My take using Sigmoid + DT4.1 and my double-white-balance as at Color calibration giving poor results for white balance - #36 by RawConvert
@hannoschwalm , I noticed something about highlight reconstruction you might be interested in. Sampling an area within the main blown part, LCh gave a white value - (405,405,405), whereas Segmentation gave (518,315,373). However it didn’t look as pink as those numbers suggest to me. But it was noticeably different in colour. This was with very few modules enabled.
Thanks @ggbutcher
DSZ_1481-Conejos-river-rawconvert-V1-S-sRGB.xmp (61.0 KB)