Guide · explorable
Spectral Metamerism
Two different brands can read the same to one cohort, and differently to another. Edit the two emission profiles and the cohort's salience weights, and watch their perceived readings collapse and separate. This is an illustration — to measure a real brand from public evidence (with noise floors and cohort resolution), use the Brand Spectrometer.
Profile A emission
Profile B emission
Cohort weights salience w · 0–1
What this shows. Two different emission profiles, projected through a
cohort's salience weights, can yield one perceived reading —
metamerism. The perceived reading is a salience-weighted mean,
perceived(P) = Σ wᵢ·Pᵢ / Σ wᵢ. Click
“Make B metameric to A”, then drag the
cohort weights: the two readings split apart. That is the point —
metamerism is observer-relative. A different cohort resolves the very
brands the first one confused; there is no observer-free “true” reading. This collapse
is exactly why a single perceived impression underdetermines the emission — the
Brand Spectrometer
triangulates across cohorts and artifacts to constrain it.
What it does not claim. This is a deliberately simple one-scalar
projection for teaching; a real reflection is cohort-resolved with a
noise floor and reported only when signal clears it. For the geometry of the projection,
start from the math path; for the brand reading, the
CMO path. Terms used: cohort, perception cloud, reflection,
metamerism, observer spectral profile.
From illustration to measurement
Collapse to one perceived scalar is the whole problem: many emission profiles share it, so a lone impression cannot recover the brand that produced it. A real reading is cohort-resolved, triangulated across public artifacts, and reports a noise floor — it declares two brands different only when the signal clears it. That is the Brand Spectrometer. For the geometry of the weighted projection, start from the math path; for what the eight dimensions are and why eight, the CMO path on the Guide.