Originally posted by bdery I've read that you can take a keyword and convert it into a "people keyword", that would be fine for me, again IF it works.
If you have a well tagged library, something like that would work. I'd see if that function was there, and how well it works for you, but I'm not around my computer at the moment.
Though, I have to ask, if your tagging method works for you, then what would the difference be? For people with good keyword discipline, I don't really see this as a killer feature, especially since it really don't work that consistently. You're still going to have to go and tag half your pictures. Outside of the fact that LRs algorithm is a bit schizophrenic (two pictures, shot 20 seconds apart, same lightning, same subject, only one has a " recognizable" face?), all face matching algorithms suffer from the limitation that they can only find faces, so if you have a half-turned profile, or a silhouette, they will never help.
I've run into this with Picasa, even if it caught everything that has an actual human face, around 30% of my real world photos need me to go in and add keywords, since people aren't always facing the camera in in-situ, non-portrait photography.