Since this topic was brought up, I'd just like to ask something.
Is --direct auto a little more efficient in 2 pass mode than crf? I ask because I notice that x264 usually selects spatial in crf, but often selects a lot more temporal in 2 pass. If there's a difference, my guess would be that it would be fairly minor, am I correct?
(note, this is from memory, from a while ago, if it's changed more recently, then my apologies)