Vanity or Vice
Skip to content
Northern Lantern House Labs (opens in a new tab)
Makeup Intelligence

Score a profile · leave with a list

The Edit

0 of 29 fields personalized · risk 43 · skin-like 57 · 3 films · 8 min applying

The answer

Build performance minimal — 6 objects, 3 films, about 8 minutes.

Why

Sebum movement is doing the most model work (+7). The model raises wear pressure because sebum is in the profile and you do not want midday intervention.

Caveat

Education only — never a diagnosis, toxin score, safety ranking or calibrated probability. Verify the unit in your hand.

Next

Read the order, then open the kit, pick one house example, and take your list to the counter.

Pancake index 43 out of 100. Skin-like index 57 out of 100. These are heuristic model scores, not measured probabilities.

Your pancake index

43

Conditional · Balanced architecture

Skin-like index 57·model baseline 30

Products

6 / 6

Layers

3

Minutes applying

8

Tension

20

Why this pancake risk

  • +7

    Sebum movement

    The model raises wear pressure because sebum is in the profile and you do not want midday intervention.

    oilStress 7 × 1

  • +4

    Maintenance tolerance

    With little or no touch-up tolerance, the model favors finishes that ask less of midday maintenance.

    8 − maintenance × 4

  • +2

    Desire

    Moderate appetite adds little pressure by itself.

    desire ≥ 3 → +9 · 0 → −10 · else +2

  • 0

    Coverage appetite

    Moderate coverage adds less all-over film pressure in this model.

    (coverage − 30) × 0.45

  • 0

    Layer count (3)

    Fewer layers reduce the number of interfaces the model has to account for.

    max(0, layers−3)×7 − max(0, 3−layers)×5

  • 0

    Dryness / exposure pressure

    The stated environment adds little dryness pressure to this model.

    dryStress 0 × 0.55

  • 0

    Time available

    The time budget is neutral in this model.

    ≤5 min → −8 · ≥25 min → +7 · else 0

Type fit — ranked

Fit · Fit is how well a product type matches this profile. It runs from a neutral baseline of 50, then signed weights stretch it.

  • Performance sport tint50

    core · base

    Delivers close to the desk's default coverage — set yours and this line becomes personal rather than assumed.

  • Pressed powder compact49

    consider · finish

    Delivers close to the desk's default coverage — set yours and this line becomes personal rather than assumed.

  • Long-wear cheek stain49

    consider · colour

    The catalog prior is relatively favorable for sebum-related wear.

  • Powder blush47

    consider · colour

    The catalog prior is relatively favorable for sebum-related wear.

  • Powder bronzer47

    consider · colour

    The catalog prior is relatively favorable for sebum-related wear.

  • HD optical blur base46

    consider · base

    Delivers close to the desk's default coverage — set yours and this line becomes personal rather than assumed.

  • Strategic setting powder45

    consider · finish

    The catalog prior is relatively favorable for sebum-related wear.

  • Accessible multi-stick45

    consider · base

    Delivers close to the desk's default coverage — set yours and this line becomes personal rather than assumed.

  • Blotting papers44

    consider · finish

    The catalog prior is relatively favorable for sebum-related wear.

  • Shine-control stick44

    consider · finish

    The catalog prior is relatively favorable for sebum-related wear.

  • Green redness corrector44

    consider · spot

    Delivers close to the desk's default coverage — set yours and this line becomes personal rather than assumed.

  • Serum skin tint43

    consider · base

    Delivers close to the desk's default coverage — set yours and this line becomes personal rather than assumed.

Alternative pathways

  • Performance minimal

    47

    Tinted SPF or sport tint, stain color, brows — sweat-honest.

    No cream highlight, no powder rebuild, no dinner-proof glam stack.

  • Prep-led base

    44

    Hydrating prep, then the thinnest base that still reads even.

    Costs a skincare step before any makeup decision.

  • Skincare-first, color later

    43

    Hydration and SPF carry the finish; one color object signals effort.

    Coverage is essentially zero — good skincare has to do the work.

  • Definition without base

    43

    Brow, liner, lash and lip. The complexion is left alone entirely.

    Tone is untouched, so the face reads styled rather than even.

  • Mature flexible film

    43

    Prep, serum tint, cream color — nothing that settles into lines.

    Matte control and full opacity are off the table on purpose.

  • Fragrance-guard capsule

    43

    Hard fragrance-free filters, short list, mineral-aware where possible.

    The shortlist is deliberately narrow — desire spends on texture, not range.

Contextual finding intelligence

The finding should tell you what to look at next.

Each panel ties the current desk context to a measurement procedure, one-variable test, explanatory plate, worked case and source trail. It is a decision aid, not a diagnosis.

audited baked fallback

catalog 2026-09-13-baked

Why this is here now

Separate broad tone evening from local contrast correction.

Coverage is still largely a model default, so start with the smallest film that proves what actually needs more opacity.

Claim evidence ladderFour steps separate ingredient presence, concentration knowledge, plausible mechanism and finished-product evidence.PRESENCEDOSE?often unknownMECHANISMplausible ≠ measuredFINISHED PRODUCTdefined use + endpointSTRONGER OUTCOME EVIDENCE →
Explanatory plate, not a measurement. The measurement procedure is listed beside it.

Measure this

From normal viewing distance, mark which zones still create unwanted contrast after one thin base pass. Add product only to those zones and compare the total film needed.

Smallest useful test

Change only the profile field behind that weight, and watch which products move and by how much.

Decision matrix

Why: a weight is a model prior, not a measurement of your face

Keep constant
Pick the single largest weight in the breakdown and read what it says it is responding to.
Change
Change only the profile field behind that weight, and watch which products move and by how much.
A useful result supports
A ranking that reorders tells you the field matters to the model. A ranking that barely moves tells you the field was not what was deciding.
It cannot establish
It cannot validate the weight. No scoring weight in this desk carries a citation, and the score is an index rather than a probability.
Prefills hypothesis, constants and the one variable to change. Saved locally only.

Sources for this finding