ATLAS Clarus Interaction Wheel — Cross-L v0.5 Diagnostics

Visible L* plane

L*=65 · 987 selectable chromatic references

A · Source B · Context A′ · Result

1 · Select role, then click wheel

Pick A on any plane. Then change L* if desired and pick B.

A · Source

Not selected

—

B · Context

Not selected

—

2 · Area relationship A : B · SURROUND

3 · A′ · Context Appearance Reference

Viewing background: H000_L050_C000 · #777777

Select A and B

A′ searches full active master

—ΔE00 A ↔ B

—|Δλmaster| A ↔ B

—ΔE00 result ↔ target

—|Δλ| result ↔ target

Reference identity is never rewritten.

4 · A′ robustness diagnostics · v0.5

— A_PRIME_TARGET_ERROR · joint score

— A_PRIME_WINNER_MARGIN · absolute / relative

— A_PRIME_LOCAL_DENSITY · 1 / median kNN distance · k=8

— median normalized kNN spacing · same ΔE00 + Δλ search space

— A_PRIME_DENSITY_STABILITY · perturbations: L*±0.25 · a*±0.25 · b*±0.25 · Δλ±0.5 nm

— A_PRIME_SWITCH_COUNT · changed winners across 8 controlled perturbations

Diagnostics only. These values describe target fit, winner separation, local master density and perturbation sensitivity. They do not modify A′ or the v0.4 full-master selection rule.

Cross-L behaviour

A and B retain their own frozen interaction references when you change the visible L* plane. Once both are selected, the model target is calculated in Lab + stored Δλ space and A′ is searched across all 13,283 valid rows of the active master. The interface then jumps to A′'s L* plane and marks the result there.

A′ is model-derived and does not overwrite A, B, or any ATLAS source identity. The interaction model is experimental and is not part of the project-internal normative v3.4.0 source-identity selection.

Method status

Target moves away from B in Lab and in the master's stored Δλ dimension. The full-master result is selected by the same normalized joint ΔE00 + Δλ score used in v0.4. v0.5 then evaluates target error, winner margin, local kNN density and controlled target perturbations without feeding those diagnostics back into selection.

Ratio→k remains provisional: 1:1=7.5%, 1:2=12%, 1:4=17.1%, 1:8=21.8%. Diagnostic perturbations: L*±0.25, a*±0.25, b*±0.25 and Δλ±0.5 nm. Not a validated human-perception law.

Viewing background · retained from v0.4

A′ is displayed on the fixed neutral master reference H000_L050_C000 · RGB 119,119,119 · #777777.

Presentation only: the viewing background is not an input to the interaction calculation and does not modify A, B, A′, the full-master search, or any frozen ATLAS identity.

Master integrity

ACTIVE MASTER BOUND
13,283 valid rows · 19 L* planes · exact PKL RGB/HEX values embedded for all references.

Master SHA-256: 8283ab91b10f89ac758d09ecf5fb4d6343536600a06dd468b1cc1ecf4ec747c4