What decides is the chromophore, not the colour
Not the colour itself, but which substance causes it. Melanin darkens the whole area and keeps a yellow-brown hue. Haemoglobin adds colour but barely darkens the skin, because the blood lies beneath an intact epidermis. Dried blood has a very dark red-brown core. A scar is recognised by what it lacks, no dark component, no skin markings.
Colorime3ka measures the robust colour of the area (median of channels), converts it to CIELAB and names it by ΔE (CIEDE2000). It adds uniformity, sharpness of the margin, surface roughness, outline and the hue gradient from centre to edge. The reticle also frames the surrounding skin the finding is compared with.
Calibration
A grey card removes the colour cast of the lighting; a 24-patch colour chart gives a correction matrix for comparable series. The printed T2/T3 target from the aPrehliadky project is found in the photo automatically.
Millimetre scale
Two taps on a ruler, the target's millimetre fiducials, or on iPhone Pro straight from LiDAR · no ruler needed. The LiDAR scale is not provided when zoomed in: a silently wrong scale is worse than none.
Atlas of 68 findings
Bilingual, with original schematic drawings: mechanical, thermal and electrical injuries, healing, pigmented lesions, efflorescences, post-mortem changes, and findings regularly mistaken for abuse.
Postoperative wound (3.1)
With the context "surgery" and the date of the operation, the app stops guessing what the finding is and describes healing on the Southampton score and the CDC/NHSN definition of infection. Never above grade II from colour: pus and dehiscence are never claimed from a photo.
Reference fingerprints
Recognition thresholds are measured on verified clinical photographs. Expert mode stores a measurement as a further reference: twelve numbers and a note, never a photo; colleagues exchange them as .ch3lab files, and received fingerprints wait in quarantine.