Redness
Elevation of the red channel (a*) vs your own facial baseline — the research standard for measuring erythema.
- Change threshold
- 17.3
- Reliability
- 0.926 very high
- Measurement noise
- 6.2
MIRRA · The Ten Metrics
The ten metrics
Every metric carries three numbers we publish: what it measures, the threshold below which we will not call a change an improvement, and how reliably it repeats on a retest. The third matters most — and we did not hide the weakest one.
Spots & pigmentation
Detection of localized dark areas — hyperpigmentation, sun damage and hormonal spots.
- Change threshold
- 20.1
- Reliability
- 0.906 very high
- Measurement noise
- 7.3
Nasolabial fold
Where the tissue sits — the jawline and the nasolabial folds, from landmark geometry. Shown, not scored: it moves with expression.
- Change threshold
- 17.4
- Reliability
- 0.904 very high
- Measurement noise
- 6.3
Fine lines
Line density around the eyes and forehead — the strongest predictor of perceived age.
- Change threshold
- 15.8
- Reliability
- 0.907 very high
- Measurement noise
- 5.7
Shine
Specular reflections in the T-zone — actual oiliness, not how it feels.
- Change threshold
- 19.7
- Reliability
- 0.938 very high
- Measurement noise
- 7.1
Pores
Pore visibility on cheeks and nose — one of the 10 signs validated against dermatologists.
- Change threshold
- 18.1
- Reliability
- 0.924 very high
- Measurement noise
- 6.5
Tone evenness
Local color homogeneity — research found it alone accounts for up to 20 years of perceived age.
- Change threshold
- 17.4
- Reliability
- 0.904 very high
- Measurement noise
- 6.3
Blemishes
Localized inflammatory lesions — counted separately from spots, because they answer a product in days rather than months.
- Change threshold
- 23.0
- Reliability
- 0.917 very high
- Measurement noise
- 8.3
Dark circles
A color reading in the tear-trough band alone — kept apart from crow’s feet, which are a texture reading over a wider area.
- Change threshold
- 20.3
- Reliability
- 0.887 high
- Measurement noise
- 7.3
Dryness
Light scattering from fine surface flaking — what reads as “flaky” before it is felt.
- Change threshold
- 18.5
- Reliability
- 0.889 high
- Measurement noise
- 6.7
Thresholds: the wider of a capture-jitter study on 50 real scans and real repeat pairs · ICC from the jitter study, not from repeat visits · the metric comes from a model trained on licensed dermatology datasets · A SCAN, not a photograph: 5–9 frames fused