{"schema_version":1,"experiment_id":"eoa_inverse_innovation_exp05_complete_proposal_portfolio20_20260803","cell_id":"invariant_mode_decomposition_design__art_aesthetics","arm":"COMPLETE_PROPOSAL_PORTFOLIO","candidate_id":"cand_art_modal_projection_proofing_01","proposal_index":1,"version":0,"title":"Modal Proofing for Multi-Projector Art Installations","problem":"During calibration of a multi-projector artwork, each correction cycle changes many coupled visual properties at once: luminance balance, chroma, black level, edge contrast, seam visibility, and the relative salience of image regions. Technicians commonly correct the most visible projector, channel, or seam independently. If the calibration process repeatedly amplifies a coordinated error pattern—such as brighter peripheries, weakened central contrast, and elevated seam chroma—those component-wise corrections can move the error between coordinates or cause oscillation without restoring the artist's intended composition.","actors":["Installation artist","Projection and color-calibration technician","Curator or exhibition producer","Invited proofing viewers"],"observable_state":"Across successive calibration passes, a fixed set of test scenes is measured in predefined image zones. The resulting state vector contains zone-level luminance, chroma, black level, edge contrast, seam contrast, and artist-marked focal-to-background contrast. A problem state is present when repeatable combinations of these measurements persist, grow, reverse, or oscillate across passes even though individual projector or channel errors appear to improve.","consequence":"Calibration can settle on technically acceptable individual readings while altering focal hierarchy, flattening gradients, exposing seams, or producing recurrent overcorrection. The artist must then choose between accepting a compositionally distorted installation and continuing correction cycles whose coupled behavior is not understood.","affected_objective":"Reproduce the artist-approved reference composition at the installed viewing position while keeping seams unobtrusive and preserving intended focal hierarchy, tonal gradients, and color relationships.","intervention":"In one nonpublic proofing session, represent a complete calibration pass as a local transition from the current zone-level error vector to the next. Apply small, artist-approved perturbations to projector gains, channel curves, masks, and blend widths; estimate an explicit local transition matrix; decompose it into invariant modes and scalar gains; and classify modes as decaying, persistent, oscillatory, or growing. Use a modal sensitivity sweep to identify which controllable combination most improves reference error and artist-scored focal hierarchy without worsening seam visibility. Test only the highest-ranked bounded correction on duplicated calibration presets, compare it with ordinary coordinate-wise correction, reconstruct the observed response from retained modes, and withhold adoption if residual, drift, or coupling exceeds preregistered tolerances.","structural_mapping":[{"archetype_element":"Transformation Scope","domain_realization":"One complete measure-diagnose-adjust-remeasure calibration cycle for a fixed artwork, projector layout, ambient-light condition, test-scene set, and viewing position."},{"archetype_element":"State-Vector Definition","domain_realization":"Signed deviations from the artist-approved reference for luminance, chroma, black level, edge contrast, seam contrast, and focal-to-background contrast in predefined zones."},{"archetype_element":"Invariant Mode Basis","domain_realization":"Coupled patterns of zone and visual-property errors whose direction is approximately preserved from one calibration pass to the next."},{"archetype_element":"Modal Gain Spectrum","domain_realization":"For each mode, the estimated multiplier indicating whether the corresponding error pattern shrinks, persists, reverses, oscillates, or grows after a correction cycle."},{"archetype_element":"Dominant Mode Selection Rule","domain_realization":"A mode becomes action-relevant only if it is reproducible across held-out perturbation trials and either fails to decay or materially changes a preregistered aesthetic or seam outcome under sensitivity testing."},{"archetype_element":"Stable/Unstable Mode Partition","domain_realization":"Modes are partitioned by their estimated repeated-cycle behavior within the tested calibration neighborhood, with uncertainty intervals retained around the stability boundary."},{"archetype_element":"Modal Intervention Map","domain_realization":"A mapping from feasible combinations of gain, curve, mask, and blend-width changes to their predicted effects on each action-relevant mode and the protected aesthetic outcomes."},{"archetype_element":"Reconstruction Residual Check","domain_realization":"The retained modes reconstruct held-out zone-level changes; unexplained magnitude and structured spatial or color residuals remain visible rather than being absorbed into the modal account."},{"archetype_element":"Mode Drift Monitor","domain_realization":"Modes, gains, and their ordering are rechecked after projector movement, lamp or laser-output change, ambient-light change, media re-encoding, or repeated prediction failure."},{"archetype_element":"Interpretation Scope Contract","domain_realization":"The decomposition describes local calibration dynamics for this work and setup; it does not define artistic value, general viewer preference, or behavior outside the tested perturbation range."},{"archetype_element":"Mode-Coupling Register","domain_realization":"Cross-effects observed when a nominally single-mode correction also changes other modes, especially seam and focal-hierarchy modes, are recorded and carried into the action decision."},{"archetype_element":"Local Linearization Window","domain_realization":"Only small parameter changes inside artist-approved bounds, under fixed content, geometry, viewing position, and ambient illumination, are covered."},{"archetype_element":"Spectral Gap Threshold","domain_realization":"A dominant-mode simplification is permitted only if a preregistered separation and direction-stability criterion distinguishes retained modes from dropped modes across resamples."}],"mechanism_mapping":[{"mechanism_slug":"eigendecomposition_workflow","role":"Decompose the estimated square calibration-transition matrix into a complete set of candidate invariant error directions and their pass-to-pass gains; report eigenvector conditioning as part of fitness for use.","counterfactual_removal":"Without it, the intervention has correlated measurements but no transformation-defined invariant directions or repeated-cycle gains."},{"mechanism_slug":"modal_stability_analysis","role":"Interpret each modal gain relative to the discrete-cycle stability boundary and identify decaying, persistent, oscillatory, and growing error patterns within the tested neighborhood.","counterfactual_removal":"Without it, the spectrum cannot distinguish a currently visible mode that will fade from a less visible mode that will accumulate across correction cycles."},{"mechanism_slug":"modal_sensitivity_sweep","role":"Perturb feasible controls in the fitted local model and bounded physical trials to rank modes by effects on reference error, focal hierarchy, and seam visibility while logging cross-mode effects.","counterfactual_removal":"Without it, mode magnitude would be mistaken for intervention value, and corrections could be chosen without knowing which protected outcomes or other modes they disturb."},{"mechanism_slug":"residual_reconstruction_test","role":"Reconstruct held-out calibration responses from retained modes and inspect both normalized error and structured zone-level residuals before allowing modal control.","counterfactual_removal":"Without it, omitted spatial or chromatic behavior could remain hidden behind an apparently concise decomposition."},{"mechanism_slug":"spectral_gap_monitor","role":"Check whether retained modes remain separated and directionally stable across resamples and after changes to hardware, geometry, content encoding, or ambient light.","counterfactual_removal":"Without it, a once-useful dominant-mode model could continue guiding calibration after its simplifying separation has disappeared."},{"mechanism_slug":"spectral_decomposition_report","role":"Give the artist and technician a bounded account of mode meanings, gains, couplings, residuals, operating conditions, and prohibited interpretations.","counterfactual_removal":"Without it, tentative local modes could be handed off as independent, causal, or generally valid properties of the artwork."}],"causal_chain":["Successive calibration passes act as a coupled local transformation on the installation's zone-level visual-error state.","Component-wise correction obscures error combinations that the transformation preserves, damps, reverses, or amplifies.","Bounded perturbation trials make a local transition matrix estimable for the fixed installation regime.","Eigendecomposition and stability classification expose repeatable error modes and their pass-to-pass gains.","Sensitivity testing identifies control combinations that act on an outcome-relevant mode while revealing cross-effects on seams and focal hierarchy.","A duplicated-preset trial compares the selected modal correction with coordinate-wise correction without changing the approved installation state.","Residual, spectral-gap, and drift checks determine whether the modal explanation remains adequate for this bounded decision.","Only if those checks pass does the artist receive the modal correction as an optional calibration preset for aesthetic review."],"baseline":"The current baseline is technician-led coordinate correction: inspect projector, channel, zone, and seam deviations; adjust the largest visible error; remeasure; and repeat until tolerances and artist review are satisfied. Evaluation compares this process with the modal preset under the same reference scenes, perturbation bounds, measurement procedure, and review conditions.","nearest_rivals":["Vendor automatic multi-projector color and brightness matching, which minimizes device-level discrepancies but need not model pass-to-pass coupled error modes or artist-defined focal hierarchy.","Global color-management or lookup-table fitting, which maps input colors to measured output but does not classify growing or oscillatory calibration directions under repeated correction.","Zone-by-zone manual proofing with side-by-side reference comparison, which preserves artistic judgment but acts on named coordinates rather than invariant combinations.","Multivariate regression from calibration controls to output errors, which can predict coupled changes but lacks an explicit repeated-transformation stability partition unless augmented with one."],"remaining_contrastive_claim":"The proposal's distinguishing testable feature is that calibration choices are selected from invariant directions of an estimated correction-cycle transformation, ranked by bounded outcome sensitivity, and rejected when reconstruction residuals, coupling, drift, or spectral separation invalidate that modal view. It does not assert superiority over the listed rivals.","authority_safety":{"decision_authority":"The installation artist retains final authority over aesthetic acceptance. The technician controls measurement and reversible preset implementation; the curator may enforce venue and equipment constraints but may not redefine the artist's reference objective.","authorized_first_step":"Run a nonpublic, reversible pilot on one artwork using duplicated calibration presets, fixed reference scenes, small preapproved perturbations, and no changes to the master media or physical installation geometry.","excluded_actions":["No permanent overwrite of the artist-approved calibration or master media.","No public deployment before artist review of the resulting preset.","No automated optimization of an unmeasured aesthetic objective.","No perturbation outside equipment-safe and artist-approved bounds.","No interpretation of a mode as an artistic intention, quality judgment, or viewer preference without separate evidence.","No reuse of the fitted modes for another artwork, venue configuration, or lighting regime without re-estimation."],"halt_rollback":"Stop if equipment limits are approached, the artist reports unacceptable visual change, held-out residual exceeds the preregistered budget, a retained mode becomes unstable across resamples, or protected seam and focal outcomes worsen. Restore the untouched approved preset, verify it against saved measurements and reference scenes, and archive the pilot outputs as nonoperative evidence."},"negative_tests":{"strongest_counterevidence":"Small replicated perturbations produce no stable coupled directions: estimated modes rotate across resamples, eigenvectors are ill-conditioned, the spectral gap is absent, or a simple independent-coordinate model predicts held-out responses equally well with lower error and complexity.","problem_falsifier":"Across repeated baseline calibration passes, errors change independently or randomly rather than persisting, growing, reversing, or oscillating in reproducible combinations; component-wise correction then adequately describes the observable process.","intervention_falsifier":"On held-out perturbations and duplicated presets, the modal model fails its preregistered prediction and residual criteria, or its selected correction does not preserve the protected aesthetic outcomes at least as well as coordinate-wise correction under blinded preset review.","risks":["The local transition estimate may be underdetermined relative to the number of visual measurements and controls.","Measurement noise or projector warm-up drift may create spurious modes.","Nearly equal eigenvalues or non-normal dynamics may make individual mode directions unstable or misleading.","Artist ratings may vary across scenes or viewing positions, making the sensitivity ranking contingent.","Optimizing measured contrast and color errors may displace unmeasured texture, temporal, or conceptual qualities.","Additional perturbation passes may consume equipment time or accelerate wear.","A concise spectral report may lend unwarranted authority to a provisional mathematical description."]},"next_evidence_step":"Pre-register a single-session pilot on one nonpublic installation: select two representative fixed scenes, twelve predefined zones, a limited state vector, and small safe perturbations of existing reversible calibration controls; reserve a subset of perturbation trials for held-out evaluation; compare the modal transition model with an independent-coordinate model on signed change prediction and structured residuals; then present randomized duplicated presets from the highest-ranked modal correction and ordinary coordinate correction to the artist and a small invited proofing group. Evidence is sufficient only to decide whether a larger within-installation test is warranted, not to infer general aesthetic benefit.","prior_art_status":"UNSEARCHED","diversity_from_prior_proposals":"This is the initial sealed proposal for the specified cell, so no prior proposal within the cell is available for comparison.","revision_record":{"parent_version":null,"progress_targets_addressed":["Initial complete proposal"],"conceptual_changes":["Established repeated multi-projector calibration as the domain transformation and coupled visual-error patterns as its candidate invariant modes."],"operational_changes":["Defined a reversible, nonpublic duplicated-preset pilot with fixed scenes, bounded perturbations, held-out trials, explicit authority, halt conditions, and rollback."],"evidence_changes":["Specified comparison against an independent-coordinate model and randomized review of modal versus coordinate-wise presets."],"claim_changes":["Limited the claim to a testable structural contrast and explicitly withheld novelty, prevalence, demand, effect-size, and superiority claims."]}}