Re: [Openexr-devel] Re: [Openexr] KeyKode offsets

[Ken McGaugh]

Kevin Wheatley wrote:
> Florian Kainz wrote:
>
> >     A) DPX-style film edge codes, as listed in section 6.1 of SMPTE 268M,
> >        "File Format for Digital Moving-Picture Exchange":
>
>
> This gets my vote if you add a 'perfs per frame' indicator in some
> way, (although scanning people will have to work out what DPX field we
> could get the scanners to put it in ideally)

I too vote for this one, especially since I've already implemented
it here (including the perfsPerFrame).  I chose to interpret perfOffset
as the number of perfs between the reference mark and the tail end of the
frame.  So for the zero-frame, the perfsPerFrame would be a number between
0 and (perfsPerFrame-1).  This is consistent with how the northlight scanner
(with the keycode option) encodes the perfOffset into dpx files.  This
method also works regadless of the film format.

Florian, note that the "count" field is actually "foot count" not "frame count".


> >     B) "Film feet edge numbers" as specified by SMPTE RP 195, "Use of the
> >        Reference Mark in Manufacturer-Printed Latent Image Key Numbers for
> >        Unambiguous Film Frame Identification".  This numbering scheme
> >        covers 35mm and 65mm film with key numbers occuring every 64, 80
> >        and 120 perforations, and with 3, 4, 8 and 15-perforation frames.
> >        (If you are voting for this one, please explain to me how the
> >        scheme works.  After reading RP 195, I still don't understand it.)
>
>
> I see this as a way of displaying the KeyKodes, however it is missing
> the film manufacturer ID code mappings which of course expand every
> time a new stock comes out.

I found RP195 very useful in understanding how to read a film, but I really
wish they hadn't used the term "perfOffset" the way they did since it
conflicts with other specs.

--Ken