search

w3c / sync-media-pub

Repository of the Synchronized Multimedia for Publications Community Group
http://w3c.github.io/sync-media-pub
Other
16 stars 4 forks source link

clock values #40

Closed iherman closed 3 years ago

iherman commented 3 years ago

At the moment, the definition of clipBegin and clipEnd refer to §B which lists a number of examples.

However... clipBegin and clipEnd are not prefixed as sync, which means that these are the SMIL attributes as defined in the SMIL Spec for clipBegin and clipEnd, respectively. In SMIL, the value of these attributes refer to SMPTE or NPT and there is a "metric specifier" as part of the value. Looking at the examples in the SMIL spec, we have things like clipBegin="smpte=10:12:33" or clipBegin="npt=123.45s".

I am not an expert in clock values, but there are questions here:

  1. Do we want to keep this complexity in Sync Media? In which cases all the examples in §B are syntactically wrong.
  2. Do we want to choose among SMPTE or NPT, thereby simplifying the attribute (ie, no 'metric specifier')? Then this should be stated explicitly, and giving a proper definition to clipBegin and clipEnd as clock values.

If the choice is (2) (which sounds find to me) then what we have is not the SMIL attribute anymore; we should not create an incompatibility (see also my comment in #35). Which would mean, in practice, to put both those attributes into the sync namespace.

marisademeglio commented 3 years ago

Agree we should formalize what we allow as clock values. I had, for now, punted on that, hence the Media Overlays reference.

I think though that conceptually we are a valid subset of SMIL, which says "If no metric specifier is given, then a default of "npt=" is presumed.".

I wonder if it would be beneficial to leave this unrestricted and accept (1) ? Since SyncMedia supports video, there might be some content creators who would want to use SMPTE frame-level references, whereas we didn't have to accommodate this in Media Overlays.

iherman commented 3 years ago

I think though that conceptually we are a valid subset of SMIL, which says "If no metric specifier is given, then a default of "npt=" is presumed.".

Ah! I did not realize this (I am not really an expert in SMIL in spite of the fact that I still belong, formally, to the CWI department that has many of the SMIL editors on board:-). But sounds then good.

I wonder if it would be beneficial to leave this unrestricted and accept (1) ? Since SyncMedia supports video, there might be some content creators who would want to use SMPTE frame-level references, whereas we didn't have to accommodate this in Media Overlays.

I cannot really comment on that, not being an expert on using these...