RESEARCH PROTOCOL · NO EXPERIMENTAL RESULTS

Spatial tolerance of narrative actions in site-specific augmented reality

A research protocol grounded in HEX001 The Awakening

Status: research protocol and case study in preparation · Version: 1.0 · Date: 22 September 2026 · Project: LABERINTO XR · Case: EL LABERINTO DEL FAUNO, HEX001 — THE AWAKENING

Scope statement. This text presents a research question, a design model, and implementation evidence. It reports no participant study results and demonstrates no perceptual robustness advantage. It has not been peer reviewed.

Abstract

In site-specific augmented reality, narrative actions require different geometric relationships with their physical surroundings. A virtual hand that appears to touch a tree trunk calls for precise proximity and orientation, whereas a distributed environmental response may communicate an event without that specific contact. We propose to examine whether an action can retain its dramatic function while requiring less geometric precision, and how its plausibility changes under spatial perturbation. HEX001 provides a developing sequence set in a real garden and an archive of design decisions, configurations, scenes, and VR releases. A historical CONTACT proposal placed the Faun's hand near a palm tree, but no executed implementation of the gesture has been recovered. Revision A.2 removed the gesture; subsequent versions implemented EL PULSO as an environmental phenomenon that the Faun observes without causing. We define a spatial coupling taxonomy and a multidimensional narrative error budget. Existing observations document technical operation and limited reception of specific releases. They do not establish spatial-error tolerance or narrative equivalence. The next step is to preserve a verifiable executable baseline and construct comparable stimuli before evaluating perception at the site.

Keywords: site-specific augmented reality; spatial narrative; registration; perceptual tolerance; XR design; HEX001; Narrative Error Budgeting.

1. Research question and scope

Primary question. Can a site-specific narrative action preserve its dramatic function while requiring less geometric precision, and how does its robustness change under spatial error?

Our proposed contribution, Narrative Error Budgeting / Spatial Coupling, specifies the relations an action requires with a site, identifies deviations that may disrupt its interpretation, and evaluates those requirements explicitly. The budget need not be a universal distance threshold. Translation, orientation, surface adherence, occlusion, and timing may affect different actions differently.

A secondary methodological principle separates technical validity from perceptual validity. A scene opening, completing, and producing publishable bundles does not establish that a visitor perceives its intended event. Prior AR research has addressed registration stability and measured whether latency, registration inaccuracy, or jitter become noticeable or disruptive in other settings [1, 2]. Our protocol applies this distinction to narrative interpretation; it does not claim a new experimental result.

Comparisons of Gaussian splats and surfels, agent productivity, Meshy, Generic versus Humanoid rigs, texture optimization, general AR/VR equivalence, and heritage interpretation effects are outside the primary contribution. They remain engineering context or possible future work.

2. Case and provenance

HEX001 — THE AWAKENING is part of the EL LABERINTO DEL FAUNO universe within LABERINTO XR. Project records identify OVRLand vascular.walnut.streak, hex 8c3919335981dff, and OVRMap e5310bde-d83d-4b93-afe8-d5d65894fdd7. Carmen de los Mártires is the intended physical AR site. VR releases show a remote representation and do not establish AR registration at that garden.

The supplied records specify Unity 6000.3.18f1, the Built-In render pipeline, OVER SDK 0.5.0 pinned to commit d18797cf5f35ba6b429ecbe2c471a10e8e1c2133, and Visual Scripting 1.9.11. An archived V03 scene has SHA-256 577b5b75a25a7f13e5ae3d05e1d850f436e44335cfb46da85cba350d53d2c363; its documented VR publication is 25303. Release records assign 45,752,901, 45,603,385, and 62,323,621 bytes to the Android, iOS, and WebGL native bundles. These are file sizes, not runtime memory or performance measurements. The handoff contains the scene, settings, and ledgers, but references large binary payloads through a separate catalog. The complete snapshot has not been reproduced in the present environment.

The A.2 sequence covers CALMA → LATIDO_01 → LATIDO_02 → FAUNO_REVEAL → MOVE_TO_STOP_A2 → STOP_PULSO → EL_PULSO → ATTEND_ENVIRONMENT_A2 → RETURN_TO_VISITOR_A2 → CLOSING → COMPLETE, with additional intermediate states depending on release. Its planned path from FAUNO_REVEAL through WALK_MID to STOP_PULSO totals 2.083694 m in planning coordinates. That calculation is not an in-situ AR motion measurement. V03 has focused local checks, but no documented subsequent Web or App perceptual acceptance.

3. From CONTACT to EL PULSO

The historical composition A proposed a contact patch on a palm trunk. It identified a point-cloud center and an approximate character foothold but left reach, animation, physical surface, and in-situ alignment unresolved. A published and tested CONTACT gesture has not been recovered. This proposal motivates the study of actions with demanding geometric dependencies.

Revision A.2 removed contact, its associated large turns, and an additional exit. Subsequent releases implemented EL PULSO as a regional environmental manifestation. The Faun stops, observes it, and redirects attention toward the visitor; the narrative explicitly says that the Faun does not trigger it through touch. Authoring records, scenes, publication logs, and limited observations exist for EL PULSO, including releases 25297 and 25300. A user reported distinguishing the phenomenon in the documented App release. The report establishes its own limited observation, not an experimental comparison.

Historical CONTACT and implemented PULSO also differ in gesture, suggested causality, attention, movement, and visual presentation. A direct comparison of those two versions could not isolate spatial coupling. A future high-coupling stimulus therefore need not recreate the old touch.

4. Proposed spatial coupling model

The taxonomy describes the spatial relationship needed to understand an action, rather than geometric complexity alone:

A higher class does not automatically imply a worse experience. Orientation, occlusion, scale, timing, and narrative meaning will be specified for each action. Historical CONTACT is a C4 candidate. The implemented environmental pulse appears primarily regional, but final classification of future stimuli requires inspection of the frozen graph and presentation.

For action (a), let (e=(\Delta x,\Delta y,\Delta z,\Delta R,\Delta t,\Delta s)), where ΔR denotes orientation change and Δs denotes surface deviation where applicable. We propose an admissible region (B_a), conditional on a particular scene and interpretation task, constrained by technical, perceptual, narrative, and safety requirements. It must be measured, not inferred from a universal standard or from the difference between stored poses.

5. Coordinate evidence and its limits

Original responses preserve a personal pose U and an official pose O for the same map. Their saved Unity translations are U (0, 0.3638, 0) and O (-0.687246323, -2.265177011, -0.049999714); both record quaternion (0, 1, 0, 0) and unit scale. Subtracting the translations gives (-0.687246323, -2.628977011, -0.049999714): 0.689063 m horizontally and 2.628977 m vertically. Project records also apply a separate vertical rebase to the VR representation.

These numbers describe stored pose configurations. They do not measure VPS error, physical registration residual, jitter, or drift. Measuring those properties requires surveyed physical references, a documented alignment procedure, timestamped poses, relocalization trials, and identified devices. Conditions near water and actual physical surfaces require on-site safety checks before any visitor trial.

6. Future study and evaluation criteria

The provisional dramatic function (F) is: after the encounter and the Faun's movement, the place manifests its own response, which the Faun perceives and which redirects the visitor's attention to the environment. This follows the A.2 design and implemented pulse, but must be fixed against a frozen narrative graph before it becomes a stimulus contract.

Two future conditions should share character, region, information, sequence, duration, agency, sound, and audiovisual intensity. In HIGH, reading the event requires precise alignment with a selected physical surface or contour (C2). In LOW, the same event remains legible within the region without precise surface adherence (C1). The high condition must not add a hand that causes the pulse; neither condition should change the story, salience, or consequence at the same time. The stimuli have not yet been implemented, and no test offsets have been set.

Future evaluation will separate (a) execution and technical state; (b) estimated spatial deviation against physical references; (c) plausibility of the scene–site relationship; (d) understanding of what initiates the event and what follows; and (e) safety. Operational criteria, exclusions, and analysis will be set before data collection. No participants have been recruited; no final sample size, effect size, or comparative results exist.

7. Limitations and current status

VR publication and local checks show that observable versions of the sequence were built. They do not certify registered AR operation at Carmen de los Mártires. V03 lacks documented perceptual acceptance after publication. CONTACT is a historical composition proposal, whereas PULSO has an implementation and limited observations; they are not two arms of a completed experiment. The complete binaries of one coherent V03 snapshot have not yet been opened in this environment.

The immediate task is to recover the original scene, dependencies, and bundles by hash, reproduce the baseline without altering production, and fix the narrative contract for (F). The two stimuli and physical registration characterization follow. Participant evaluation and its safeguards come afterward.

8. Conclusion

HEX001 is a concrete case for studying how a dramatic action depends on its spatial relationship to a real place. Its documented transition from proposed contact to implemented environmental response supports a testable hypothesis and the design of more controlled comparators. The hypothesis remains open: we do not yet know whether reducing coupling will preserve comprehension and plausibility under spatial deviations, or what expressive costs it might incur.

References and project materials

1. Bajura, M., and Neumann, U. (1995). Dynamic registration correction in augmented-reality systems. IEEE VRAIS. https://ieeexplore.ieee.org/document/512495/ — technical background on registration and stability.
2. Lo, W. H., Regenbrecht, H., and Zollmann, S. (2023). Sports Visualization in the Wild: The Impact of Technical Factors on User Experience in Augmented Reality Sports Spectating. IEEE Computer Graphics and Applications. https://ieeexplore.ieee.org/document/10232848/ — an empirical example on latency, registration accuracy, and jitter in another context; its thresholds do not transfer to HEX001.
3. AdjustAR: AI-Driven In-Situ Adjustment of Site-Specific Augmented Reality Content (2025), preprint. https://arxiv.org/abs/2508.06826 — related work on site adjustment; it does not validate our narrative hypothesis.
4. Primary project material: ASTRA_HEX_STUDY_HANDOFF.zip, Documentation/08_HEX001_SPATIAL_STUDY.md, Documentation/Hex001TechPreview/a2-official-placement.json, .cache/hex001-streak/reposition.json, Documentation/16_HEX001_NARRATIVE.md, Documentation/17_FAUNO_LOCOMOTION_V03.md, and Documentation/Releases/HEX001_LOCOMOTION_V03/RELEASE_LEDGER.json. Project files remain subject to access and redistribution rights; the handoff ZIP is not a complete executable replica.

Suggested citation: LABERINTO XR. Spatial tolerance of narrative actions in site-specific augmented reality: a research protocol grounded in HEX001 The Awakening. Version 1.0, 2026-09-22. Permanent URL: https://laberintoxr.io/en/research/hex001-spatial-tolerance. No DOI or peer review.