Maybe you just create a basic film, then create a new film with some subtitles, then in the second one you "link" the first somehow. Then the second film creates a DCP which depends on the DCP in the first.

Not sure how this would work. You possibly only know the required IDs once the OV DCP has been created. If this is the case one workflow would be:

1. create OV film and make DCP
2. create subbed film; add OV's DCP to a list of dependencies, make DCP
   -- second DCP depends on the first.

This is quite fiddly; it would nicer if you could set everything up and press go. You could do the link with films rather than DCPs.

1. create OV film
2. create subbed film; add OV film to dependencies; make DCP would perhaps make both ...
3. or you could add subbed and OV films to the batch converter and it would Do The Right Thing (TM).

The problem with this approach, perhaps, is that it's quite long-winded. e.g. making a OV DCP and then two VFs, one with subs1 and one with subs2:

1. Create OV DOM project, add video/audio, save.
2. Create new DOM project, add same audio/video (and set it up the same way), add subs1, link to (1), save.
3. Create new DOM project, add same audio/video (and set it up the same way), add subs2, link to (1), save.
4. Load (2) and (3) into batch converter and click go.

Maybe there could be an OV content tab, multiple VF content tabs which always have the OV's content, then multiple DCP tabs. Then:

1. Create OV DOM project, add audio/video.
2. Click "Make VF" which sets up the tabs. Then add subs1 to the VF content tab.
3. Click "Make VF" and add subs2 to the second VF content tab.
4. Set up DCP prefs (somehow) and click go.

DCP prefs are awkward too; you would frequently want them to be the same e.g. an "expected" arrangement might be:

All DCPs share

• content type
• signed/encrypted flags
• standard
• container
• frame rate
• subtitle burn
• 3D
• resolution
• J2K bandwidth
• audio channels
• audio processing

Different

• names
• keys (although the shared assets would have the same keys)

but you could just about imagine VFs with different containers, frame rates ... although perhaps you'd argue that for this first cut our motive is always to be sharing the video.

to create supplemental DCP (VF) from external source (i.e. not made by dcpomatic) probably we need have possibility to choose dependencies - OV film (made by dcpomatic) or DCP (external). What do you think?

True, we probably need to be able to link to an existing DCP or a DCP-o-matic project.

Another way might be to just allow multiple Content/DCP pairs in a Film, then make them as DCPs extracting common parts to a OV.

Internally, this might mean moving all the film's "DCP" settings inside Playlist, then Film just contains multiple Playlists, each is represented in the UI by Content/DCP tabs. Allow versions to be copied from existing versions, linked etc., perhaps.

Unfortunately this is quite a bit of internal upheaval of metadata, since settings would move from Film to Playlist. Maybe the movement of stuff Film -> Playlist should happen before V2 is released...

A first (perhaps useful) cut might be just a "make vf" switch which tries to re-use existing DCP content rather than rewrapping it. Then if the content is just a DCP with some more bits it would be quite easy to implement. Harder if there are multiple pieces of content; this would need multi-reel.

4e6f15f602c605804f95c6b06af9bf79eaf2dde1 and nearby stuff allows imported DCP content to be referenced rather than re-wrapped.