How it works
Colour is a measurement, not an adjective.
Every claim Vhue makes reduces to two numbers: how far a trend colour sits from a shade you own, and what closing that distance is worth. Everything below exists to make those two numbers trustworthy.
Extraction
Frames are sampled from posts driving a trend and the dominant shade is taken from the pixels themselves. A caption saying “cherry cola” is a guess; #4A1521 is a fact. Skin, background and lighting are separated out before the colour is trusted.
Conversion
sRGB is device space, not perceptual space — two colours the same distance apart in RGB can be obviously different or invisibly close. Every shade, trend and SKU alike, is converted to CIELAB under a D65 white point before anything is compared.
Comparison
Distance is CIEDE2000, the standard the colour industry actually settled on, with its corrections for hue, chroma and lightness. It is computed against your catalogue, stored at write time, so a match is a query rather than a conversion run.
Valuation
A match with no figure attached is trivia. The Margin Engine takes the matched SKU's real cost, price, stock and velocity and models the run in whole cents — never floating point, because money that drifts is worse than money that is missing.
The verdict scale
Five answers, and only one of them costs you a season.
The guardrail
The engine refuses rather than flatters.
A shade whose unit cost exceeds its retail price cannot be modelled, so Vhue declines to model it and says why, at the point of entry. A declining trend does not return an opportunity — it returns markdown exposure, as a negative figure, labelled as what it is.
Every projection carries the inputs it used, frozen at the moment it was computed, and the engine version that produced it. A figure you cannot audit six weeks later is a figure you should not have presented to a board.