Group-defined field map

Due to an miscommunication that was not caught early enough, I have fieldmaps in only one encoding direction for a subset of my participants. They have the same encoding direction as my bold images, so I can’t just extract the first few volumes from the bold images to have a complete pair. We are currently comparing PEPOLAR to Syn fieldmap-less correction, but I wanted to see if anyone has tried to develop a group-level field map. There is a mention on the HCP Pipeline site that this is being worked on, but I am curious if anyone has tried to do this and has any advice/feedback. I have paired images for most of my participants, so I would hopefully be able to develop a fairly reliable group average.