How metrics connect

Many skin metrics are not independent — some are the inverse of another, some are derived from a single parent, and some are combinations of several. Requesting the right one for your use case is easier with the whole picture in one place. This page is that map; each metric links to its own concept page for detail.

The map at a glance

shine ─┬─► dull_skin
       ├─► radiance
       ├─► skin_type
       └─► hyperpigmentation_adjusted

uneven_skin_tone ──► smoothness    ◄ inverse pair
                 └─► radiance

wrinkles + eyes ──► skin_sagging ──► skin_firmness    ◄ inverse pair

redness    ──► redness_absolute
dark_spots ──► dark_spots_adjusted

pigmentation analysis ──► { hyperpigmentation · dark_spots · freckles · melasma }

Inverse pairs

Two views of the same analysis: a severity score and its “higher is better” complement. Request whichever matches your UX — never both in the same view (users find the inversion confusing).

Severity metric	Inverse (“good number goes up”)
`uneven_skin_tone`	`smoothness`
`skin_sagging`	`skin_firmness`

The two carry the same information, so there is nothing to gain from requesting both — and a region or whole-face score that is unavailable in one is unavailable in the other.

Derived from a single parent

A transform of one other metric — most often a cross-skin-tone-comparable rescale.

Metric	Derived from	Purpose
`dull_skin`	`shine`	Absence of luminosity
`redness_absolute`	`redness`	Cross-tone-comparable redness
`dark_spots_adjusted`	`dark_spots`	Cross-tone-comparable dark spots

Combination metrics

Built from several inputs together.

Metric	Combines
`radiance`	`smoothness` + `shine`
`radiance_adjusted`	A cross-skin-tone-comparable radiance variant
`skin_type`	`shine` + `pore_dilation` + `redness` + `dull_skin`
`hyperpigmentation_adjusted`	`shine` + `dark_spots` + `melasma`
`skin_sagging`	`wrinkles` + `eyes`

The pigmentation family

hyperpigmentation, dark_spots, freckles, and melasma all come out of one pigmentation analysis. They describe different facets of the same signal — cumulative pigment, large distinct spots, small freckle clusters, and diffuse patches. Request any subset; they share the same underlying pass.

Which to request for your UX

Diagnostic / “what’s wrong” UIs — use the severity metric (uneven_skin_tone, skin_sagging, dull_skin).
Improvement-over-time / “good number goes up” UIs — use the inverse (smoothness, skin_firmness, radiance).
Cross-user analytics or cohort comparison — use the cross-tone *_adjusted / *_absolute variants, which are normalized for comparability across skin tones.