alongside the 1.5.0 cut, so a first-time user can step through a working graph rather than authoring one from scratch.

(Methodological details are in the Extended Matrix manual and on the dev tracker.)

axisCase study: Great Temple

Editor’s note: this case-study block is the placeholder Emanuel will replace with the production text. The frame below sketches the intended structure.

The Great Temple is the first public reconstruction authored end-to-end on EM 1.5 — from the source dossier, through the proxy and the documentary evidence, to a published scene a peer can open in a browser. Each visible piece of the model resolves back, through the graph, to the units it represents, the documents that describe them, the author who proposed the reconstruction and the license under which it is shared. The reconstructive narrative — what was known when, on what evidence, by whom — is encoded in the graph rather than in surrounding commentary.

The Great Temple was not modellable end-to-end on EM 1.4: [TBD — name the precise gap, in terms of what the archaeologist could not do. Candidates: there was no place to track documents alongside the 3D objects they describe; the proxy could not be drawn in a way that carried its own evidence; there was no chronology slider to navigate the multi-phase narrative; the source dossier could not be queried as structured data. Pick the gap the case study actually closes and rewrite this paragraph around what the modeller can now do that they previously had to work around.]

Migrating from 1.4

EM 1.5 is fully backward-compatible with 1.4. There are no breaking changes: a graph authored against 1.4 opens, renders and round-trips cleanly through 1.5 tooling, and the DP-26 spolia work landing in the cut keeps this guarantee.

What 1.5 introduces are new and simpler ways to express the same things, not replacements. The clearest example is the Paradata NodeGroup (DP-60): the per-US paradata that in 1.4 hung around each stratigraphic unit as scattered nodes can now be folded into a single, openable cluster — the graph stays readable as the project grows, without losing any of the underlying detail. Old-style scattered paradata still parses and renders correctly; new graphs benefit from the cleaner aggregation by default, with no work required from the author.

If you have in-flight 1.4 projects, no migration is required — open them in 1.5 and continue authoring. The new constructs become available organically as you iterate.

Tip: see DP-60 — Paradata NodeGroup for the formalisation that drove the visual decluttering work.

Patch history

Patches to the 1.5 line are listed here as they are released. The detailed commit-level changelog lives in the EM Tools changelog.

1.5.0 — 2026-05-15

• Initial 1.5.0 stable release. Becomes the recommended LTS line for new projects. Targets Blender 5.1 (current stable); the LTS binding will follow the next Blender Foundation LTS in the 5.x series. Blender 4.5 LTS remains a supported tested combination for users who need an LTS-pinned environment today.

Editors: add ### 1.5.1, ### 1.5.2, etc. above this line as patches ship.

Full changelog (draft)

This changelog is a draft compiled from the development projects tracker. For the canonical source see the Extended Matrix dev page.

Incorporated development projects

• DP-07 — Document Node as a Spatio-Temporal Entity — three-axis classification (role / content nature / geometry) formalised in the yEd palette, in s3Dgraphy constructors, in the StratiMiner prompt and in the Document Manager Edit Classification operator.
• DP-10 — Multigraph Project — Landscape Mode, CronoFilter horizons with auto-from-epochs and .cf.json save/load, s3Dgraphy multi-graph API, per-graph publishable flag, graph-code object prefixing, and multi-graph Heriverse JSON export.
• DP-11 — Heriverse Project — exporter refactored into the export_operators/heriverse/ subpackage; multi-graph JSON, GLTF + Draco + optional GPU instancing, separate texture pipeline, ParaData (RMDoc) and Special Finds export, plugin-style provider registry in the Export Manager panel.
• DP-32 — Propagative Metadata System — hierarchical scope resolver (node → swimlane → graph) with five built-in rules (chronology, author, license, embargo), bidirectional TPQ/TAQ closure, hard-policy paradox detection with claim attribution, and 13 locked synthetic regression scenarios.
• DP-35 — UI/UX Refactoring — modular package architecture across every former monolithic top-level file, plugin-style Export Manager with ExportProvider registry, uniform experimental gating, Visual Manager with property-based colouring, Blender Extension packaging with multi-platform CI/CD.
• DP-36 — Container Nodes — VSF / US / USD container groups shipped end-to-end across yEd, s3Dgraphy and EM Tools, with containment filtering and instance-chain edges; Anastylosis Manager refactored to consume the new container model.
• DP-39 — Transformation Connector (Dotted) — changed_from edge type with instance-chain helpers, GraphML round-trip preservation and instance-chain filtering in the Stratigraphy and RM Managers; absorbs the documentation work of the former DP-44.
• DP-40 — Canvas Header Metadata Tags — graph-level [ID], [ORCID], [License], [Embargo] tags on the canvas title, consumed by the DP-32 resolver as graph-scope defaults.
• DP-43 — Group Nodes — Time Branch (green), Activity, and Paradata group nodes formalised in the palette and round-tripped end-to-end; the Activity Manager’s “By Activity” filter and the Heriverse exporter’s per-source ParaData bundling are driven by these groups.
• DP-45 — XLSX Auxiliary Data & US Image Resources — XLSX import and merge wizard with conflict resolution, paired with the s3Dgraphy Graph Merger (v0.1.33). US image resources and the Image Viewing System are deferred to 1.6.
• DP-46 — Proxy Box Creator — 7-point measurement system with the Extractor / Combiner paradata chain wrapped in a per-US ParadataNodeGroup, document-instance cloning at every run, GraphML write-lock guard, save-after-create on by default, and a chain-summary narrative in the panel.
• DP-47 — RMDoc Manager (3D Document Manager) — image import, quad and camera creation with DOSCO path resolution, master/instance model with certainty classification, three-axis Edit Classification operator, and an opt-in colour-by-geometry toggle that mirrors the document’s geometry on the viewport quad.
• DP-48 — GraphML Writer & Updater — exporter + incremental patcher used by Create Host, Bake and Save GraphML; visual rules honoured end-to-end. Multi-source assembly pipeline split out to DP-53 for 1.6.
• DP-50 — Surface Areale System — projective and shrinkwrap surface-proxy strategies on Representation Models, with the full US → Property → Extractor → Document → RM paradata chain, an A/B/C complexity classifier, and the new Working Unit (UL) node type. Boolean + LOD strategy split out to DP-54 for 1.6.
• DP-51 — Author Node Formalization — palette-backed AuthorNode, AuthorAINode, LicenseNode and EmbargoNode classes with has_author / has_license / has_embargo edges, dashed-connector reclassification by target class, and integration with the DP-32 resolver. (Operational at the graph layer; surfacing in EM Tools UIs is incremental and continues into 1.6.)
• DP-55 — Stratigraphic Unit Creation Workflow Unification — a single shared Add-US floating dialog used by Stratigraphy Manager, Proxy Box Creator and Surface Areas; JSON-driven US-type registry, gap-aware shared-pool numbering with legacy PREFIX_ALIASES. Builds on the s3Dgraphy v1.5.2 datamodel patch (additive family + is_series metadata per subtype — the EM formalism stays at 1.5, no breaking changes).
• DP-60 — Paradata NodeGroup — group node for aggregating per-US paradata into a single visual container. Strong visual decluttering of the EM graph; backward-compatible with 1.4 scattered-style paradata. Sibling of DP-43 (Group Nodes), consumed by DP-19 (Swimlane Paradata Node Group), DP-46 (Proxy Box Creator) and the Heriverse exporter.

Pending before the 1.5 cut

• DP-26 — Spolia Project — last development project before the 1.5 release: introduces the Reused Special Find (RSF) node with red border for re-used architectural and decorative elements (spolia), reinforcing the “US container” formalism in language and code. Lead: Nicola Delbarba (Ph.D. thesis).