The extraction for fiberId 85 in 97127 r1 seems to have problems. It's a single brightish object, and seems to have extracted badly and messed up 84/86 too. b doesn't look great either.