DPX transfer characteristics code 3 (logarithm)


(Chad Fogg) #1

Does any one know the formula for DPX transfer characteristics code = 3?
There’s a cryptic message in the SMPTE 268 that defers to I23 subgroup for a definition.

Is it this log10 function implied by this CTL script ?

I am trying to decypher the signal container used in these HDR clips captured from Sony F55 & F65:

http://medialab.sjtu.edu.cn/HDR/

The “Camp Fire" DPX files are here (33 Mbytes each).

ftp://202.120.39.226/dpx

Regards,

–Chad


(Nick Shaw) #2

Unfortunately you cannot reliably infer anything from the DPX Transfer metadata. Frequently the data in there may be wrong, and the code 3, meaning “log”, if used accurately at all, just gets used to indicate that one of many possible log transfer functions has been used.

The DPX spec was defined back when the only common log format was Cineon log from film scans, and it does not have the flexibility to indicate whether the log transfer function used is Cineon, LogC, S-Log3 etc. Accurate notes, or other “human” communication is required!


(Jim Houston) #3

I agree, it is hard to tell what the intent of the party writing this file from that field.
In general, for scans, 0 is often used. 1 was intended for Cineon style curves (based on a
printing density for film). 3 was intended for camera log formats (Viper, et.al.), However the
committee referenced in the document never came to a consensus on this subject so all that you can
say is that the file is logarithmic (with a toe or not is unknown ). It is a problem to put things in standards
that are incomplete.

There is currently a SMPTE Constrained DPX standards group working on putting HDR encodings at the end of the table (PQ,etc…)

Because you know the source, I would look at these files with an assumption that they are SLOG3.
(though SLOG2 is possible)

Jim