[Dan Davis] ” I would have thought that the 10bit file would retain any information.”
A matter of some curiosity to myself as well. There are some compromises in this approach IMO, because I am not convinced that the “ranging” and gamma-transfer curves are equally valid for both data formats. Bringing in DnX-“baked” MXF containers does not yield the same output values in Resolve that are being generated in NLE applications using their versions of, for example, any of the Sony S-Log profiles. This is easily tested by comparing a baked offline chase video with a side-by-side grade in the color correction timeline. The disparities that I see can be somewhat matched by cascading a number of profiles — using two successive OFX ColorManagement nodes can come close — experimentally I can recreate more or less the same Media Composer output with a “Rec-709” concatenated with a “SonySlog” gamma transfer, but this is still two nodes before I can start where the editor left off. If there is a fairly large exposure-delta, the anti-log transform still might be distorting the transfer curve to a point where I may as well have re-engineered my own linearization process – and for which the camera originals may have the best chance of success.
Bottom line, though, is if you are in the middle, not really any differences to speak of, but if the shot needs to be pushed or pulled some distance, worth it to reconform.
jPo
“I always pass on free advice — its never of any use to me” Oscar Wilde.