Drop a folder of Lightroom presets (.xmp) or a .zip of them. Each preset is read setting by setting and run through an approximate model of Lightroom's develop pipeline, then grouped into recipe families with a card for each one. Your files never leave this page: nothing is uploaded.
Families are grouped by recipe (curves, calibration and color grade), not by name. Averages are per family. Shadow floor is the output brightness of the darkest 10% of tones. Hue shifts are in Lab: for skin, + means toward yellow (golden); for foliage, + means toward cyan; for sky, − means toward cyan.
These presets have identical settings apart from name, group and housekeeping fields, or differ only in the fields listed.
With white balance and exposure set aside, each group differs by less than about 6 ΔE on its worst test patch: the same look with small tweaks.
These numbers come from a model of Lightroom built from each preset's settings, not from Lightroom renders. Use them to compare presets with each other. Camera-matching and third-party profiles (Cobalt, Fuji film simulations and so on) can't be simulated; those cards are marked and show the recipe on top of Adobe Color. Lightroom doesn't document whether the master point curve runs before or after the R/G/B curves; this page assumes master first. Families, tiers (fade, contrast, brightness) and close variants are relative to the set you drop, so the same preset can land in a different tier in a different set.