End-to-End: A Complete Site Workflow

The previous intermediate tutorials each focused on one slice of the EM workflow. This one stitches them together on a real (small) site so you can see the whole life-cycle: from the field report and the photogrammetric survey to the published reconstruction with auditable paradata.

It is the longest tutorial in the manual and is meant to be taken in three sessions of 60–90 minutes each.

Learning objectives

By the end you will have:

a complete EM graph for a real (training) site, with units, relations, sources, paradata and TSU;
a Blender scene where every proxy is bound to its graph node and every selection cross-walks both ways;
a multi-temporal reconstruction with at least two epochs and one alternative hypothesis;
a Heriverse export that opens in the web viewer;
a small report (auto-generated CSV + your commentary) summarising the project’s quantitative side.

Prerequisites

All four intermediate tutorials:

Building an Extended Matrix Manually, Unit by Unit
Linking EM to Your 2D and 3D Documentation
Managing Sources, Paradata and Metadata
familiarity with at least one of the advanced tutorials (Multi-Temporal 3D Visualization Demo is the most relevant).

Dataset

The case study uses the Casa di Esempio training dataset, which includes:

18 stratigraphic units (US, USV, USD, SF) covering one room across three phases;
one TSU campaign (decay survey) on the south wall;
12 source PDFs (excavation report, drawings, archival photographs, one material analysis);
a photogrammetric mesh (~250k tris) and an orthophoto of the floor;
two reference reconstructions (one published, one alternative).

Todo

The training dataset is being assembled and will be downloadable from the Extended Matrix website. Until then, follow the sessions below using your own equivalent material — every step is dataset-agnostic.

Session 1 — Authoring (~90 min)

Goal: get the graph and the proxies in place.

Apply Building an Extended Matrix Manually, Unit by Unit to the 18 units of the case study. Save the GraphML.
Apply Linking EM to Your 2D and 3D Documentation: bring in the orthophoto, the section drawing and the photogrammetric mesh; create proxies for every US that has a physical body.
Save the project.

Acceptance check. Selecting any US in the Stratigraphy Manager reveals at least one source and (for physical units) flies the camera to the proxy.

Session 2 — Reasoning (~90 min)

Goal: turn the units into justified claims.

Apply Managing Sources, Paradata and Metadata to the five units whose interpretation depends on more than one source.
Define the three epochs of the case study in Epochs Manager; assign every US to its has_first_epoch (and has_last_epoch where it differs).
Build the alternative reconstruction hypothesis for the southern wall: duplicate the RM, mark the divergence with a Property node pointing to the supporting Document.
Add the TSU campaign on the south wall using Conservation Workflow (TSU).

Acceptance check. The Paradata Manager filter view shows full chains for the five interpreted units; the Cronofilter steps cleanly through the three epochs; the TSU materials are visible on the south-wall proxy.

Session 3 — Publication (~60 min)

Goal: produce the deliverables.

Run a full Heriverse export (Export Manager); open the result in the web viewer, verify epoch switching and source pop-ups work.
Run the table exports (EM (csv), US/USV, Sources); check that every row carries the right epoch, type and source reference.
Run Export Statistics (Experimental) — volumes by epoch and by certainty.
Write a 1-page commentary describing what the numbers mean (this is the part that no exporter can do for you).

Acceptance check. A colleague who has never seen the project can open the Heriverse package and reach, for any selected unit, the supporting source within two clicks.

What to take away

A site that looks small (18 units) generates a non-trivial graph (~50 nodes including paradata). EM rewards discipline.
The interesting part of the workflow is not the panels. It is the moments where you decide which type a unit is, what an extractor extracted, whether a divergence is a real hypothesis or a stylistic preference. The panels make those decisions traceable.
The deliverables you produce here (Heriverse package + CSVs + commentary) are a complete archival unit on their own — they can be deposited in a repository alongside the field documentation.

Where to go next

Export Guide — deeper export options for archive deposit, Zenodo, etc.
CronoFilter — fine-grained chronological horizons across multiple sites.
The advanced tutorials (14–18) — each picks one of the patterns you used here and explores it in depth.