Both files I opened load OK. Unfortunately as far as I can see these are being decoded completely incorrectly. Rendering them back to a dpx sequence they do not match the input as compared using AE either. Only tried using 32bit float (full) so far. Things look so mangled I don't know if this is even worth spending anymore time on. It would seem that these are being treated as linear instead of log.
Haven't checked the V9 manual but the online help offers no clue as to how this is expected to work.

Bob.