EX-ITTEN-006 — OLED/LCD appearance equivalence

Question and falsifier

When emitted colorimetry is matched, which spectral, black-level, reflection, off-axis, temporal, and local-dimming variables change appearance or performance? The class-label account fails if within-class device variance rivals or exceeds the OLED-versus-LCD mean.

Work completed

A transparent engineering parameter sweep crossed two illustrative device profiles, four ambient levels, and three white luminances (24 conditions). Effective black was modeled as emitted black plus Lambertian ambient reflection.

Pilot results

At 300-nit white, modeled effective contrast changed from 600,000:1 to 79.5:1 for the illustrative OLED profile between 0 and 1,000 lux, and from 3,750:1 to 38.3:1 for the LCD profile. The values are parameter consequences, not measurements of named products. They demonstrate that emitted matching cannot establish viewed equivalence.

Device experiment ready to run

• Minimum sample: three OLED and three LCD models, with multiple units if feasible.
• Instruments: spectroradiometer, imaging photometer, lux meter, temporal light meter, and goniometric/off-axis fixture.
• Match center-screen XYZ and luminance, then measure spectra, black, flare/reflection, viewing angle, temporal modulation, ABL, local dimming, and tone/gamut mapping.
• Human tasks: dark-detail detection, color matching, small-text reading, halo/local contrast judgment, and discomfort under 0/50/200/1000 lux.
• Analysis: multilevel model with device technology and unit/model random effects. Attribute effects to measured variables, not the marketing class.

Current position / continuation

Stage: apparatus equations and condition matrix complete; no devices measured. Next action: inventory available displays/instruments, record exact model/firmware, warm up and calibrate, then run repeatability measurements before participant viewing. Data: ../experiments/data/ex-itten-006.json.