two or more tungsten, HMI, or LED as blackbody-approx CCTs — relative illuminance from power and distance, mixed Kelvin, mired, and a CTO/CTB starting point for a target white point
Illuminance at the subject is modeled as P/d². Mixing is in XYZ from each CCT on a planck-approx locus, then a nearest locus T — good for gaffer estimates, not a full spectral solve (fluorescent, gelled sources, and bounce are different).
Illuminance share: 36.0% @ 5600 K · 64.0% @ 3200 K
About 1/2 CTB (≈68 mired) is a solid start; you need about -76 mired total (cooling (lower mired, higher K)). Refine with 1/4 or 1/8 cuts. Treat as a single gel in front of the path you are correcting (lens, fixture, or bounce), not a precise spectrum match — mixed HMI + tungsten is not one black body. Lee / Rosco mired differ slightly by line.